Upload 48 files

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 0.28,
+    "train_loss": 1.2429857851664226,
+    "train_runtime": 27386.6257,
+    "train_samples": 173235,
+    "train_samples_per_second": 1.753,
+    "train_steps_per_second": 0.11
+}

checkpoint-1000/config.json

@@ -0,0 +1,44 @@
+{
+  "_name_or_path": "E:/PycharmProjects/dl_models/chatglm2-6b-int4",
+  "add_bias_linear": false,
+  "add_qkv_bias": true,
+  "apply_query_key_layer_scaling": true,
+  "apply_residual_connection_post_layernorm": false,
+  "architectures": [
+    "ChatGLMForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "attention_softmax_in_fp32": true,
+  "auto_map": {
+    "AutoConfig": "configuration_chatglm.ChatGLMConfig",
+    "AutoModel": "modeling_chatglm.ChatGLMForConditionalGeneration",
+    "AutoModelForSeq2SeqLM": "modeling_chatglm.ChatGLMForConditionalGeneration"
+  },
+  "bias_dropout_fusion": true,
+  "eos_token_id": 2,
+  "ffn_hidden_size": 13696,
+  "fp32_residual_connection": false,
+  "hidden_dropout": 0.0,
+  "hidden_size": 4096,
+  "kv_channels": 128,
+  "layernorm_epsilon": 1e-05,
+  "model_type": "chatglm",
+  "multi_query_attention": true,
+  "multi_query_group_num": 2,
+  "num_attention_heads": 32,
+  "num_layers": 28,
+  "original_rope": true,
+  "pad_token_id": 0,
+  "padded_vocab_size": 65024,
+  "post_layer_norm": true,
+  "pre_seq_len": 128,
+  "prefix_projection": false,
+  "quantization_bit": 4,
+  "rmsnorm": true,
+  "seq_length": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.31.0",
+  "use_cache": true,
+  "vocab_size": 65024
+}

checkpoint-1000/configuration_chatglm.py

@@ -0,0 +1,59 @@
+from transformers import PretrainedConfig
+
+
+class ChatGLMConfig(PretrainedConfig):
+    model_type = "chatglm"
+    def __init__(
+        self,
+        num_layers=28,
+        padded_vocab_size=65024,
+        hidden_size=4096,
+        ffn_hidden_size=13696,
+        kv_channels=128,
+        num_attention_heads=32,
+        seq_length=2048,
+        hidden_dropout=0.0,
+        attention_dropout=0.0,
+        layernorm_epsilon=1e-5,
+        rmsnorm=True,
+        apply_residual_connection_post_layernorm=False,
+        post_layer_norm=True,
+        add_bias_linear=False,
+        add_qkv_bias=False,
+        bias_dropout_fusion=True,
+        multi_query_attention=False,
+        multi_query_group_num=1,
+        apply_query_key_layer_scaling=True,
+        attention_softmax_in_fp32=True,
+        fp32_residual_connection=False,
+        quantization_bit=0,
+        pre_seq_len=None,
+        prefix_projection=False,
+        **kwargs
+    ):
+        self.num_layers = num_layers
+        self.vocab_size = padded_vocab_size
+        self.padded_vocab_size = padded_vocab_size
+        self.hidden_size = hidden_size
+        self.ffn_hidden_size = ffn_hidden_size
+        self.kv_channels = kv_channels
+        self.num_attention_heads = num_attention_heads
+        self.seq_length = seq_length
+        self.hidden_dropout = hidden_dropout
+        self.attention_dropout = attention_dropout
+        self.layernorm_epsilon = layernorm_epsilon
+        self.rmsnorm = rmsnorm
+        self.apply_residual_connection_post_layernorm = apply_residual_connection_post_layernorm
+        self.post_layer_norm = post_layer_norm
+        self.add_bias_linear = add_bias_linear
+        self.add_qkv_bias = add_qkv_bias
+        self.bias_dropout_fusion = bias_dropout_fusion
+        self.multi_query_attention = multi_query_attention
+        self.multi_query_group_num = multi_query_group_num
+        self.apply_query_key_layer_scaling = apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = attention_softmax_in_fp32
+        self.fp32_residual_connection = fp32_residual_connection
+        self.quantization_bit = quantization_bit
+        self.pre_seq_len = pre_seq_len
+        self.prefix_projection = prefix_projection
+        super().__init__(**kwargs)

checkpoint-1000/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.31.0"
+}

checkpoint-1000/modeling_chatglm.py

@@ -0,0 +1,1285 @@
+""" PyTorch ChatGLM model. """
+
+import math
+import copy
+import warnings
+import re
+import sys
+
+import torch
+import torch.utils.checkpoint
+import torch.nn.functional as F
+from torch import nn
+from torch.nn import CrossEntropyLoss, LayerNorm
+from torch.nn import CrossEntropyLoss, LayerNorm, MSELoss, BCEWithLogitsLoss
+from torch.nn.utils import skip_init
+from typing import Optional, Tuple, Union, List, Callable, Dict, Any
+
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+from transformers.generation.logits_process import LogitsProcessor
+from transformers.generation.utils import LogitsProcessorList, StoppingCriteriaList, GenerationConfig, ModelOutput
+
+from .configuration_chatglm import ChatGLMConfig
+
+# flags required to enable jit fusion kernels
+
+if sys.platform != 'darwin':
+    torch._C._jit_set_profiling_mode(False)
+    torch._C._jit_set_profiling_executor(False)
+    torch._C._jit_override_can_fuse_on_cpu(True)
+    torch._C._jit_override_can_fuse_on_gpu(True)
+
+logger = logging.get_logger(__name__)
+
+_CHECKPOINT_FOR_DOC = "THUDM/ChatGLM2-6B"
+_CONFIG_FOR_DOC = "ChatGLM6BConfig"
+
+CHATGLM_6B_PRETRAINED_MODEL_ARCHIVE_LIST = [
+    "THUDM/chatglm2-6b",
+    # See all ChatGLM models at https://huggingface.co/models?filter=chatglm
+]
+
+
+def default_init(cls, *args, **kwargs):
+    return cls(*args, **kwargs)
+
+
+class InvalidScoreLogitsProcessor(LogitsProcessor):
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
+        if torch.isnan(scores).any() or torch.isinf(scores).any():
+            scores.zero_()
+            scores[..., 5] = 5e4
+        return scores
+
+
+class PrefixEncoder(torch.nn.Module):
+    """
+    The torch.nn model to encode the prefix
+    Input shape: (batch-size, prefix-length)
+    Output shape: (batch-size, prefix-length, 2*layers*hidden)
+    """
+
+    def __init__(self, config: ChatGLMConfig):
+        super().__init__()
+        self.prefix_projection = config.prefix_projection
+        if self.prefix_projection:
+            # Use a two-layer MLP to encode the prefix
+            kv_size = config.num_layers * config.kv_channels * config.multi_query_group_num * 2
+            self.embedding = torch.nn.Embedding(config.pre_seq_len, kv_size)
+            self.trans = torch.nn.Sequential(
+                torch.nn.Linear(kv_size, config.hidden_size),
+                torch.nn.Tanh(),
+                torch.nn.Linear(config.hidden_size, kv_size)
+            )
+        else:
+            self.embedding = torch.nn.Embedding(config.pre_seq_len,
+                                                config.num_layers * config.kv_channels * config.multi_query_group_num * 2)
+
+    def forward(self, prefix: torch.Tensor):
+        if self.prefix_projection:
+            prefix_tokens = self.embedding(prefix)
+            past_key_values = self.trans(prefix_tokens)
+        else:
+            past_key_values = self.embedding(prefix)
+        return past_key_values
+
+
+def split_tensor_along_last_dim(
+        tensor: torch.Tensor,
+        num_partitions: int,
+        contiguous_split_chunks: bool = False,
+) -> List[torch.Tensor]:
+    """Split a tensor along its last dimension.
+
+    Arguments:
+        tensor: input tensor.
+        num_partitions: number of partitions to split the tensor
+        contiguous_split_chunks: If True, make each chunk contiguous
+                                 in memory.
+
+    Returns:
+        A list of Tensors
+    """
+    # Get the size and dimension.
+    last_dim = tensor.dim() - 1
+    last_dim_size = tensor.size()[last_dim] // num_partitions
+    # Split.
+    tensor_list = torch.split(tensor, last_dim_size, dim=last_dim)
+    # Note: torch.split does not create contiguous tensors by default.
+    if contiguous_split_chunks:
+        return tuple(chunk.contiguous() for chunk in tensor_list)
+
+    return tensor_list
+
+
+class RotaryEmbedding(nn.Module):
+    def __init__(self, dim, original_impl=False, device=None, dtype=None):
+        super().__init__()
+        inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2, device=device).to(dtype=dtype) / dim))
+        self.register_buffer("inv_freq", inv_freq)
+        self.dim = dim
+        self.original_impl = original_impl
+
+    def forward_impl(
+            self, seq_len: int, n_elem: int, dtype: torch.dtype, device: torch.device, base: int = 10000
+    ):
+        """Enhanced Transformer with Rotary Position Embedding.
+
+        Derived from: https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/labml_nn/
+        transformers/rope/__init__.py. MIT License:
+        https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/license.
+        """
+        # $\Theta = {\theta_i = 10000^{\frac{2(i-1)}{d}}, i \in [1, 2, ..., \frac{d}{2}]}$
+        theta = 1.0 / (base ** (torch.arange(0, n_elem, 2, dtype=dtype, device=device) / n_elem))
+
+        # Create position indexes `[0, 1, ..., seq_len - 1]`
+        seq_idx = torch.arange(seq_len, dtype=dtype, device=device)
+
+        # Calculate the product of position index and $\theta_i$
+        idx_theta = torch.outer(seq_idx, theta).float()
+
+        cache = torch.stack([torch.cos(idx_theta), torch.sin(idx_theta)], dim=-1)
+
+        # this is to mimic the behaviour of complex32, else we will get different results
+        if dtype in (torch.float16, torch.bfloat16, torch.int8):
+            cache = cache.bfloat16() if dtype == torch.bfloat16 else cache.half()
+        return cache
+
+    def forward(self, max_seq_len, offset=0):
+        return self.forward_impl(
+            max_seq_len, self.dim, dtype=self.inv_freq.dtype, device=self.inv_freq.device
+        )
+
+
+@torch.jit.script
+def apply_rotary_pos_emb(x: torch.Tensor, rope_cache: torch.Tensor) -> torch.Tensor:
+    # x: [sq, b, np, hn]
+    sq, b, np, hn = x.size(0), x.size(1), x.size(2), x.size(3)
+    rot_dim = rope_cache.shape[-2] * 2
+    x, x_pass = x[..., :rot_dim], x[..., rot_dim:]
+    # truncate to support variable sizes
+    rope_cache = rope_cache[:sq]
+    xshaped = x.reshape(sq, -1, np, rot_dim // 2, 2)
+    rope_cache = rope_cache.view(sq, -1, 1, xshaped.size(3), 2)
+    x_out2 = torch.stack(
+        [
+            xshaped[..., 0] * rope_cache[..., 0] - xshaped[..., 1] * rope_cache[..., 1],
+            xshaped[..., 1] * rope_cache[..., 0] + xshaped[..., 0] * rope_cache[..., 1],
+        ],
+        -1,
+    )
+    x_out2 = x_out2.flatten(3)
+    return torch.cat((x_out2, x_pass), dim=-1)
+
+
+class RMSNorm(torch.nn.Module):
+    def __init__(self, normalized_shape, eps=1e-5, device=None, dtype=None, **kwargs):
+        super().__init__()
+        self.weight = torch.nn.Parameter(torch.empty(normalized_shape, device=device, dtype=dtype))
+        self.eps = eps
+
+    def forward(self, hidden_states: torch.Tensor):
+        input_dtype = hidden_states.dtype
+        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
+
+        return (self.weight * hidden_states).to(input_dtype)
+
+
+class CoreAttention(torch.nn.Module):
+    def __init__(self, config: ChatGLMConfig, layer_number):
+        super(CoreAttention, self).__init__()
+
+        self.apply_query_key_layer_scaling = config.apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
+        if self.apply_query_key_layer_scaling:
+            self.attention_softmax_in_fp32 = True
+        self.layer_number = max(1, layer_number)
+
+        projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_partition = projection_size
+        self.hidden_size_per_attention_head = projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        coeff = None
+        self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
+        if self.apply_query_key_layer_scaling:
+            coeff = self.layer_number
+            self.norm_factor *= coeff
+        self.coeff = coeff
+
+        self.attention_dropout = torch.nn.Dropout(config.attention_dropout)
+
+    def forward(self, query_layer, key_layer, value_layer, attention_mask):
+        pytorch_major_version = int(torch.__version__.split('.')[0])
+        if pytorch_major_version >= 2:
+            query_layer, key_layer, value_layer = [k.permute(1, 2, 0, 3) for k in [query_layer, key_layer, value_layer]]
+            if attention_mask is None and query_layer.shape[2] == key_layer.shape[2]:
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 is_causal=True)
+            else:
+                if attention_mask is not None:
+                    attention_mask = ~attention_mask
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 attention_mask)
+            context_layer = context_layer.permute(2, 0, 1, 3)
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.reshape(*new_context_layer_shape)
+        else:
+            # Raw attention scores
+
+            # [b, np, sq, sk]
+            output_size = (query_layer.size(1), query_layer.size(2), query_layer.size(0), key_layer.size(0))
+
+            # [sq, b, np, hn] -> [sq, b * np, hn]
+            query_layer = query_layer.view(output_size[2], output_size[0] * output_size[1], -1)
+            # [sk, b, np, hn] -> [sk, b * np, hn]
+            key_layer = key_layer.view(output_size[3], output_size[0] * output_size[1], -1)
+
+            # preallocting input tensor: [b * np, sq, sk]
+            matmul_input_buffer = torch.empty(
+                output_size[0] * output_size[1], output_size[2], output_size[3], dtype=query_layer.dtype,
+                device=query_layer.device
+            )
+
+            # Raw attention scores. [b * np, sq, sk]
+            matmul_result = torch.baddbmm(
+                matmul_input_buffer,
+                query_layer.transpose(0, 1),  # [b * np, sq, hn]
+                key_layer.transpose(0, 1).transpose(1, 2),  # [b * np, hn, sk]
+                beta=0.0,
+                alpha=(1.0 / self.norm_factor),
+            )
+
+            # change view to [b, np, sq, sk]
+            attention_scores = matmul_result.view(*output_size)
+
+            # ===========================
+            # Attention probs and dropout
+            # ===========================
+
+            # attention scores and attention mask [b, np, sq, sk]
+            if self.attention_softmax_in_fp32:
+                attention_scores = attention_scores.float()
+            if self.coeff is not None:
+                attention_scores = attention_scores * self.coeff
+            if attention_mask is None and attention_scores.shape[2] == attention_scores.shape[3]:
+                attention_mask = torch.ones(output_size[0], 1, output_size[2], output_size[3],
+                                            device=attention_scores.device, dtype=torch.bool)
+                attention_mask.tril_()
+                attention_mask = ~attention_mask
+            if attention_mask is not None:
+                attention_scores = attention_scores.masked_fill(attention_mask, float("-inf"))
+            attention_probs = F.softmax(attention_scores, dim=-1)
+            attention_probs = attention_probs.type_as(value_layer)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.attention_dropout(attention_probs)
+            # =========================
+            # Context layer. [sq, b, hp]
+            # =========================
+
+            # value_layer -> context layer.
+            # [sk, b, np, hn] --> [b, np, sq, hn]
+
+            # context layer shape: [b, np, sq, hn]
+            output_size = (value_layer.size(1), value_layer.size(2), query_layer.size(0), value_layer.size(3))
+            # change view [sk, b * np, hn]
+            value_layer = value_layer.view(value_layer.size(0), output_size[0] * output_size[1], -1)
+            # change view [b * np, sq, sk]
+            attention_probs = attention_probs.view(output_size[0] * output_size[1], output_size[2], -1)
+            # matmul: [b * np, sq, hn]
+            context_layer = torch.bmm(attention_probs, value_layer.transpose(0, 1))
+            # change view [b, np, sq, hn]
+            context_layer = context_layer.view(*output_size)
+            # [b, np, sq, hn] --> [sq, b, np, hn]
+            context_layer = context_layer.permute(2, 0, 1, 3).contiguous()
+            # [sq, b, np, hn] --> [sq, b, hp]
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.view(*new_context_layer_shape)
+
+        return context_layer
+
+
+class SelfAttention(torch.nn.Module):
+    """Parallel self-attention layer abstract class.
+
+    Self-attention layer takes input with size [s, b, h]
+    and returns output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(SelfAttention, self).__init__()
+        self.layer_number = max(1, layer_number)
+
+        self.projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_attention_head = self.projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        self.multi_query_attention = config.multi_query_attention
+        self.qkv_hidden_size = 3 * self.projection_size
+        if self.multi_query_attention:
+            self.num_multi_query_groups_per_partition = config.multi_query_group_num
+            self.qkv_hidden_size = (
+                    self.projection_size + 2 * self.hidden_size_per_attention_head * config.multi_query_group_num
+            )
+        self.query_key_value = nn.Linear(config.hidden_size, self.qkv_hidden_size,
+                                         bias=config.add_bias_linear or config.add_qkv_bias,
+                                         device=device, **_config_to_kwargs(config)
+                                         )
+
+        self.core_attention = CoreAttention(config, self.layer_number)
+
+        # Output.
+        self.dense = nn.Linear(self.projection_size, config.hidden_size, bias=config.add_bias_linear,
+                               device=device, **_config_to_kwargs(config)
+                               )
+
+    def _allocate_memory(self, inference_max_sequence_len, batch_size, device=None, dtype=None):
+        if self.multi_query_attention:
+            num_attention_heads = self.num_multi_query_groups_per_partition
+        else:
+            num_attention_heads = self.num_attention_heads_per_partition
+        return torch.empty(
+            inference_max_sequence_len,
+            batch_size,
+            num_attention_heads,
+            self.hidden_size_per_attention_head,
+            dtype=dtype,
+            device=device,
+        )
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True
+    ):
+        # hidden_states: [sq, b, h]
+
+        # =================================================
+        # Pre-allocate memory for key-values for inference.
+        # =================================================
+        # =====================
+        # Query, Key, and Value
+        # =====================
+
+        # Attention heads [sq, b, h] --> [sq, b, (np * 3 * hn)]
+        mixed_x_layer = self.query_key_value(hidden_states)
+
+        if self.multi_query_attention:
+            (query_layer, key_layer, value_layer) = mixed_x_layer.split(
+                [
+                    self.num_attention_heads_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                ],
+                dim=-1,
+            )
+            query_layer = query_layer.view(
+                query_layer.size()[:-1] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            key_layer = key_layer.view(
+                key_layer.size()[:-1] + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.view(
+                value_layer.size()[:-1]
+                + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+        else:
+            new_tensor_shape = mixed_x_layer.size()[:-1] + \
+                               (self.num_attention_heads_per_partition,
+                                3 * self.hidden_size_per_attention_head)
+            mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
+
+            # [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
+            (query_layer, key_layer, value_layer) = split_tensor_along_last_dim(mixed_x_layer, 3)
+
+        # apply relative positional encoding (rotary embedding)
+        if rotary_pos_emb is not None:
+            query_layer = apply_rotary_pos_emb(query_layer, rotary_pos_emb)
+            key_layer = apply_rotary_pos_emb(key_layer, rotary_pos_emb)
+
+        # adjust key and value for inference
+        if kv_cache is not None:
+            cache_k, cache_v = kv_cache
+            key_layer = torch.cat((cache_k, key_layer), dim=0)
+            value_layer = torch.cat((cache_v, value_layer), dim=0)
+        if use_cache:
+            kv_cache = (key_layer, value_layer)
+        else:
+            kv_cache = None
+
+        if self.multi_query_attention:
+            key_layer = key_layer.unsqueeze(-2)
+            key_layer = key_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            key_layer = key_layer.contiguous().view(
+                key_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.unsqueeze(-2)
+            value_layer = value_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            value_layer = value_layer.contiguous().view(
+                value_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+
+        # ==================================
+        # core attention computation
+        # ==================================
+
+        context_layer = self.core_attention(query_layer, key_layer, value_layer, attention_mask)
+
+        # =================
+        # Output. [sq, b, h]
+        # =================
+
+        output = self.dense(context_layer)
+
+        return output, kv_cache
+
+
+def _config_to_kwargs(args):
+    common_kwargs = {
+        "dtype": args.torch_dtype,
+    }
+    return common_kwargs
+
+
+class MLP(torch.nn.Module):
+    """MLP.
+
+    MLP will take the input with h hidden state, project it to 4*h
+    hidden dimension, perform nonlinear transformation, and project the
+    state back into h hidden dimension.
+    """
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(MLP, self).__init__()
+
+        self.add_bias = config.add_bias_linear
+
+        # Project to 4h. If using swiglu double the output width, see https://arxiv.org/pdf/2002.05202.pdf
+        self.dense_h_to_4h = nn.Linear(
+            config.hidden_size,
+            config.ffn_hidden_size * 2,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+        def swiglu(x):
+            x = torch.chunk(x, 2, dim=-1)
+            return F.silu(x[0]) * x[1]
+
+        self.activation_func = swiglu
+
+        # Project back to h.
+        self.dense_4h_to_h = nn.Linear(
+            config.ffn_hidden_size,
+            config.hidden_size,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+    def forward(self, hidden_states):
+        # [s, b, 4hp]
+        intermediate_parallel = self.dense_h_to_4h(hidden_states)
+        intermediate_parallel = self.activation_func(intermediate_parallel)
+        # [s, b, h]
+        output = self.dense_4h_to_h(intermediate_parallel)
+        return output
+
+
+class GLMBlock(torch.nn.Module):
+    """A single transformer layer.
+
+    Transformer layer takes input with size [s, b, h] and returns an
+    output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(GLMBlock, self).__init__()
+        self.layer_number = layer_number
+
+        self.apply_residual_connection_post_layernorm = config.apply_residual_connection_post_layernorm
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+        LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+        # Layernorm on the input data.
+        self.input_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                             dtype=config.torch_dtype)
+
+        # Self attention.
+        self.self_attention = SelfAttention(config, layer_number, device=device)
+        self.hidden_dropout = config.hidden_dropout
+
+        # Layernorm on the attention output
+        self.post_attention_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                      dtype=config.torch_dtype)
+
+        # MLP
+        self.mlp = MLP(config, device=device)
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True,
+    ):
+        # hidden_states: [s, b, h]
+
+        # Layer norm at the beginning of the transformer layer.
+        layernorm_output = self.input_layernorm(hidden_states)
+        # Self attention.
+        attention_output, kv_cache = self.self_attention(
+            layernorm_output,
+            attention_mask,
+            rotary_pos_emb,
+            kv_cache=kv_cache,
+            use_cache=use_cache
+        )
+
+        # Residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = hidden_states
+
+        layernorm_input = torch.nn.functional.dropout(attention_output, p=self.hidden_dropout, training=self.training)
+        layernorm_input = residual + layernorm_input
+
+        # Layer norm post the self attention.
+        layernorm_output = self.post_attention_layernorm(layernorm_input)
+
+        # MLP.
+        mlp_output = self.mlp(layernorm_output)
+
+        # Second residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = layernorm_input
+
+        output = torch.nn.functional.dropout(mlp_output, p=self.hidden_dropout, training=self.training)
+        output = residual + output
+
+        return output, kv_cache
+
+
+class GLMTransformer(torch.nn.Module):
+    """Transformer class."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(GLMTransformer, self).__init__()
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+        self.post_layer_norm = config.post_layer_norm
+
+        # Number of layers.
+        self.num_layers = config.num_layers
+
+        # Transformer layers.
+        def build_layer(layer_number):
+            return GLMBlock(config, layer_number, device=device)
+
+        self.layers = torch.nn.ModuleList([build_layer(i + 1) for i in range(self.num_layers)])
+
+        if self.post_layer_norm:
+            LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+            # Final layer norm before output.
+            self.final_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                 dtype=config.torch_dtype)
+
+        self.gradient_checkpointing = False
+
+    def _get_layer(self, layer_number):
+        return self.layers[layer_number]
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_caches=None,
+            use_cache: Optional[bool] = True,
+            output_hidden_states: Optional[bool] = False,
+    ):
+        if not kv_caches:
+            kv_caches = [None for _ in range(self.num_layers)]
+        presents = () if use_cache else None
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning_once(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        all_self_attentions = None
+        all_hidden_states = () if output_hidden_states else None
+        for index in range(self.num_layers):
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            layer = self._get_layer(index)
+            if self.gradient_checkpointing and self.training:
+                layer_ret = torch.utils.checkpoint.checkpoint(
+                    layer,
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_caches[index],
+                    use_cache
+                )
+            else:
+                layer_ret = layer(
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_cache=kv_caches[index],
+                    use_cache=use_cache
+                )
+            hidden_states, kv_cache = layer_ret
+            if use_cache:
+                presents = presents + (kv_cache,)
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        # Final layer norm.
+        if self.post_layer_norm:
+            hidden_states = self.final_layernorm(hidden_states)
+
+        return hidden_states, presents, all_hidden_states, all_self_attentions
+
+
+class ChatGLMPreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and
+    a simple interface for downloading and loading pretrained models.
+    """
+
+    is_parallelizable = False
+    supports_gradient_checkpointing = True
+    config_class = ChatGLMConfig
+    base_model_prefix = "transformer"
+    _no_split_modules = ["GLMBlock"]
+
+    def _init_weights(self, module: nn.Module):
+        """Initialize the weights."""
+        return
+
+    def get_masks(self, input_ids, past_key_values, padding_mask=None):
+        batch_size, seq_length = input_ids.shape
+        full_attention_mask = torch.ones(batch_size, seq_length, seq_length, device=input_ids.device)
+        full_attention_mask.tril_()
+        past_length = 0
+        if past_key_values:
+            past_length = past_key_values[0][0].shape[0]
+        if past_length:
+            full_attention_mask = torch.cat((torch.ones(batch_size, seq_length, past_length,
+                                                        device=input_ids.device), full_attention_mask), dim=-1)
+        if padding_mask is not None:
+            full_attention_mask = full_attention_mask * padding_mask.unsqueeze(1)
+        if not past_length and padding_mask is not None:
+            full_attention_mask -= padding_mask.unsqueeze(-1) - 1
+        full_attention_mask = (full_attention_mask < 0.5).bool()
+        full_attention_mask.unsqueeze_(1)
+        return full_attention_mask
+
+    def get_position_ids(self, input_ids, device):
+        batch_size, seq_length = input_ids.shape
+        position_ids = torch.arange(seq_length, dtype=torch.long, device=device).unsqueeze(0).repeat(batch_size, 1)
+        return position_ids
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, GLMTransformer):
+            module.gradient_checkpointing = value
+
+
+class Embedding(torch.nn.Module):
+    """Language model embeddings."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(Embedding, self).__init__()
+
+        self.hidden_size = config.hidden_size
+        # Word embeddings (parallel).
+        self.word_embeddings = nn.Embedding(
+            config.padded_vocab_size,
+            self.hidden_size,
+            dtype=config.torch_dtype,
+            device=device
+        )
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+    def forward(self, input_ids):
+        # Embeddings.
+        words_embeddings = self.word_embeddings(input_ids)
+        embeddings = words_embeddings
+        # Data format change to avoid explicit tranposes : [b s h] --> [s b h].
+        embeddings = embeddings.transpose(0, 1).contiguous()
+        # If the input flag for fp32 residual connection is set, convert for float.
+        if self.fp32_residual_connection:
+            embeddings = embeddings.float()
+        return embeddings
+
+
+class ChatGLMModel(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, device=None, empty_init=True):
+        super().__init__(config)
+        if empty_init:
+            init_method = skip_init
+        else:
+            init_method = default_init
+        init_kwargs = {}
+        if device is not None:
+            init_kwargs["device"] = device
+        self.embedding = init_method(Embedding, config, **init_kwargs)
+        self.num_layers = config.num_layers
+        self.multi_query_group_num = config.multi_query_group_num
+        self.kv_channels = config.kv_channels
+
+        # Rotary positional embeddings
+        self.seq_length = config.seq_length
+        rotary_dim = (
+            config.hidden_size // config.num_attention_heads if config.kv_channels is None else config.kv_channels
+        )
+
+        self.rotary_pos_emb = RotaryEmbedding(rotary_dim // 2, original_impl=config.original_rope, device=device,
+                                              dtype=config.torch_dtype)
+        self.encoder = init_method(GLMTransformer, config, **init_kwargs)
+        self.output_layer = init_method(nn.Linear, config.hidden_size, config.padded_vocab_size, bias=False,
+                                        dtype=config.torch_dtype, **init_kwargs)
+        self.pre_seq_len = config.pre_seq_len
+        self.prefix_projection = config.prefix_projection
+        if self.pre_seq_len is not None:
+            for param in self.parameters():
+                param.requires_grad = False
+            self.prefix_tokens = torch.arange(self.pre_seq_len).long()
+            self.prefix_encoder = PrefixEncoder(config)
+            self.dropout = torch.nn.Dropout(0.1)
+
+    def get_input_embeddings(self):
+        return self.embedding.word_embeddings
+
+    def get_prompt(self, batch_size, device, dtype=torch.half):
+        prefix_tokens = self.prefix_tokens.unsqueeze(0).expand(batch_size, -1).to(device)
+        past_key_values = self.prefix_encoder(prefix_tokens).type(dtype)
+        past_key_values = past_key_values.view(
+            batch_size,
+            self.pre_seq_len,
+            self.num_layers * 2,
+            self.multi_query_group_num,
+            self.kv_channels
+        )
+        # seq_len, b, nh, hidden_size
+        past_key_values = self.dropout(past_key_values)
+        past_key_values = past_key_values.permute([2, 1, 0, 3, 4]).split(2)
+        return past_key_values
+
+    def forward(
+            self,
+            input_ids,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.BoolTensor] = None,
+            full_attention_mask: Optional[torch.BoolTensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ):
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        batch_size, seq_length = input_ids.shape
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embedding(input_ids)
+
+        if self.pre_seq_len is not None:
+            if past_key_values is None:
+                past_key_values = self.get_prompt(batch_size=batch_size, device=input_ids.device,
+                                                  dtype=inputs_embeds.dtype)
+            if attention_mask is not None:
+                attention_mask = torch.cat([attention_mask.new_ones((batch_size, self.pre_seq_len)),
+                                            attention_mask], dim=-1)
+
+        if full_attention_mask is None:
+            if (attention_mask is not None and not attention_mask.all()) or (past_key_values and seq_length != 1):
+                full_attention_mask = self.get_masks(input_ids, past_key_values, padding_mask=attention_mask)
+
+        # Rotary positional embeddings
+        rotary_pos_emb = self.rotary_pos_emb(self.seq_length)
+        if position_ids is not None:
+            rotary_pos_emb = rotary_pos_emb[position_ids]
+        else:
+            rotary_pos_emb = rotary_pos_emb[None, :seq_length]
+        rotary_pos_emb = rotary_pos_emb.transpose(0, 1).contiguous()
+
+        # Run encoder.
+        hidden_states, presents, all_hidden_states, all_self_attentions = self.encoder(
+            inputs_embeds, full_attention_mask, rotary_pos_emb=rotary_pos_emb,
+            kv_caches=past_key_values, use_cache=use_cache, output_hidden_states=output_hidden_states
+        )
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, presents, all_hidden_states, all_self_attentions] if v is not None)
+
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=presents,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+        )
+
+    def quantize(self, weight_bit_width: int):
+        from .quantization import quantize
+        quantize(self.encoder, weight_bit_width)
+        return self
+
+
+class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.max_sequence_length = config.max_length
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+        self.config = config
+        self.quantized = False
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def _update_model_kwargs_for_generation(
+            self,
+            outputs: ModelOutput,
+            model_kwargs: Dict[str, Any],
+            is_encoder_decoder: bool = False,
+            standardize_cache_format: bool = False,
+    ) -> Dict[str, Any]:
+        # update past_key_values
+        model_kwargs["past_key_values"] = self._extract_past_from_model_output(
+            outputs, standardize_cache_format=standardize_cache_format
+        )
+
+        # update attention mask
+        if "attention_mask" in model_kwargs:
+            attention_mask = model_kwargs["attention_mask"]
+            model_kwargs["attention_mask"] = torch.cat(
+                [attention_mask, attention_mask.new_ones((attention_mask.shape[0], 1))], dim=-1
+            )
+
+        # update position ids
+        if "position_ids" in model_kwargs:
+            position_ids = model_kwargs["position_ids"]
+            new_position_id = position_ids[..., -1:].clone()
+            new_position_id += 1
+            model_kwargs["position_ids"] = torch.cat(
+                [position_ids, new_position_id], dim=-1
+            )
+
+        model_kwargs["is_first_forward"] = False
+        return model_kwargs
+
+    def prepare_inputs_for_generation(
+            self,
+            input_ids: torch.LongTensor,
+            past_key_values: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            is_first_forward: bool = True,
+            **kwargs
+    ) -> dict:
+        # only last token for input_ids if past is not None
+        if position_ids is None:
+            position_ids = self.get_position_ids(input_ids, device=input_ids.device)
+        if not is_first_forward:
+            if past_key_values is not None:
+                position_ids = position_ids[..., -1:]
+                input_ids = input_ids[:, -1:]
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past_key_values,
+            "position_ids": position_ids,
+            "attention_mask": attention_mask,
+            "return_last_logit": True,
+            "use_cache": use_cache
+        }
+
+    def forward(
+            self,
+            input_ids: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[torch.FloatTensor]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            labels: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_attentions: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+            return_last_logit: Optional[bool] = False,
+    ):
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        if return_last_logit:
+            hidden_states = hidden_states[-1:]
+        lm_logits = self.transformer.output_layer(hidden_states)
+        lm_logits = lm_logits.transpose(0, 1).contiguous()
+
+        loss = None
+        if labels is not None:
+            lm_logits = lm_logits.to(torch.float32)
+
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss(ignore_index=-100)
+            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+
+            lm_logits = lm_logits.to(hidden_states.dtype)
+            loss = loss.to(hidden_states.dtype)
+
+        if not return_dict:
+            output = (lm_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+    @staticmethod
+    def _reorder_cache(
+            past: Tuple[Tuple[torch.Tensor, torch.Tensor], ...], beam_idx: torch.LongTensor
+    ) -> Tuple[Tuple[torch.Tensor, torch.Tensor], ...]:
+        """
+        This function is used to re-order the `past_key_values` cache if [`~PreTrainedModel.beam_search`] or
+        [`~PreTrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
+        beam_idx at every generation step.
+
+        Output shares the same memory storage as `past`.
+        """
+        return tuple(
+            (
+                layer_past[0].index_select(1, beam_idx.to(layer_past[0].device)),
+                layer_past[1].index_select(1, beam_idx.to(layer_past[1].device)),
+            )
+            for layer_past in past
+        )
+
+    def process_response(self, response):
+        response = response.strip()
+        response = response.replace("[[训练时间]]", "2023年")
+        return response
+
+    def build_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        prompt = tokenizer.build_prompt(query, history=history)
+        inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    def build_stream_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        if history:
+            prompt = "\n\n[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            input_ids = tokenizer.encode(prompt, add_special_tokens=False)
+            input_ids = input_ids[1:]
+            inputs = tokenizer.batch_encode_plus([(input_ids, None)], return_tensors="pt", add_special_tokens=False)
+        else:
+            prompt = "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    @torch.inference_mode()
+    def chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, max_length: int = 8192, num_beams=1,
+             do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "num_beams": num_beams, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        inputs = self.build_inputs(tokenizer, query, history=history)
+        outputs = self.generate(**inputs, **gen_kwargs)
+        outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+        response = tokenizer.decode(outputs)
+        response = self.process_response(response)
+        history = history + [(query, response)]
+        return response, history
+
+    @torch.inference_mode()
+    def stream_chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, past_key_values=None,
+                    max_length: int = 8192, do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None,
+                    return_past_key_values=False, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        if past_key_values is None and not return_past_key_values:
+            inputs = self.build_inputs(tokenizer, query, history=history)
+        else:
+            inputs = self.build_stream_inputs(tokenizer, query, history=history)
+        if past_key_values is not None:
+            past_length = past_key_values[0][0].shape[0]
+            if self.transformer.pre_seq_len is not None:
+                past_length -= self.transformer.pre_seq_len
+            inputs.position_ids += past_length
+            attention_mask = inputs.attention_mask
+            attention_mask = torch.cat((attention_mask.new_ones(1, past_length), attention_mask), dim=1)
+            inputs['attention_mask'] = attention_mask
+        for outputs in self.stream_generate(**inputs, past_key_values=past_key_values,
+                                            return_past_key_values=return_past_key_values, **gen_kwargs):
+            if return_past_key_values:
+                outputs, past_key_values = outputs
+            outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+            response = tokenizer.decode(outputs)
+            if response and response[-1] != "�":
+                response = self.process_response(response)
+                new_history = history + [(query, response)]
+                if return_past_key_values:
+                    yield response, new_history, past_key_values
+                else:
+                    yield response, new_history
+
+    @torch.inference_mode()
+    def stream_generate(
+            self,
+            input_ids,
+            generation_config: Optional[GenerationConfig] = None,
+            logits_processor: Optional[LogitsProcessorList] = None,
+            stopping_criteria: Optional[StoppingCriteriaList] = None,
+            prefix_allowed_tokens_fn: Optional[Callable[[int, torch.Tensor], List[int]]] = None,
+            return_past_key_values=False,
+            **kwargs,
+    ):
+        batch_size, input_ids_seq_length = input_ids.shape[0], input_ids.shape[-1]
+
+        if generation_config is None:
+            generation_config = self.generation_config
+        generation_config = copy.deepcopy(generation_config)
+        model_kwargs = generation_config.update(**kwargs)
+        model_kwargs["use_cache"] = generation_config.use_cache
+        bos_token_id, eos_token_id = generation_config.bos_token_id, generation_config.eos_token_id
+
+        if isinstance(eos_token_id, int):
+            eos_token_id = [eos_token_id]
+
+        has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
+        if has_default_max_length and generation_config.max_new_tokens is None:
+            warnings.warn(
+                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
+                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
+                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                UserWarning,
+            )
+        elif generation_config.max_new_tokens is not None:
+            generation_config.max_length = generation_config.max_new_tokens + input_ids_seq_length
+            if not has_default_max_length:
+                logger.warn(
+                    f"Both `max_new_tokens` (={generation_config.max_new_tokens}) and `max_length`(="
+                    f"{generation_config.max_length}) seem to have been set. `max_new_tokens` will take precedence. "
+                    "Please refer to the documentation for more information. "
+                    "(https://huggingface.co/docs/transformers/main/en/main_classes/text_generation)",
+                    UserWarning,
+                )
+
+        if input_ids_seq_length >= generation_config.max_length:
+            input_ids_string = "decoder_input_ids" if self.config.is_encoder_decoder else "input_ids"
+            logger.warning(
+                f"Input length of {input_ids_string} is {input_ids_seq_length}, but `max_length` is set to"
+                f" {generation_config.max_length}. This can lead to unexpected behavior. You should consider"
+                " increasing `max_new_tokens`."
+            )
+
+        # 2. Set generation parameters if not already defined
+        logits_processor = logits_processor if logits_processor is not None else LogitsProcessorList()
+        stopping_criteria = stopping_criteria if stopping_criteria is not None else StoppingCriteriaList()
+
+        logits_processor = self._get_logits_processor(
+            generation_config=generation_config,
+            input_ids_seq_length=input_ids_seq_length,
+            encoder_input_ids=input_ids,
+            prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
+            logits_processor=logits_processor,
+        )
+
+        stopping_criteria = self._get_stopping_criteria(
+            generation_config=generation_config, stopping_criteria=stopping_criteria
+        )
+        logits_warper = self._get_logits_warper(generation_config)
+
+        unfinished_sequences = input_ids.new(input_ids.shape[0]).fill_(1)
+        scores = None
+        while True:
+            model_inputs = self.prepare_inputs_for_generation(input_ids, **model_kwargs)
+            # forward pass to get next token
+            outputs = self(
+                **model_inputs,
+                return_dict=True,
+                output_attentions=False,
+                output_hidden_states=False,
+            )
+
+            next_token_logits = outputs.logits[:, -1, :]
+
+            # pre-process distribution
+            next_token_scores = logits_processor(input_ids, next_token_logits)
+            next_token_scores = logits_warper(input_ids, next_token_scores)
+
+            # sample
+            probs = nn.functional.softmax(next_token_scores, dim=-1)
+            if generation_config.do_sample:
+                next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
+            else:
+                next_tokens = torch.argmax(probs, dim=-1)
+
+            # update generated ids, model inputs, and length for next step
+            input_ids = torch.cat([input_ids, next_tokens[:, None]], dim=-1)
+            model_kwargs = self._update_model_kwargs_for_generation(
+                outputs, model_kwargs, is_encoder_decoder=self.config.is_encoder_decoder
+            )
+            unfinished_sequences = unfinished_sequences.mul((sum(next_tokens != i for i in eos_token_id)).long())
+            if return_past_key_values:
+                yield input_ids, outputs.past_key_values
+            else:
+                yield input_ids
+            # stop when each sentence is finished, or if we exceed the maximum length
+            if unfinished_sequences.max() == 0 or stopping_criteria(input_ids, scores):
+                break
+
+    def quantize(self, bits: int, empty_init=False, device=None, **kwargs):
+        if bits == 0:
+            return
+
+        from .quantization import quantize
+
+        if self.quantized:
+            logger.info("Already quantized.")
+            return self
+
+        self.quantized = True
+
+        self.config.quantization_bit = bits
+
+        self.transformer.encoder = quantize(self.transformer.encoder, bits, empty_init=empty_init, device=device,
+                                            **kwargs)
+        return self
+
+
+class ChatGLMForSequenceClassification(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.num_labels = config.num_labels
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+
+        self.classifier_head = nn.Linear(config.hidden_size, config.num_labels, bias=True, dtype=torch.half)
+        if config.classifier_dropout is not None:
+            self.dropout = nn.Dropout(config.classifier_dropout)
+        else:
+            self.dropout = None
+        self.config = config
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def forward(
+            self,
+            input_ids: Optional[torch.LongTensor] = None,
+            position_ids: Optional[torch.LongTensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            full_attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.LongTensor] = None,
+            labels: Optional[torch.LongTensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor, ...], SequenceClassifierOutputWithPast]:
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            full_attention_mask=full_attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        pooled_hidden_states = hidden_states[-1]
+        if self.dropout is not None:
+            pooled_hidden_states = self.dropout(pooled_hidden_states)
+        logits = self.classifier_head(pooled_hidden_states)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze().float(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits.float(), labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(logits.view(-1, self.num_labels).float(), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits.float(), labels.view(-1, self.num_labels))
+
+        if not return_dict:
+            output = (logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )

checkpoint-1000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:683581a470952333639ab32852612a097f78c7f0a93e0490bab027c65fea6271
+size 14681455

checkpoint-1000/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e9e5f0ed00486492dce195bfca9bdafe810811fbe498ad167145f41f3d1f89a
+size 7340861

checkpoint-1000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:637499f18ba38b03c7fc48a0c3d2efcb736559501b48a1e7306e5e2b1d2a20ce
+size 14575

checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b8b2da7849875ff5dc7c5782252c5e9dcc13e8a73ea1fba2a34327b3fe26b3a
+size 627

checkpoint-1000/special_tokens_map.json

@@ -0,0 +1 @@
+{}

checkpoint-1000/tokenization_chatglm.py

@@ -0,0 +1,257 @@
+import os
+import torch
+from typing import List, Optional, Union, Dict
+from sentencepiece import SentencePieceProcessor
+from transformers import PreTrainedTokenizer
+from transformers.utils import logging, PaddingStrategy
+from transformers.tokenization_utils_base import EncodedInput, BatchEncoding
+
+
+class SPTokenizer:
+    def __init__(self, model_path: str):
+        # reload tokenizer
+        assert os.path.isfile(model_path), model_path
+        self.sp_model = SentencePieceProcessor(model_file=model_path)
+
+        # BOS / EOS token IDs
+        self.n_words: int = self.sp_model.vocab_size()
+        self.bos_id: int = self.sp_model.bos_id()
+        self.eos_id: int = self.sp_model.eos_id()
+        self.pad_id: int = self.sp_model.unk_id()
+        assert self.sp_model.vocab_size() == self.sp_model.get_piece_size()
+
+        special_tokens = ["[MASK]", "[gMASK]", "[sMASK]", "sop", "eop"]
+        self.special_tokens = {}
+        self.index_special_tokens = {}
+        for token in special_tokens:
+            self.special_tokens[token] = self.n_words
+            self.index_special_tokens[self.n_words] = token
+            self.n_words += 1
+
+    def tokenize(self, s: str):
+        return self.sp_model.EncodeAsPieces(s)
+
+    def encode(self, s: str, bos: bool = False, eos: bool = False) -> List[int]:
+        assert type(s) is str
+        t = self.sp_model.encode(s)
+        if bos:
+            t = [self.bos_id] + t
+        if eos:
+            t = t + [self.eos_id]
+        return t
+
+    def decode(self, t: List[int]) -> str:
+        return self.sp_model.decode(t)
+
+    def decode_tokens(self, tokens: List[str]) -> str:
+        text = self.sp_model.DecodePieces(tokens)
+        return text
+
+    def convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        return self.sp_model.PieceToId(token)
+
+    def convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        if index in self.index_special_tokens or index in [self.eos_id, self.bos_id, self.pad_id] or index < 0:
+            return ""
+        return self.sp_model.IdToPiece(index)
+
+
+class ChatGLMTokenizer(PreTrainedTokenizer):
+    vocab_files_names = {"vocab_file": "tokenizer.model"}
+
+    model_input_names = ["input_ids", "attention_mask", "position_ids"]
+
+    def __init__(self, vocab_file, padding_side="left", clean_up_tokenization_spaces=False, **kwargs):
+        self.name = "GLMTokenizer"
+
+        self.vocab_file = vocab_file
+        self.tokenizer = SPTokenizer(vocab_file)
+        self.special_tokens = {
+            "<bos>": self.tokenizer.bos_id,
+            "<eos>": self.tokenizer.eos_id,
+            "<pad>": self.tokenizer.pad_id
+        }
+        super().__init__(padding_side=padding_side, clean_up_tokenization_spaces=clean_up_tokenization_spaces, **kwargs)
+
+    def get_command(self, token):
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        assert token in self.tokenizer.special_tokens, f"{token} is not a special token for {self.name}"
+        return self.tokenizer.special_tokens[token]
+
+    @property
+    def unk_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token_id(self):
+        return self.get_command("<pad>")
+
+    @property
+    def eos_token(self) -> str:
+        return "</s>"
+
+    @property
+    def eos_token_id(self):
+        return self.get_command("<eos>")
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_words
+
+    def get_vocab(self):
+        """ Returns vocab as a dict """
+        vocab = {self._convert_id_to_token(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text, **kwargs):
+        return self.tokenizer.tokenize(text)
+
+    def _convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        return self.tokenizer.convert_token_to_id(token)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        return self.tokenizer.convert_id_to_token(index)
+
+    def convert_tokens_to_string(self, tokens: List[str]) -> str:
+        return self.tokenizer.decode_tokens(tokens)
+
+    def save_vocabulary(self, save_directory, filename_prefix=None):
+        """
+        Save the vocabulary and special tokens file to a directory.
+
+        Args:
+            save_directory (`str`):
+                The directory in which to save the vocabulary.
+            filename_prefix (`str`, *optional*):
+                An optional prefix to add to the named of the saved files.
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        if os.path.isdir(save_directory):
+            vocab_file = os.path.join(
+                save_directory, self.vocab_files_names["vocab_file"]
+            )
+        else:
+            vocab_file = save_directory
+
+        with open(self.vocab_file, 'rb') as fin:
+            proto_str = fin.read()
+
+        with open(vocab_file, "wb") as writer:
+            writer.write(proto_str)
+
+        return (vocab_file,)
+
+    def get_prefix_tokens(self):
+        prefix_tokens = [self.get_command("[gMASK]"), self.get_command("sop")]
+        return prefix_tokens
+
+    def build_prompt(self, query, history=None):
+        if history is None:
+            history = []
+        prompt = ""
+        for i, (old_query, response) in enumerate(history):
+            prompt += "[Round {}]\n\n问：{}\n\n答：{}\n\n".format(i + 1, old_query, response)
+        prompt += "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+        return prompt
+
+    def build_inputs_with_special_tokens(
+            self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. A BERT sequence has the following format:
+
+        - single sequence: `[CLS] X [SEP]`
+        - pair of sequences: `[CLS] A [SEP] B [SEP]`
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        prefix_tokens = self.get_prefix_tokens()
+        token_ids_0 = prefix_tokens + token_ids_0
+        if token_ids_1 is not None:
+            token_ids_0 = token_ids_0 + token_ids_1 + [self.get_command("<eos>")]
+        return token_ids_0
+
+    def _pad(
+            self,
+            encoded_inputs: Union[Dict[str, EncodedInput], BatchEncoding],
+            max_length: Optional[int] = None,
+            padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
+            pad_to_multiple_of: Optional[int] = None,
+            return_attention_mask: Optional[bool] = None,
+    ) -> dict:
+        """
+        Pad encoded inputs (on left/right and up to predefined length or max length in the batch)
+
+        Args:
+            encoded_inputs:
+                Dictionary of tokenized inputs (`List[int]`) or batch of tokenized inputs (`List[List[int]]`).
+            max_length: maximum length of the returned list and optionally padding length (see below).
+                Will truncate by taking into account the special tokens.
+            padding_strategy: PaddingStrategy to use for padding.
+
+                - PaddingStrategy.LONGEST Pad to the longest sequence in the batch
+                - PaddingStrategy.MAX_LENGTH: Pad to the max length (default)
+                - PaddingStrategy.DO_NOT_PAD: Do not pad
+                The tokenizer padding sides are defined in self.padding_side:
+
+                    - 'left': pads on the left of the sequences
+                    - 'right': pads on the right of the sequences
+            pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
+                This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
+                `>= 7.5` (Volta).
+            return_attention_mask:
+                (optional) Set to False to avoid returning attention mask (default: set to model specifics)
+        """
+        # Load from model defaults
+        assert self.padding_side == "left"
+
+        required_input = encoded_inputs[self.model_input_names[0]]
+        seq_length = len(required_input)
+
+        if padding_strategy == PaddingStrategy.LONGEST:
+            max_length = len(required_input)
+
+        if max_length is not None and pad_to_multiple_of is not None and (max_length % pad_to_multiple_of != 0):
+            max_length = ((max_length // pad_to_multiple_of) + 1) * pad_to_multiple_of
+
+        needs_to_be_padded = padding_strategy != PaddingStrategy.DO_NOT_PAD and len(required_input) != max_length
+
+        # Initialize attention mask if not present.
+        if "attention_mask" not in encoded_inputs:
+            encoded_inputs["attention_mask"] = [1] * seq_length
+
+        if "position_ids" not in encoded_inputs:
+            encoded_inputs["position_ids"] = list(range(seq_length))
+
+        if needs_to_be_padded:
+            difference = max_length - len(required_input)
+
+            if "attention_mask" in encoded_inputs:
+                encoded_inputs["attention_mask"] = [0] * difference + encoded_inputs["attention_mask"]
+            if "position_ids" in encoded_inputs:
+                encoded_inputs["position_ids"] = [0] * difference + encoded_inputs["position_ids"]
+            encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input
+
+        return encoded_inputs

checkpoint-1000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7dc4c393423b76e4373e5157ddc34803a0189ba96b21ddbb40269d31468a6f2
+size 1018370

checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_chatglm.ChatGLMTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "padding_side": "left",
+  "remove_space": false,
+  "tokenizer_class": "ChatGLMTokenizer"
+}

checkpoint-1000/trainer_state.json

@@ -0,0 +1,616 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.09235955575053684,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019933333333333334,
+      "loss": 2.6644,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019866666666666668,
+      "loss": 1.7151,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0198,
+      "loss": 1.6228,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019733333333333335,
+      "loss": 1.401,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019666666666666666,
+      "loss": 1.6172,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0196,
+      "loss": 1.4695,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019533333333333333,
+      "loss": 1.5137,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019466666666666667,
+      "loss": 1.5425,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0194,
+      "loss": 1.4272,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019333333333333334,
+      "loss": 1.3727,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019266666666666668,
+      "loss": 1.3114,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0192,
+      "loss": 1.4758,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019133333333333332,
+      "loss": 1.5219,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.01906666666666667,
+      "loss": 1.376,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019,
+      "loss": 1.4257,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.018933333333333333,
+      "loss": 1.3474,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018866666666666667,
+      "loss": 1.2929,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0188,
+      "loss": 1.3208,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018733333333333334,
+      "loss": 1.3381,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018666666666666668,
+      "loss": 1.3644,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018600000000000002,
+      "loss": 1.2932,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018533333333333332,
+      "loss": 1.4092,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018466666666666666,
+      "loss": 1.3006,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0184,
+      "loss": 1.4572,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018333333333333333,
+      "loss": 1.2789,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018266666666666667,
+      "loss": 1.4444,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0182,
+      "loss": 1.4511,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01813333333333333,
+      "loss": 1.3541,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01806666666666667,
+      "loss": 1.3228,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.018000000000000002,
+      "loss": 1.3185,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017933333333333332,
+      "loss": 1.199,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017866666666666666,
+      "loss": 1.3417,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0178,
+      "loss": 1.4251,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017733333333333334,
+      "loss": 1.3574,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017666666666666667,
+      "loss": 1.2547,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0176,
+      "loss": 1.2651,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017533333333333335,
+      "loss": 1.3414,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017466666666666665,
+      "loss": 1.3322,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0174,
+      "loss": 1.4147,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017333333333333336,
+      "loss": 1.2813,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017266666666666666,
+      "loss": 1.3687,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0172,
+      "loss": 1.5593,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017133333333333334,
+      "loss": 1.3073,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017066666666666667,
+      "loss": 1.2359,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017,
+      "loss": 1.2474,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.016933333333333335,
+      "loss": 1.3874,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.01686666666666667,
+      "loss": 1.3203,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0168,
+      "loss": 1.2875,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016733333333333333,
+      "loss": 1.2767,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016666666666666666,
+      "loss": 1.3017,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0166,
+      "loss": 1.2321,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016533333333333334,
+      "loss": 1.1719,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016466666666666668,
+      "loss": 1.2552,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016399999999999998,
+      "loss": 1.3816,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01633333333333333,
+      "loss": 1.2956,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01626666666666667,
+      "loss": 1.2061,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016200000000000003,
+      "loss": 1.2086,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016133333333333333,
+      "loss": 1.1633,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016066666666666667,
+      "loss": 1.2638,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.016,
+      "loss": 1.3441,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015933333333333334,
+      "loss": 1.2924,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015866666666666668,
+      "loss": 1.1818,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0158,
+      "loss": 1.3918,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015733333333333332,
+      "loss": 1.2232,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015666666666666666,
+      "loss": 1.2472,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015600000000000001,
+      "loss": 1.2398,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015533333333333333,
+      "loss": 1.3649,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015466666666666667,
+      "loss": 1.2302,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0154,
+      "loss": 1.2053,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015333333333333334,
+      "loss": 1.2974,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015266666666666666,
+      "loss": 1.3036,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0152,
+      "loss": 1.3162,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015133333333333334,
+      "loss": 1.2567,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015066666666666666,
+      "loss": 1.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015,
+      "loss": 1.2692,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014933333333333335,
+      "loss": 1.1332,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014866666666666667,
+      "loss": 1.2949,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0148,
+      "loss": 1.2703,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014733333333333334,
+      "loss": 1.3891,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014666666666666666,
+      "loss": 1.3594,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0146,
+      "loss": 1.166,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014533333333333334,
+      "loss": 1.3256,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014466666666666668,
+      "loss": 1.2669,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0144,
+      "loss": 1.241,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014333333333333333,
+      "loss": 1.2591,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014266666666666667,
+      "loss": 1.238,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014199999999999999,
+      "loss": 1.3583,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014133333333333333,
+      "loss": 1.164,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014066666666666668,
+      "loss": 1.2367,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013999999999999999,
+      "loss": 1.1864,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013933333333333334,
+      "loss": 1.2259,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013866666666666668,
+      "loss": 1.2129,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0138,
+      "loss": 1.2085,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013733333333333334,
+      "loss": 1.2316,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013666666666666667,
+      "loss": 1.2721,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013600000000000001,
+      "loss": 1.2428,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013533333333333333,
+      "loss": 1.2126,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013466666666666667,
+      "loss": 1.1583,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0134,
+      "loss": 1.2776,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013333333333333332,
+      "loss": 1.2317,
+      "step": 1000
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 1,
+  "total_flos": 5.7864771305472e+16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70ac9eb43a2d446e07cde6bbbb21250fe0373a4093de76a8aa1f7223e3836bcd
+size 4155

checkpoint-2000/config.json

@@ -0,0 +1,44 @@
+{
+  "_name_or_path": "E:/PycharmProjects/dl_models/chatglm2-6b-int4",
+  "add_bias_linear": false,
+  "add_qkv_bias": true,
+  "apply_query_key_layer_scaling": true,
+  "apply_residual_connection_post_layernorm": false,
+  "architectures": [
+    "ChatGLMForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "attention_softmax_in_fp32": true,
+  "auto_map": {
+    "AutoConfig": "configuration_chatglm.ChatGLMConfig",
+    "AutoModel": "modeling_chatglm.ChatGLMForConditionalGeneration",
+    "AutoModelForSeq2SeqLM": "modeling_chatglm.ChatGLMForConditionalGeneration"
+  },
+  "bias_dropout_fusion": true,
+  "eos_token_id": 2,
+  "ffn_hidden_size": 13696,
+  "fp32_residual_connection": false,
+  "hidden_dropout": 0.0,
+  "hidden_size": 4096,
+  "kv_channels": 128,
+  "layernorm_epsilon": 1e-05,
+  "model_type": "chatglm",
+  "multi_query_attention": true,
+  "multi_query_group_num": 2,
+  "num_attention_heads": 32,
+  "num_layers": 28,
+  "original_rope": true,
+  "pad_token_id": 0,
+  "padded_vocab_size": 65024,
+  "post_layer_norm": true,
+  "pre_seq_len": 128,
+  "prefix_projection": false,
+  "quantization_bit": 4,
+  "rmsnorm": true,
+  "seq_length": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.31.0",
+  "use_cache": true,
+  "vocab_size": 65024
+}

checkpoint-2000/configuration_chatglm.py

@@ -0,0 +1,59 @@
+from transformers import PretrainedConfig
+
+
+class ChatGLMConfig(PretrainedConfig):
+    model_type = "chatglm"
+    def __init__(
+        self,
+        num_layers=28,
+        padded_vocab_size=65024,
+        hidden_size=4096,
+        ffn_hidden_size=13696,
+        kv_channels=128,
+        num_attention_heads=32,
+        seq_length=2048,
+        hidden_dropout=0.0,
+        attention_dropout=0.0,
+        layernorm_epsilon=1e-5,
+        rmsnorm=True,
+        apply_residual_connection_post_layernorm=False,
+        post_layer_norm=True,
+        add_bias_linear=False,
+        add_qkv_bias=False,
+        bias_dropout_fusion=True,
+        multi_query_attention=False,
+        multi_query_group_num=1,
+        apply_query_key_layer_scaling=True,
+        attention_softmax_in_fp32=True,
+        fp32_residual_connection=False,
+        quantization_bit=0,
+        pre_seq_len=None,
+        prefix_projection=False,
+        **kwargs
+    ):
+        self.num_layers = num_layers
+        self.vocab_size = padded_vocab_size
+        self.padded_vocab_size = padded_vocab_size
+        self.hidden_size = hidden_size
+        self.ffn_hidden_size = ffn_hidden_size
+        self.kv_channels = kv_channels
+        self.num_attention_heads = num_attention_heads
+        self.seq_length = seq_length
+        self.hidden_dropout = hidden_dropout
+        self.attention_dropout = attention_dropout
+        self.layernorm_epsilon = layernorm_epsilon
+        self.rmsnorm = rmsnorm
+        self.apply_residual_connection_post_layernorm = apply_residual_connection_post_layernorm
+        self.post_layer_norm = post_layer_norm
+        self.add_bias_linear = add_bias_linear
+        self.add_qkv_bias = add_qkv_bias
+        self.bias_dropout_fusion = bias_dropout_fusion
+        self.multi_query_attention = multi_query_attention
+        self.multi_query_group_num = multi_query_group_num
+        self.apply_query_key_layer_scaling = apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = attention_softmax_in_fp32
+        self.fp32_residual_connection = fp32_residual_connection
+        self.quantization_bit = quantization_bit
+        self.pre_seq_len = pre_seq_len
+        self.prefix_projection = prefix_projection
+        super().__init__(**kwargs)

checkpoint-2000/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.31.0"
+}

checkpoint-2000/modeling_chatglm.py

@@ -0,0 +1,1285 @@
+""" PyTorch ChatGLM model. """
+
+import math
+import copy
+import warnings
+import re
+import sys
+
+import torch
+import torch.utils.checkpoint
+import torch.nn.functional as F
+from torch import nn
+from torch.nn import CrossEntropyLoss, LayerNorm
+from torch.nn import CrossEntropyLoss, LayerNorm, MSELoss, BCEWithLogitsLoss
+from torch.nn.utils import skip_init
+from typing import Optional, Tuple, Union, List, Callable, Dict, Any
+
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+from transformers.generation.logits_process import LogitsProcessor
+from transformers.generation.utils import LogitsProcessorList, StoppingCriteriaList, GenerationConfig, ModelOutput
+
+from .configuration_chatglm import ChatGLMConfig
+
+# flags required to enable jit fusion kernels
+
+if sys.platform != 'darwin':
+    torch._C._jit_set_profiling_mode(False)
+    torch._C._jit_set_profiling_executor(False)
+    torch._C._jit_override_can_fuse_on_cpu(True)
+    torch._C._jit_override_can_fuse_on_gpu(True)
+
+logger = logging.get_logger(__name__)
+
+_CHECKPOINT_FOR_DOC = "THUDM/ChatGLM2-6B"
+_CONFIG_FOR_DOC = "ChatGLM6BConfig"
+
+CHATGLM_6B_PRETRAINED_MODEL_ARCHIVE_LIST = [
+    "THUDM/chatglm2-6b",
+    # See all ChatGLM models at https://huggingface.co/models?filter=chatglm
+]
+
+
+def default_init(cls, *args, **kwargs):
+    return cls(*args, **kwargs)
+
+
+class InvalidScoreLogitsProcessor(LogitsProcessor):
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
+        if torch.isnan(scores).any() or torch.isinf(scores).any():
+            scores.zero_()
+            scores[..., 5] = 5e4
+        return scores
+
+
+class PrefixEncoder(torch.nn.Module):
+    """
+    The torch.nn model to encode the prefix
+    Input shape: (batch-size, prefix-length)
+    Output shape: (batch-size, prefix-length, 2*layers*hidden)
+    """
+
+    def __init__(self, config: ChatGLMConfig):
+        super().__init__()
+        self.prefix_projection = config.prefix_projection
+        if self.prefix_projection:
+            # Use a two-layer MLP to encode the prefix
+            kv_size = config.num_layers * config.kv_channels * config.multi_query_group_num * 2
+            self.embedding = torch.nn.Embedding(config.pre_seq_len, kv_size)
+            self.trans = torch.nn.Sequential(
+                torch.nn.Linear(kv_size, config.hidden_size),
+                torch.nn.Tanh(),
+                torch.nn.Linear(config.hidden_size, kv_size)
+            )
+        else:
+            self.embedding = torch.nn.Embedding(config.pre_seq_len,
+                                                config.num_layers * config.kv_channels * config.multi_query_group_num * 2)
+
+    def forward(self, prefix: torch.Tensor):
+        if self.prefix_projection:
+            prefix_tokens = self.embedding(prefix)
+            past_key_values = self.trans(prefix_tokens)
+        else:
+            past_key_values = self.embedding(prefix)
+        return past_key_values
+
+
+def split_tensor_along_last_dim(
+        tensor: torch.Tensor,
+        num_partitions: int,
+        contiguous_split_chunks: bool = False,
+) -> List[torch.Tensor]:
+    """Split a tensor along its last dimension.
+
+    Arguments:
+        tensor: input tensor.
+        num_partitions: number of partitions to split the tensor
+        contiguous_split_chunks: If True, make each chunk contiguous
+                                 in memory.
+
+    Returns:
+        A list of Tensors
+    """
+    # Get the size and dimension.
+    last_dim = tensor.dim() - 1
+    last_dim_size = tensor.size()[last_dim] // num_partitions
+    # Split.
+    tensor_list = torch.split(tensor, last_dim_size, dim=last_dim)
+    # Note: torch.split does not create contiguous tensors by default.
+    if contiguous_split_chunks:
+        return tuple(chunk.contiguous() for chunk in tensor_list)
+
+    return tensor_list
+
+
+class RotaryEmbedding(nn.Module):
+    def __init__(self, dim, original_impl=False, device=None, dtype=None):
+        super().__init__()
+        inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2, device=device).to(dtype=dtype) / dim))
+        self.register_buffer("inv_freq", inv_freq)
+        self.dim = dim
+        self.original_impl = original_impl
+
+    def forward_impl(
+            self, seq_len: int, n_elem: int, dtype: torch.dtype, device: torch.device, base: int = 10000
+    ):
+        """Enhanced Transformer with Rotary Position Embedding.
+
+        Derived from: https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/labml_nn/
+        transformers/rope/__init__.py. MIT License:
+        https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/license.
+        """
+        # $\Theta = {\theta_i = 10000^{\frac{2(i-1)}{d}}, i \in [1, 2, ..., \frac{d}{2}]}$
+        theta = 1.0 / (base ** (torch.arange(0, n_elem, 2, dtype=dtype, device=device) / n_elem))
+
+        # Create position indexes `[0, 1, ..., seq_len - 1]`
+        seq_idx = torch.arange(seq_len, dtype=dtype, device=device)
+
+        # Calculate the product of position index and $\theta_i$
+        idx_theta = torch.outer(seq_idx, theta).float()
+
+        cache = torch.stack([torch.cos(idx_theta), torch.sin(idx_theta)], dim=-1)
+
+        # this is to mimic the behaviour of complex32, else we will get different results
+        if dtype in (torch.float16, torch.bfloat16, torch.int8):
+            cache = cache.bfloat16() if dtype == torch.bfloat16 else cache.half()
+        return cache
+
+    def forward(self, max_seq_len, offset=0):
+        return self.forward_impl(
+            max_seq_len, self.dim, dtype=self.inv_freq.dtype, device=self.inv_freq.device
+        )
+
+
+@torch.jit.script
+def apply_rotary_pos_emb(x: torch.Tensor, rope_cache: torch.Tensor) -> torch.Tensor:
+    # x: [sq, b, np, hn]
+    sq, b, np, hn = x.size(0), x.size(1), x.size(2), x.size(3)
+    rot_dim = rope_cache.shape[-2] * 2
+    x, x_pass = x[..., :rot_dim], x[..., rot_dim:]
+    # truncate to support variable sizes
+    rope_cache = rope_cache[:sq]
+    xshaped = x.reshape(sq, -1, np, rot_dim // 2, 2)
+    rope_cache = rope_cache.view(sq, -1, 1, xshaped.size(3), 2)
+    x_out2 = torch.stack(
+        [
+            xshaped[..., 0] * rope_cache[..., 0] - xshaped[..., 1] * rope_cache[..., 1],
+            xshaped[..., 1] * rope_cache[..., 0] + xshaped[..., 0] * rope_cache[..., 1],
+        ],
+        -1,
+    )
+    x_out2 = x_out2.flatten(3)
+    return torch.cat((x_out2, x_pass), dim=-1)
+
+
+class RMSNorm(torch.nn.Module):
+    def __init__(self, normalized_shape, eps=1e-5, device=None, dtype=None, **kwargs):
+        super().__init__()
+        self.weight = torch.nn.Parameter(torch.empty(normalized_shape, device=device, dtype=dtype))
+        self.eps = eps
+
+    def forward(self, hidden_states: torch.Tensor):
+        input_dtype = hidden_states.dtype
+        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
+
+        return (self.weight * hidden_states).to(input_dtype)
+
+
+class CoreAttention(torch.nn.Module):
+    def __init__(self, config: ChatGLMConfig, layer_number):
+        super(CoreAttention, self).__init__()
+
+        self.apply_query_key_layer_scaling = config.apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
+        if self.apply_query_key_layer_scaling:
+            self.attention_softmax_in_fp32 = True
+        self.layer_number = max(1, layer_number)
+
+        projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_partition = projection_size
+        self.hidden_size_per_attention_head = projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        coeff = None
+        self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
+        if self.apply_query_key_layer_scaling:
+            coeff = self.layer_number
+            self.norm_factor *= coeff
+        self.coeff = coeff
+
+        self.attention_dropout = torch.nn.Dropout(config.attention_dropout)
+
+    def forward(self, query_layer, key_layer, value_layer, attention_mask):
+        pytorch_major_version = int(torch.__version__.split('.')[0])
+        if pytorch_major_version >= 2:
+            query_layer, key_layer, value_layer = [k.permute(1, 2, 0, 3) for k in [query_layer, key_layer, value_layer]]
+            if attention_mask is None and query_layer.shape[2] == key_layer.shape[2]:
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 is_causal=True)
+            else:
+                if attention_mask is not None:
+                    attention_mask = ~attention_mask
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 attention_mask)
+            context_layer = context_layer.permute(2, 0, 1, 3)
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.reshape(*new_context_layer_shape)
+        else:
+            # Raw attention scores
+
+            # [b, np, sq, sk]
+            output_size = (query_layer.size(1), query_layer.size(2), query_layer.size(0), key_layer.size(0))
+
+            # [sq, b, np, hn] -> [sq, b * np, hn]
+            query_layer = query_layer.view(output_size[2], output_size[0] * output_size[1], -1)
+            # [sk, b, np, hn] -> [sk, b * np, hn]
+            key_layer = key_layer.view(output_size[3], output_size[0] * output_size[1], -1)
+
+            # preallocting input tensor: [b * np, sq, sk]
+            matmul_input_buffer = torch.empty(
+                output_size[0] * output_size[1], output_size[2], output_size[3], dtype=query_layer.dtype,
+                device=query_layer.device
+            )
+
+            # Raw attention scores. [b * np, sq, sk]
+            matmul_result = torch.baddbmm(
+                matmul_input_buffer,
+                query_layer.transpose(0, 1),  # [b * np, sq, hn]
+                key_layer.transpose(0, 1).transpose(1, 2),  # [b * np, hn, sk]
+                beta=0.0,
+                alpha=(1.0 / self.norm_factor),
+            )
+
+            # change view to [b, np, sq, sk]
+            attention_scores = matmul_result.view(*output_size)
+
+            # ===========================
+            # Attention probs and dropout
+            # ===========================
+
+            # attention scores and attention mask [b, np, sq, sk]
+            if self.attention_softmax_in_fp32:
+                attention_scores = attention_scores.float()
+            if self.coeff is not None:
+                attention_scores = attention_scores * self.coeff
+            if attention_mask is None and attention_scores.shape[2] == attention_scores.shape[3]:
+                attention_mask = torch.ones(output_size[0], 1, output_size[2], output_size[3],
+                                            device=attention_scores.device, dtype=torch.bool)
+                attention_mask.tril_()
+                attention_mask = ~attention_mask
+            if attention_mask is not None:
+                attention_scores = attention_scores.masked_fill(attention_mask, float("-inf"))
+            attention_probs = F.softmax(attention_scores, dim=-1)
+            attention_probs = attention_probs.type_as(value_layer)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.attention_dropout(attention_probs)
+            # =========================
+            # Context layer. [sq, b, hp]
+            # =========================
+
+            # value_layer -> context layer.
+            # [sk, b, np, hn] --> [b, np, sq, hn]
+
+            # context layer shape: [b, np, sq, hn]
+            output_size = (value_layer.size(1), value_layer.size(2), query_layer.size(0), value_layer.size(3))
+            # change view [sk, b * np, hn]
+            value_layer = value_layer.view(value_layer.size(0), output_size[0] * output_size[1], -1)
+            # change view [b * np, sq, sk]
+            attention_probs = attention_probs.view(output_size[0] * output_size[1], output_size[2], -1)
+            # matmul: [b * np, sq, hn]
+            context_layer = torch.bmm(attention_probs, value_layer.transpose(0, 1))
+            # change view [b, np, sq, hn]
+            context_layer = context_layer.view(*output_size)
+            # [b, np, sq, hn] --> [sq, b, np, hn]
+            context_layer = context_layer.permute(2, 0, 1, 3).contiguous()
+            # [sq, b, np, hn] --> [sq, b, hp]
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.view(*new_context_layer_shape)
+
+        return context_layer
+
+
+class SelfAttention(torch.nn.Module):
+    """Parallel self-attention layer abstract class.
+
+    Self-attention layer takes input with size [s, b, h]
+    and returns output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(SelfAttention, self).__init__()
+        self.layer_number = max(1, layer_number)
+
+        self.projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_attention_head = self.projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        self.multi_query_attention = config.multi_query_attention
+        self.qkv_hidden_size = 3 * self.projection_size
+        if self.multi_query_attention:
+            self.num_multi_query_groups_per_partition = config.multi_query_group_num
+            self.qkv_hidden_size = (
+                    self.projection_size + 2 * self.hidden_size_per_attention_head * config.multi_query_group_num
+            )
+        self.query_key_value = nn.Linear(config.hidden_size, self.qkv_hidden_size,
+                                         bias=config.add_bias_linear or config.add_qkv_bias,
+                                         device=device, **_config_to_kwargs(config)
+                                         )
+
+        self.core_attention = CoreAttention(config, self.layer_number)
+
+        # Output.
+        self.dense = nn.Linear(self.projection_size, config.hidden_size, bias=config.add_bias_linear,
+                               device=device, **_config_to_kwargs(config)
+                               )
+
+    def _allocate_memory(self, inference_max_sequence_len, batch_size, device=None, dtype=None):
+        if self.multi_query_attention:
+            num_attention_heads = self.num_multi_query_groups_per_partition
+        else:
+            num_attention_heads = self.num_attention_heads_per_partition
+        return torch.empty(
+            inference_max_sequence_len,
+            batch_size,
+            num_attention_heads,
+            self.hidden_size_per_attention_head,
+            dtype=dtype,
+            device=device,
+        )
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True
+    ):
+        # hidden_states: [sq, b, h]
+
+        # =================================================
+        # Pre-allocate memory for key-values for inference.
+        # =================================================
+        # =====================
+        # Query, Key, and Value
+        # =====================
+
+        # Attention heads [sq, b, h] --> [sq, b, (np * 3 * hn)]
+        mixed_x_layer = self.query_key_value(hidden_states)
+
+        if self.multi_query_attention:
+            (query_layer, key_layer, value_layer) = mixed_x_layer.split(
+                [
+                    self.num_attention_heads_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                ],
+                dim=-1,
+            )
+            query_layer = query_layer.view(
+                query_layer.size()[:-1] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            key_layer = key_layer.view(
+                key_layer.size()[:-1] + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.view(
+                value_layer.size()[:-1]
+                + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+        else:
+            new_tensor_shape = mixed_x_layer.size()[:-1] + \
+                               (self.num_attention_heads_per_partition,
+                                3 * self.hidden_size_per_attention_head)
+            mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
+
+            # [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
+            (query_layer, key_layer, value_layer) = split_tensor_along_last_dim(mixed_x_layer, 3)
+
+        # apply relative positional encoding (rotary embedding)
+        if rotary_pos_emb is not None:
+            query_layer = apply_rotary_pos_emb(query_layer, rotary_pos_emb)
+            key_layer = apply_rotary_pos_emb(key_layer, rotary_pos_emb)
+
+        # adjust key and value for inference
+        if kv_cache is not None:
+            cache_k, cache_v = kv_cache
+            key_layer = torch.cat((cache_k, key_layer), dim=0)
+            value_layer = torch.cat((cache_v, value_layer), dim=0)
+        if use_cache:
+            kv_cache = (key_layer, value_layer)
+        else:
+            kv_cache = None
+
+        if self.multi_query_attention:
+            key_layer = key_layer.unsqueeze(-2)
+            key_layer = key_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            key_layer = key_layer.contiguous().view(
+                key_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.unsqueeze(-2)
+            value_layer = value_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            value_layer = value_layer.contiguous().view(
+                value_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+
+        # ==================================
+        # core attention computation
+        # ==================================
+
+        context_layer = self.core_attention(query_layer, key_layer, value_layer, attention_mask)
+
+        # =================
+        # Output. [sq, b, h]
+        # =================
+
+        output = self.dense(context_layer)
+
+        return output, kv_cache
+
+
+def _config_to_kwargs(args):
+    common_kwargs = {
+        "dtype": args.torch_dtype,
+    }
+    return common_kwargs
+
+
+class MLP(torch.nn.Module):
+    """MLP.
+
+    MLP will take the input with h hidden state, project it to 4*h
+    hidden dimension, perform nonlinear transformation, and project the
+    state back into h hidden dimension.
+    """
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(MLP, self).__init__()
+
+        self.add_bias = config.add_bias_linear
+
+        # Project to 4h. If using swiglu double the output width, see https://arxiv.org/pdf/2002.05202.pdf
+        self.dense_h_to_4h = nn.Linear(
+            config.hidden_size,
+            config.ffn_hidden_size * 2,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+        def swiglu(x):
+            x = torch.chunk(x, 2, dim=-1)
+            return F.silu(x[0]) * x[1]
+
+        self.activation_func = swiglu
+
+        # Project back to h.
+        self.dense_4h_to_h = nn.Linear(
+            config.ffn_hidden_size,
+            config.hidden_size,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+    def forward(self, hidden_states):
+        # [s, b, 4hp]
+        intermediate_parallel = self.dense_h_to_4h(hidden_states)
+        intermediate_parallel = self.activation_func(intermediate_parallel)
+        # [s, b, h]
+        output = self.dense_4h_to_h(intermediate_parallel)
+        return output
+
+
+class GLMBlock(torch.nn.Module):
+    """A single transformer layer.
+
+    Transformer layer takes input with size [s, b, h] and returns an
+    output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(GLMBlock, self).__init__()
+        self.layer_number = layer_number
+
+        self.apply_residual_connection_post_layernorm = config.apply_residual_connection_post_layernorm
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+        LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+        # Layernorm on the input data.
+        self.input_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                             dtype=config.torch_dtype)
+
+        # Self attention.
+        self.self_attention = SelfAttention(config, layer_number, device=device)
+        self.hidden_dropout = config.hidden_dropout
+
+        # Layernorm on the attention output
+        self.post_attention_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                      dtype=config.torch_dtype)
+
+        # MLP
+        self.mlp = MLP(config, device=device)
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True,
+    ):
+        # hidden_states: [s, b, h]
+
+        # Layer norm at the beginning of the transformer layer.
+        layernorm_output = self.input_layernorm(hidden_states)
+        # Self attention.
+        attention_output, kv_cache = self.self_attention(
+            layernorm_output,
+            attention_mask,
+            rotary_pos_emb,
+            kv_cache=kv_cache,
+            use_cache=use_cache
+        )
+
+        # Residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = hidden_states
+
+        layernorm_input = torch.nn.functional.dropout(attention_output, p=self.hidden_dropout, training=self.training)
+        layernorm_input = residual + layernorm_input
+
+        # Layer norm post the self attention.
+        layernorm_output = self.post_attention_layernorm(layernorm_input)
+
+        # MLP.
+        mlp_output = self.mlp(layernorm_output)
+
+        # Second residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = layernorm_input
+
+        output = torch.nn.functional.dropout(mlp_output, p=self.hidden_dropout, training=self.training)
+        output = residual + output
+
+        return output, kv_cache
+
+
+class GLMTransformer(torch.nn.Module):
+    """Transformer class."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(GLMTransformer, self).__init__()
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+        self.post_layer_norm = config.post_layer_norm
+
+        # Number of layers.
+        self.num_layers = config.num_layers
+
+        # Transformer layers.
+        def build_layer(layer_number):
+            return GLMBlock(config, layer_number, device=device)
+
+        self.layers = torch.nn.ModuleList([build_layer(i + 1) for i in range(self.num_layers)])
+
+        if self.post_layer_norm:
+            LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+            # Final layer norm before output.
+            self.final_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                 dtype=config.torch_dtype)
+
+        self.gradient_checkpointing = False
+
+    def _get_layer(self, layer_number):
+        return self.layers[layer_number]
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_caches=None,
+            use_cache: Optional[bool] = True,
+            output_hidden_states: Optional[bool] = False,
+    ):
+        if not kv_caches:
+            kv_caches = [None for _ in range(self.num_layers)]
+        presents = () if use_cache else None
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning_once(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        all_self_attentions = None
+        all_hidden_states = () if output_hidden_states else None
+        for index in range(self.num_layers):
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            layer = self._get_layer(index)
+            if self.gradient_checkpointing and self.training:
+                layer_ret = torch.utils.checkpoint.checkpoint(
+                    layer,
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_caches[index],
+                    use_cache
+                )
+            else:
+                layer_ret = layer(
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_cache=kv_caches[index],
+                    use_cache=use_cache
+                )
+            hidden_states, kv_cache = layer_ret
+            if use_cache:
+                presents = presents + (kv_cache,)
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        # Final layer norm.
+        if self.post_layer_norm:
+            hidden_states = self.final_layernorm(hidden_states)
+
+        return hidden_states, presents, all_hidden_states, all_self_attentions
+
+
+class ChatGLMPreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and
+    a simple interface for downloading and loading pretrained models.
+    """
+
+    is_parallelizable = False
+    supports_gradient_checkpointing = True
+    config_class = ChatGLMConfig
+    base_model_prefix = "transformer"
+    _no_split_modules = ["GLMBlock"]
+
+    def _init_weights(self, module: nn.Module):
+        """Initialize the weights."""
+        return
+
+    def get_masks(self, input_ids, past_key_values, padding_mask=None):
+        batch_size, seq_length = input_ids.shape
+        full_attention_mask = torch.ones(batch_size, seq_length, seq_length, device=input_ids.device)
+        full_attention_mask.tril_()
+        past_length = 0
+        if past_key_values:
+            past_length = past_key_values[0][0].shape[0]
+        if past_length:
+            full_attention_mask = torch.cat((torch.ones(batch_size, seq_length, past_length,
+                                                        device=input_ids.device), full_attention_mask), dim=-1)
+        if padding_mask is not None:
+            full_attention_mask = full_attention_mask * padding_mask.unsqueeze(1)
+        if not past_length and padding_mask is not None:
+            full_attention_mask -= padding_mask.unsqueeze(-1) - 1
+        full_attention_mask = (full_attention_mask < 0.5).bool()
+        full_attention_mask.unsqueeze_(1)
+        return full_attention_mask
+
+    def get_position_ids(self, input_ids, device):
+        batch_size, seq_length = input_ids.shape
+        position_ids = torch.arange(seq_length, dtype=torch.long, device=device).unsqueeze(0).repeat(batch_size, 1)
+        return position_ids
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, GLMTransformer):
+            module.gradient_checkpointing = value
+
+
+class Embedding(torch.nn.Module):
+    """Language model embeddings."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(Embedding, self).__init__()
+
+        self.hidden_size = config.hidden_size
+        # Word embeddings (parallel).
+        self.word_embeddings = nn.Embedding(
+            config.padded_vocab_size,
+            self.hidden_size,
+            dtype=config.torch_dtype,
+            device=device
+        )
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+    def forward(self, input_ids):
+        # Embeddings.
+        words_embeddings = self.word_embeddings(input_ids)
+        embeddings = words_embeddings
+        # Data format change to avoid explicit tranposes : [b s h] --> [s b h].
+        embeddings = embeddings.transpose(0, 1).contiguous()
+        # If the input flag for fp32 residual connection is set, convert for float.
+        if self.fp32_residual_connection:
+            embeddings = embeddings.float()
+        return embeddings
+
+
+class ChatGLMModel(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, device=None, empty_init=True):
+        super().__init__(config)
+        if empty_init:
+            init_method = skip_init
+        else:
+            init_method = default_init
+        init_kwargs = {}
+        if device is not None:
+            init_kwargs["device"] = device
+        self.embedding = init_method(Embedding, config, **init_kwargs)
+        self.num_layers = config.num_layers
+        self.multi_query_group_num = config.multi_query_group_num
+        self.kv_channels = config.kv_channels
+
+        # Rotary positional embeddings
+        self.seq_length = config.seq_length
+        rotary_dim = (
+            config.hidden_size // config.num_attention_heads if config.kv_channels is None else config.kv_channels
+        )
+
+        self.rotary_pos_emb = RotaryEmbedding(rotary_dim // 2, original_impl=config.original_rope, device=device,
+                                              dtype=config.torch_dtype)
+        self.encoder = init_method(GLMTransformer, config, **init_kwargs)
+        self.output_layer = init_method(nn.Linear, config.hidden_size, config.padded_vocab_size, bias=False,
+                                        dtype=config.torch_dtype, **init_kwargs)
+        self.pre_seq_len = config.pre_seq_len
+        self.prefix_projection = config.prefix_projection
+        if self.pre_seq_len is not None:
+            for param in self.parameters():
+                param.requires_grad = False
+            self.prefix_tokens = torch.arange(self.pre_seq_len).long()
+            self.prefix_encoder = PrefixEncoder(config)
+            self.dropout = torch.nn.Dropout(0.1)
+
+    def get_input_embeddings(self):
+        return self.embedding.word_embeddings
+
+    def get_prompt(self, batch_size, device, dtype=torch.half):
+        prefix_tokens = self.prefix_tokens.unsqueeze(0).expand(batch_size, -1).to(device)
+        past_key_values = self.prefix_encoder(prefix_tokens).type(dtype)
+        past_key_values = past_key_values.view(
+            batch_size,
+            self.pre_seq_len,
+            self.num_layers * 2,
+            self.multi_query_group_num,
+            self.kv_channels
+        )
+        # seq_len, b, nh, hidden_size
+        past_key_values = self.dropout(past_key_values)
+        past_key_values = past_key_values.permute([2, 1, 0, 3, 4]).split(2)
+        return past_key_values
+
+    def forward(
+            self,
+            input_ids,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.BoolTensor] = None,
+            full_attention_mask: Optional[torch.BoolTensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ):
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        batch_size, seq_length = input_ids.shape
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embedding(input_ids)
+
+        if self.pre_seq_len is not None:
+            if past_key_values is None:
+                past_key_values = self.get_prompt(batch_size=batch_size, device=input_ids.device,
+                                                  dtype=inputs_embeds.dtype)
+            if attention_mask is not None:
+                attention_mask = torch.cat([attention_mask.new_ones((batch_size, self.pre_seq_len)),
+                                            attention_mask], dim=-1)
+
+        if full_attention_mask is None:
+            if (attention_mask is not None and not attention_mask.all()) or (past_key_values and seq_length != 1):
+                full_attention_mask = self.get_masks(input_ids, past_key_values, padding_mask=attention_mask)
+
+        # Rotary positional embeddings
+        rotary_pos_emb = self.rotary_pos_emb(self.seq_length)
+        if position_ids is not None:
+            rotary_pos_emb = rotary_pos_emb[position_ids]
+        else:
+            rotary_pos_emb = rotary_pos_emb[None, :seq_length]
+        rotary_pos_emb = rotary_pos_emb.transpose(0, 1).contiguous()
+
+        # Run encoder.
+        hidden_states, presents, all_hidden_states, all_self_attentions = self.encoder(
+            inputs_embeds, full_attention_mask, rotary_pos_emb=rotary_pos_emb,
+            kv_caches=past_key_values, use_cache=use_cache, output_hidden_states=output_hidden_states
+        )
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, presents, all_hidden_states, all_self_attentions] if v is not None)
+
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=presents,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+        )
+
+    def quantize(self, weight_bit_width: int):
+        from .quantization import quantize
+        quantize(self.encoder, weight_bit_width)
+        return self
+
+
+class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.max_sequence_length = config.max_length
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+        self.config = config
+        self.quantized = False
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def _update_model_kwargs_for_generation(
+            self,
+            outputs: ModelOutput,
+            model_kwargs: Dict[str, Any],
+            is_encoder_decoder: bool = False,
+            standardize_cache_format: bool = False,
+    ) -> Dict[str, Any]:
+        # update past_key_values
+        model_kwargs["past_key_values"] = self._extract_past_from_model_output(
+            outputs, standardize_cache_format=standardize_cache_format
+        )
+
+        # update attention mask
+        if "attention_mask" in model_kwargs:
+            attention_mask = model_kwargs["attention_mask"]
+            model_kwargs["attention_mask"] = torch.cat(
+                [attention_mask, attention_mask.new_ones((attention_mask.shape[0], 1))], dim=-1
+            )
+
+        # update position ids
+        if "position_ids" in model_kwargs:
+            position_ids = model_kwargs["position_ids"]
+            new_position_id = position_ids[..., -1:].clone()
+            new_position_id += 1
+            model_kwargs["position_ids"] = torch.cat(
+                [position_ids, new_position_id], dim=-1
+            )
+
+        model_kwargs["is_first_forward"] = False
+        return model_kwargs
+
+    def prepare_inputs_for_generation(
+            self,
+            input_ids: torch.LongTensor,
+            past_key_values: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            is_first_forward: bool = True,
+            **kwargs
+    ) -> dict:
+        # only last token for input_ids if past is not None
+        if position_ids is None:
+            position_ids = self.get_position_ids(input_ids, device=input_ids.device)
+        if not is_first_forward:
+            if past_key_values is not None:
+                position_ids = position_ids[..., -1:]
+                input_ids = input_ids[:, -1:]
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past_key_values,
+            "position_ids": position_ids,
+            "attention_mask": attention_mask,
+            "return_last_logit": True,
+            "use_cache": use_cache
+        }
+
+    def forward(
+            self,
+            input_ids: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[torch.FloatTensor]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            labels: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_attentions: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+            return_last_logit: Optional[bool] = False,
+    ):
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        if return_last_logit:
+            hidden_states = hidden_states[-1:]
+        lm_logits = self.transformer.output_layer(hidden_states)
+        lm_logits = lm_logits.transpose(0, 1).contiguous()
+
+        loss = None
+        if labels is not None:
+            lm_logits = lm_logits.to(torch.float32)
+
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss(ignore_index=-100)
+            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+
+            lm_logits = lm_logits.to(hidden_states.dtype)
+            loss = loss.to(hidden_states.dtype)
+
+        if not return_dict:
+            output = (lm_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+    @staticmethod
+    def _reorder_cache(
+            past: Tuple[Tuple[torch.Tensor, torch.Tensor], ...], beam_idx: torch.LongTensor
+    ) -> Tuple[Tuple[torch.Tensor, torch.Tensor], ...]:
+        """
+        This function is used to re-order the `past_key_values` cache if [`~PreTrainedModel.beam_search`] or
+        [`~PreTrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
+        beam_idx at every generation step.
+
+        Output shares the same memory storage as `past`.
+        """
+        return tuple(
+            (
+                layer_past[0].index_select(1, beam_idx.to(layer_past[0].device)),
+                layer_past[1].index_select(1, beam_idx.to(layer_past[1].device)),
+            )
+            for layer_past in past
+        )
+
+    def process_response(self, response):
+        response = response.strip()
+        response = response.replace("[[训练时间]]", "2023年")
+        return response
+
+    def build_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        prompt = tokenizer.build_prompt(query, history=history)
+        inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    def build_stream_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        if history:
+            prompt = "\n\n[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            input_ids = tokenizer.encode(prompt, add_special_tokens=False)
+            input_ids = input_ids[1:]
+            inputs = tokenizer.batch_encode_plus([(input_ids, None)], return_tensors="pt", add_special_tokens=False)
+        else:
+            prompt = "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    @torch.inference_mode()
+    def chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, max_length: int = 8192, num_beams=1,
+             do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "num_beams": num_beams, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        inputs = self.build_inputs(tokenizer, query, history=history)
+        outputs = self.generate(**inputs, **gen_kwargs)
+        outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+        response = tokenizer.decode(outputs)
+        response = self.process_response(response)
+        history = history + [(query, response)]
+        return response, history
+
+    @torch.inference_mode()
+    def stream_chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, past_key_values=None,
+                    max_length: int = 8192, do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None,
+                    return_past_key_values=False, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        if past_key_values is None and not return_past_key_values:
+            inputs = self.build_inputs(tokenizer, query, history=history)
+        else:
+            inputs = self.build_stream_inputs(tokenizer, query, history=history)
+        if past_key_values is not None:
+            past_length = past_key_values[0][0].shape[0]
+            if self.transformer.pre_seq_len is not None:
+                past_length -= self.transformer.pre_seq_len
+            inputs.position_ids += past_length
+            attention_mask = inputs.attention_mask
+            attention_mask = torch.cat((attention_mask.new_ones(1, past_length), attention_mask), dim=1)
+            inputs['attention_mask'] = attention_mask
+        for outputs in self.stream_generate(**inputs, past_key_values=past_key_values,
+                                            return_past_key_values=return_past_key_values, **gen_kwargs):
+            if return_past_key_values:
+                outputs, past_key_values = outputs
+            outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+            response = tokenizer.decode(outputs)
+            if response and response[-1] != "�":
+                response = self.process_response(response)
+                new_history = history + [(query, response)]
+                if return_past_key_values:
+                    yield response, new_history, past_key_values
+                else:
+                    yield response, new_history
+
+    @torch.inference_mode()
+    def stream_generate(
+            self,
+            input_ids,
+            generation_config: Optional[GenerationConfig] = None,
+            logits_processor: Optional[LogitsProcessorList] = None,
+            stopping_criteria: Optional[StoppingCriteriaList] = None,
+            prefix_allowed_tokens_fn: Optional[Callable[[int, torch.Tensor], List[int]]] = None,
+            return_past_key_values=False,
+            **kwargs,
+    ):
+        batch_size, input_ids_seq_length = input_ids.shape[0], input_ids.shape[-1]
+
+        if generation_config is None:
+            generation_config = self.generation_config
+        generation_config = copy.deepcopy(generation_config)
+        model_kwargs = generation_config.update(**kwargs)
+        model_kwargs["use_cache"] = generation_config.use_cache
+        bos_token_id, eos_token_id = generation_config.bos_token_id, generation_config.eos_token_id
+
+        if isinstance(eos_token_id, int):
+            eos_token_id = [eos_token_id]
+
+        has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
+        if has_default_max_length and generation_config.max_new_tokens is None:
+            warnings.warn(
+                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
+                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
+                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                UserWarning,
+            )
+        elif generation_config.max_new_tokens is not None:
+            generation_config.max_length = generation_config.max_new_tokens + input_ids_seq_length
+            if not has_default_max_length:
+                logger.warn(
+                    f"Both `max_new_tokens` (={generation_config.max_new_tokens}) and `max_length`(="
+                    f"{generation_config.max_length}) seem to have been set. `max_new_tokens` will take precedence. "
+                    "Please refer to the documentation for more information. "
+                    "(https://huggingface.co/docs/transformers/main/en/main_classes/text_generation)",
+                    UserWarning,
+                )
+
+        if input_ids_seq_length >= generation_config.max_length:
+            input_ids_string = "decoder_input_ids" if self.config.is_encoder_decoder else "input_ids"
+            logger.warning(
+                f"Input length of {input_ids_string} is {input_ids_seq_length}, but `max_length` is set to"
+                f" {generation_config.max_length}. This can lead to unexpected behavior. You should consider"
+                " increasing `max_new_tokens`."
+            )
+
+        # 2. Set generation parameters if not already defined
+        logits_processor = logits_processor if logits_processor is not None else LogitsProcessorList()
+        stopping_criteria = stopping_criteria if stopping_criteria is not None else StoppingCriteriaList()
+
+        logits_processor = self._get_logits_processor(
+            generation_config=generation_config,
+            input_ids_seq_length=input_ids_seq_length,
+            encoder_input_ids=input_ids,
+            prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
+            logits_processor=logits_processor,
+        )
+
+        stopping_criteria = self._get_stopping_criteria(
+            generation_config=generation_config, stopping_criteria=stopping_criteria
+        )
+        logits_warper = self._get_logits_warper(generation_config)
+
+        unfinished_sequences = input_ids.new(input_ids.shape[0]).fill_(1)
+        scores = None
+        while True:
+            model_inputs = self.prepare_inputs_for_generation(input_ids, **model_kwargs)
+            # forward pass to get next token
+            outputs = self(
+                **model_inputs,
+                return_dict=True,
+                output_attentions=False,
+                output_hidden_states=False,
+            )
+
+            next_token_logits = outputs.logits[:, -1, :]
+
+            # pre-process distribution
+            next_token_scores = logits_processor(input_ids, next_token_logits)
+            next_token_scores = logits_warper(input_ids, next_token_scores)
+
+            # sample
+            probs = nn.functional.softmax(next_token_scores, dim=-1)
+            if generation_config.do_sample:
+                next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
+            else:
+                next_tokens = torch.argmax(probs, dim=-1)
+
+            # update generated ids, model inputs, and length for next step
+            input_ids = torch.cat([input_ids, next_tokens[:, None]], dim=-1)
+            model_kwargs = self._update_model_kwargs_for_generation(
+                outputs, model_kwargs, is_encoder_decoder=self.config.is_encoder_decoder
+            )
+            unfinished_sequences = unfinished_sequences.mul((sum(next_tokens != i for i in eos_token_id)).long())
+            if return_past_key_values:
+                yield input_ids, outputs.past_key_values
+            else:
+                yield input_ids
+            # stop when each sentence is finished, or if we exceed the maximum length
+            if unfinished_sequences.max() == 0 or stopping_criteria(input_ids, scores):
+                break
+
+    def quantize(self, bits: int, empty_init=False, device=None, **kwargs):
+        if bits == 0:
+            return
+
+        from .quantization import quantize
+
+        if self.quantized:
+            logger.info("Already quantized.")
+            return self
+
+        self.quantized = True
+
+        self.config.quantization_bit = bits
+
+        self.transformer.encoder = quantize(self.transformer.encoder, bits, empty_init=empty_init, device=device,
+                                            **kwargs)
+        return self
+
+
+class ChatGLMForSequenceClassification(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.num_labels = config.num_labels
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+
+        self.classifier_head = nn.Linear(config.hidden_size, config.num_labels, bias=True, dtype=torch.half)
+        if config.classifier_dropout is not None:
+            self.dropout = nn.Dropout(config.classifier_dropout)
+        else:
+            self.dropout = None
+        self.config = config
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def forward(
+            self,
+            input_ids: Optional[torch.LongTensor] = None,
+            position_ids: Optional[torch.LongTensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            full_attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.LongTensor] = None,
+            labels: Optional[torch.LongTensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor, ...], SequenceClassifierOutputWithPast]:
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            full_attention_mask=full_attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        pooled_hidden_states = hidden_states[-1]
+        if self.dropout is not None:
+            pooled_hidden_states = self.dropout(pooled_hidden_states)
+        logits = self.classifier_head(pooled_hidden_states)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze().float(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits.float(), labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(logits.view(-1, self.num_labels).float(), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits.float(), labels.view(-1, self.num_labels))
+
+        if not return_dict:
+            output = (logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )

checkpoint-2000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7141e4be1f44b2119e3e32755baacf7a756a25ff21af1f5fb79facaecbbb97a0
+size 14681455

checkpoint-2000/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2d64d7dc915cba98541c03f637ad8167c792d0b5b9611aff477356d7beeed9f
+size 7340861

checkpoint-2000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e769ab142cc628982263acecebfd601643b5ad22d48163da36d4246c8d70d5b9
+size 14575

checkpoint-2000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2181151705bc21abd4b28f2057fc0eb4770b8b2410347d913d2ae5fb451e29d
+size 627

checkpoint-2000/special_tokens_map.json

@@ -0,0 +1 @@
+{}

checkpoint-2000/tokenization_chatglm.py

@@ -0,0 +1,257 @@
+import os
+import torch
+from typing import List, Optional, Union, Dict
+from sentencepiece import SentencePieceProcessor
+from transformers import PreTrainedTokenizer
+from transformers.utils import logging, PaddingStrategy
+from transformers.tokenization_utils_base import EncodedInput, BatchEncoding
+
+
+class SPTokenizer:
+    def __init__(self, model_path: str):
+        # reload tokenizer
+        assert os.path.isfile(model_path), model_path
+        self.sp_model = SentencePieceProcessor(model_file=model_path)
+
+        # BOS / EOS token IDs
+        self.n_words: int = self.sp_model.vocab_size()
+        self.bos_id: int = self.sp_model.bos_id()
+        self.eos_id: int = self.sp_model.eos_id()
+        self.pad_id: int = self.sp_model.unk_id()
+        assert self.sp_model.vocab_size() == self.sp_model.get_piece_size()
+
+        special_tokens = ["[MASK]", "[gMASK]", "[sMASK]", "sop", "eop"]
+        self.special_tokens = {}
+        self.index_special_tokens = {}
+        for token in special_tokens:
+            self.special_tokens[token] = self.n_words
+            self.index_special_tokens[self.n_words] = token
+            self.n_words += 1
+
+    def tokenize(self, s: str):
+        return self.sp_model.EncodeAsPieces(s)
+
+    def encode(self, s: str, bos: bool = False, eos: bool = False) -> List[int]:
+        assert type(s) is str
+        t = self.sp_model.encode(s)
+        if bos:
+            t = [self.bos_id] + t
+        if eos:
+            t = t + [self.eos_id]
+        return t
+
+    def decode(self, t: List[int]) -> str:
+        return self.sp_model.decode(t)
+
+    def decode_tokens(self, tokens: List[str]) -> str:
+        text = self.sp_model.DecodePieces(tokens)
+        return text
+
+    def convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        return self.sp_model.PieceToId(token)
+
+    def convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        if index in self.index_special_tokens or index in [self.eos_id, self.bos_id, self.pad_id] or index < 0:
+            return ""
+        return self.sp_model.IdToPiece(index)
+
+
+class ChatGLMTokenizer(PreTrainedTokenizer):
+    vocab_files_names = {"vocab_file": "tokenizer.model"}
+
+    model_input_names = ["input_ids", "attention_mask", "position_ids"]
+
+    def __init__(self, vocab_file, padding_side="left", clean_up_tokenization_spaces=False, **kwargs):
+        self.name = "GLMTokenizer"
+
+        self.vocab_file = vocab_file
+        self.tokenizer = SPTokenizer(vocab_file)
+        self.special_tokens = {
+            "<bos>": self.tokenizer.bos_id,
+            "<eos>": self.tokenizer.eos_id,
+            "<pad>": self.tokenizer.pad_id
+        }
+        super().__init__(padding_side=padding_side, clean_up_tokenization_spaces=clean_up_tokenization_spaces, **kwargs)
+
+    def get_command(self, token):
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        assert token in self.tokenizer.special_tokens, f"{token} is not a special token for {self.name}"
+        return self.tokenizer.special_tokens[token]
+
+    @property
+    def unk_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token_id(self):
+        return self.get_command("<pad>")
+
+    @property
+    def eos_token(self) -> str:
+        return "</s>"
+
+    @property
+    def eos_token_id(self):
+        return self.get_command("<eos>")
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_words
+
+    def get_vocab(self):
+        """ Returns vocab as a dict """
+        vocab = {self._convert_id_to_token(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text, **kwargs):
+        return self.tokenizer.tokenize(text)
+
+    def _convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        return self.tokenizer.convert_token_to_id(token)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        return self.tokenizer.convert_id_to_token(index)
+
+    def convert_tokens_to_string(self, tokens: List[str]) -> str:
+        return self.tokenizer.decode_tokens(tokens)
+
+    def save_vocabulary(self, save_directory, filename_prefix=None):
+        """
+        Save the vocabulary and special tokens file to a directory.
+
+        Args:
+            save_directory (`str`):
+                The directory in which to save the vocabulary.
+            filename_prefix (`str`, *optional*):
+                An optional prefix to add to the named of the saved files.
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        if os.path.isdir(save_directory):
+            vocab_file = os.path.join(
+                save_directory, self.vocab_files_names["vocab_file"]
+            )
+        else:
+            vocab_file = save_directory
+
+        with open(self.vocab_file, 'rb') as fin:
+            proto_str = fin.read()
+
+        with open(vocab_file, "wb") as writer:
+            writer.write(proto_str)
+
+        return (vocab_file,)
+
+    def get_prefix_tokens(self):
+        prefix_tokens = [self.get_command("[gMASK]"), self.get_command("sop")]
+        return prefix_tokens
+
+    def build_prompt(self, query, history=None):
+        if history is None:
+            history = []
+        prompt = ""
+        for i, (old_query, response) in enumerate(history):
+            prompt += "[Round {}]\n\n问：{}\n\n答：{}\n\n".format(i + 1, old_query, response)
+        prompt += "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+        return prompt
+
+    def build_inputs_with_special_tokens(
+            self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. A BERT sequence has the following format:
+
+        - single sequence: `[CLS] X [SEP]`
+        - pair of sequences: `[CLS] A [SEP] B [SEP]`
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        prefix_tokens = self.get_prefix_tokens()
+        token_ids_0 = prefix_tokens + token_ids_0
+        if token_ids_1 is not None:
+            token_ids_0 = token_ids_0 + token_ids_1 + [self.get_command("<eos>")]
+        return token_ids_0
+
+    def _pad(
+            self,
+            encoded_inputs: Union[Dict[str, EncodedInput], BatchEncoding],
+            max_length: Optional[int] = None,
+            padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
+            pad_to_multiple_of: Optional[int] = None,
+            return_attention_mask: Optional[bool] = None,
+    ) -> dict:
+        """
+        Pad encoded inputs (on left/right and up to predefined length or max length in the batch)
+
+        Args:
+            encoded_inputs:
+                Dictionary of tokenized inputs (`List[int]`) or batch of tokenized inputs (`List[List[int]]`).
+            max_length: maximum length of the returned list and optionally padding length (see below).
+                Will truncate by taking into account the special tokens.
+            padding_strategy: PaddingStrategy to use for padding.
+
+                - PaddingStrategy.LONGEST Pad to the longest sequence in the batch
+                - PaddingStrategy.MAX_LENGTH: Pad to the max length (default)
+                - PaddingStrategy.DO_NOT_PAD: Do not pad
+                The tokenizer padding sides are defined in self.padding_side:
+
+                    - 'left': pads on the left of the sequences
+                    - 'right': pads on the right of the sequences
+            pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
+                This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
+                `>= 7.5` (Volta).
+            return_attention_mask:
+                (optional) Set to False to avoid returning attention mask (default: set to model specifics)
+        """
+        # Load from model defaults
+        assert self.padding_side == "left"
+
+        required_input = encoded_inputs[self.model_input_names[0]]
+        seq_length = len(required_input)
+
+        if padding_strategy == PaddingStrategy.LONGEST:
+            max_length = len(required_input)
+
+        if max_length is not None and pad_to_multiple_of is not None and (max_length % pad_to_multiple_of != 0):
+            max_length = ((max_length // pad_to_multiple_of) + 1) * pad_to_multiple_of
+
+        needs_to_be_padded = padding_strategy != PaddingStrategy.DO_NOT_PAD and len(required_input) != max_length
+
+        # Initialize attention mask if not present.
+        if "attention_mask" not in encoded_inputs:
+            encoded_inputs["attention_mask"] = [1] * seq_length
+
+        if "position_ids" not in encoded_inputs:
+            encoded_inputs["position_ids"] = list(range(seq_length))
+
+        if needs_to_be_padded:
+            difference = max_length - len(required_input)
+
+            if "attention_mask" in encoded_inputs:
+                encoded_inputs["attention_mask"] = [0] * difference + encoded_inputs["attention_mask"]
+            if "position_ids" in encoded_inputs:
+                encoded_inputs["position_ids"] = [0] * difference + encoded_inputs["position_ids"]
+            encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input
+
+        return encoded_inputs

checkpoint-2000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7dc4c393423b76e4373e5157ddc34803a0189ba96b21ddbb40269d31468a6f2
+size 1018370

checkpoint-2000/tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_chatglm.ChatGLMTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "padding_side": "left",
+  "remove_space": false,
+  "tokenizer_class": "ChatGLMTokenizer"
+}

checkpoint-2000/trainer_state.json

@@ -0,0 +1,1216 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18471911150107367,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019933333333333334,
+      "loss": 2.6644,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019866666666666668,
+      "loss": 1.7151,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0198,
+      "loss": 1.6228,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019733333333333335,
+      "loss": 1.401,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019666666666666666,
+      "loss": 1.6172,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0196,
+      "loss": 1.4695,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019533333333333333,
+      "loss": 1.5137,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019466666666666667,
+      "loss": 1.5425,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0194,
+      "loss": 1.4272,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019333333333333334,
+      "loss": 1.3727,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019266666666666668,
+      "loss": 1.3114,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0192,
+      "loss": 1.4758,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019133333333333332,
+      "loss": 1.5219,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.01906666666666667,
+      "loss": 1.376,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019,
+      "loss": 1.4257,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.018933333333333333,
+      "loss": 1.3474,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018866666666666667,
+      "loss": 1.2929,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0188,
+      "loss": 1.3208,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018733333333333334,
+      "loss": 1.3381,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018666666666666668,
+      "loss": 1.3644,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018600000000000002,
+      "loss": 1.2932,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018533333333333332,
+      "loss": 1.4092,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018466666666666666,
+      "loss": 1.3006,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0184,
+      "loss": 1.4572,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018333333333333333,
+      "loss": 1.2789,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018266666666666667,
+      "loss": 1.4444,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0182,
+      "loss": 1.4511,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01813333333333333,
+      "loss": 1.3541,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01806666666666667,
+      "loss": 1.3228,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.018000000000000002,
+      "loss": 1.3185,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017933333333333332,
+      "loss": 1.199,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017866666666666666,
+      "loss": 1.3417,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0178,
+      "loss": 1.4251,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017733333333333334,
+      "loss": 1.3574,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017666666666666667,
+      "loss": 1.2547,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0176,
+      "loss": 1.2651,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017533333333333335,
+      "loss": 1.3414,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017466666666666665,
+      "loss": 1.3322,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0174,
+      "loss": 1.4147,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017333333333333336,
+      "loss": 1.2813,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017266666666666666,
+      "loss": 1.3687,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0172,
+      "loss": 1.5593,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017133333333333334,
+      "loss": 1.3073,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017066666666666667,
+      "loss": 1.2359,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017,
+      "loss": 1.2474,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.016933333333333335,
+      "loss": 1.3874,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.01686666666666667,
+      "loss": 1.3203,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0168,
+      "loss": 1.2875,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016733333333333333,
+      "loss": 1.2767,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016666666666666666,
+      "loss": 1.3017,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0166,
+      "loss": 1.2321,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016533333333333334,
+      "loss": 1.1719,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016466666666666668,
+      "loss": 1.2552,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016399999999999998,
+      "loss": 1.3816,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01633333333333333,
+      "loss": 1.2956,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01626666666666667,
+      "loss": 1.2061,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016200000000000003,
+      "loss": 1.2086,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016133333333333333,
+      "loss": 1.1633,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016066666666666667,
+      "loss": 1.2638,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.016,
+      "loss": 1.3441,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015933333333333334,
+      "loss": 1.2924,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015866666666666668,
+      "loss": 1.1818,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0158,
+      "loss": 1.3918,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015733333333333332,
+      "loss": 1.2232,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015666666666666666,
+      "loss": 1.2472,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015600000000000001,
+      "loss": 1.2398,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015533333333333333,
+      "loss": 1.3649,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015466666666666667,
+      "loss": 1.2302,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0154,
+      "loss": 1.2053,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015333333333333334,
+      "loss": 1.2974,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015266666666666666,
+      "loss": 1.3036,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0152,
+      "loss": 1.3162,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015133333333333334,
+      "loss": 1.2567,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015066666666666666,
+      "loss": 1.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015,
+      "loss": 1.2692,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014933333333333335,
+      "loss": 1.1332,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014866666666666667,
+      "loss": 1.2949,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0148,
+      "loss": 1.2703,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014733333333333334,
+      "loss": 1.3891,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014666666666666666,
+      "loss": 1.3594,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0146,
+      "loss": 1.166,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014533333333333334,
+      "loss": 1.3256,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014466666666666668,
+      "loss": 1.2669,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0144,
+      "loss": 1.241,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014333333333333333,
+      "loss": 1.2591,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014266666666666667,
+      "loss": 1.238,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014199999999999999,
+      "loss": 1.3583,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014133333333333333,
+      "loss": 1.164,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014066666666666668,
+      "loss": 1.2367,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013999999999999999,
+      "loss": 1.1864,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013933333333333334,
+      "loss": 1.2259,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013866666666666668,
+      "loss": 1.2129,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0138,
+      "loss": 1.2085,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013733333333333334,
+      "loss": 1.2316,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013666666666666667,
+      "loss": 1.2721,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013600000000000001,
+      "loss": 1.2428,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013533333333333333,
+      "loss": 1.2126,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013466666666666667,
+      "loss": 1.1583,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0134,
+      "loss": 1.2776,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013333333333333332,
+      "loss": 1.2317,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013266666666666666,
+      "loss": 1.2047,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013200000000000002,
+      "loss": 1.202,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013133333333333332,
+      "loss": 1.1877,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013066666666666667,
+      "loss": 1.3071,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013000000000000001,
+      "loss": 1.2976,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012933333333333333,
+      "loss": 1.4156,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012866666666666667,
+      "loss": 1.173,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0128,
+      "loss": 1.2263,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012733333333333334,
+      "loss": 1.3235,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012666666666666666,
+      "loss": 1.1923,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0126,
+      "loss": 1.2879,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012533333333333334,
+      "loss": 1.196,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012466666666666666,
+      "loss": 1.2605,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0124,
+      "loss": 1.2303,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012333333333333335,
+      "loss": 1.2381,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012266666666666665,
+      "loss": 1.2709,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0122,
+      "loss": 1.2121,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012133333333333335,
+      "loss": 1.3713,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012066666666666668,
+      "loss": 1.2923,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012,
+      "loss": 1.2947,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011933333333333334,
+      "loss": 1.1538,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011866666666666668,
+      "loss": 1.1312,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0118,
+      "loss": 1.1807,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011733333333333333,
+      "loss": 1.2729,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011666666666666667,
+      "loss": 1.21,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0116,
+      "loss": 1.1986,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011533333333333333,
+      "loss": 1.2003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011466666666666667,
+      "loss": 1.1773,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011399999999999999,
+      "loss": 1.3241,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011333333333333332,
+      "loss": 1.2157,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011266666666666668,
+      "loss": 1.2549,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011200000000000002,
+      "loss": 1.3245,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011133333333333334,
+      "loss": 1.2109,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011066666666666667,
+      "loss": 1.1979,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011000000000000001,
+      "loss": 1.2804,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010933333333333333,
+      "loss": 1.2655,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010866666666666667,
+      "loss": 1.1264,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0108,
+      "loss": 1.2949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010733333333333333,
+      "loss": 1.2038,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010666666666666666,
+      "loss": 1.2514,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0106,
+      "loss": 1.1692,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010533333333333332,
+      "loss": 1.1947,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010466666666666666,
+      "loss": 1.3294,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010400000000000001,
+      "loss": 1.2169,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010333333333333335,
+      "loss": 1.3113,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010266666666666667,
+      "loss": 1.1322,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0102,
+      "loss": 1.4228,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010133333333333334,
+      "loss": 1.2384,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010066666666666666,
+      "loss": 1.2107,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.01,
+      "loss": 1.2655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009933333333333334,
+      "loss": 1.2991,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009866666666666668,
+      "loss": 1.324,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0098,
+      "loss": 1.3443,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009733333333333333,
+      "loss": 1.1389,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009666666666666667,
+      "loss": 1.2308,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0096,
+      "loss": 1.1847,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009533333333333335,
+      "loss": 1.3154,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009466666666666667,
+      "loss": 1.233,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0094,
+      "loss": 1.147,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009333333333333334,
+      "loss": 1.1824,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009266666666666666,
+      "loss": 1.2093,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0092,
+      "loss": 1.2111,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009133333333333334,
+      "loss": 1.0596,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009066666666666666,
+      "loss": 1.3025,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009000000000000001,
+      "loss": 1.1726,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008933333333333333,
+      "loss": 1.2078,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008866666666666667,
+      "loss": 1.2652,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0088,
+      "loss": 1.2033,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008733333333333333,
+      "loss": 1.1598,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008666666666666668,
+      "loss": 1.1904,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0086,
+      "loss": 1.242,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008533333333333334,
+      "loss": 1.3042,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008466666666666667,
+      "loss": 1.3653,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0084,
+      "loss": 1.1784,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008333333333333333,
+      "loss": 1.2306,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008266666666666667,
+      "loss": 1.2139,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008199999999999999,
+      "loss": 1.1891,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008133333333333334,
+      "loss": 1.2619,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008066666666666666,
+      "loss": 1.0873,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008,
+      "loss": 1.2537,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007933333333333334,
+      "loss": 1.2575,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007866666666666666,
+      "loss": 1.1043,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0078000000000000005,
+      "loss": 1.2063,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007733333333333333,
+      "loss": 1.1602,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007666666666666667,
+      "loss": 1.1474,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0076,
+      "loss": 1.1482,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007533333333333333,
+      "loss": 1.2124,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074666666666666675,
+      "loss": 1.195,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074,
+      "loss": 1.1426,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007333333333333333,
+      "loss": 1.2067,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007266666666666667,
+      "loss": 1.1649,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0072,
+      "loss": 1.0978,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0071333333333333335,
+      "loss": 1.2298,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007066666666666666,
+      "loss": 1.195,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006999999999999999,
+      "loss": 1.2032,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006933333333333334,
+      "loss": 1.1134,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006866666666666667,
+      "loss": 1.2925,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0068000000000000005,
+      "loss": 1.1389,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006733333333333333,
+      "loss": 1.1952,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006666666666666666,
+      "loss": 1.0672,
+      "step": 2000
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 1,
+  "total_flos": 1.15729542610944e+17,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70ac9eb43a2d446e07cde6bbbb21250fe0373a4093de76a8aa1f7223e3836bcd
+size 4155

checkpoint-3000/config.json

@@ -0,0 +1,44 @@
+{
+  "_name_or_path": "E:/PycharmProjects/dl_models/chatglm2-6b-int4",
+  "add_bias_linear": false,
+  "add_qkv_bias": true,
+  "apply_query_key_layer_scaling": true,
+  "apply_residual_connection_post_layernorm": false,
+  "architectures": [
+    "ChatGLMForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "attention_softmax_in_fp32": true,
+  "auto_map": {
+    "AutoConfig": "configuration_chatglm.ChatGLMConfig",
+    "AutoModel": "modeling_chatglm.ChatGLMForConditionalGeneration",
+    "AutoModelForSeq2SeqLM": "modeling_chatglm.ChatGLMForConditionalGeneration"
+  },
+  "bias_dropout_fusion": true,
+  "eos_token_id": 2,
+  "ffn_hidden_size": 13696,
+  "fp32_residual_connection": false,
+  "hidden_dropout": 0.0,
+  "hidden_size": 4096,
+  "kv_channels": 128,
+  "layernorm_epsilon": 1e-05,
+  "model_type": "chatglm",
+  "multi_query_attention": true,
+  "multi_query_group_num": 2,
+  "num_attention_heads": 32,
+  "num_layers": 28,
+  "original_rope": true,
+  "pad_token_id": 0,
+  "padded_vocab_size": 65024,
+  "post_layer_norm": true,
+  "pre_seq_len": 128,
+  "prefix_projection": false,
+  "quantization_bit": 4,
+  "rmsnorm": true,
+  "seq_length": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.31.0",
+  "use_cache": true,
+  "vocab_size": 65024
+}

checkpoint-3000/configuration_chatglm.py

@@ -0,0 +1,59 @@
+from transformers import PretrainedConfig
+
+
+class ChatGLMConfig(PretrainedConfig):
+    model_type = "chatglm"
+    def __init__(
+        self,
+        num_layers=28,
+        padded_vocab_size=65024,
+        hidden_size=4096,
+        ffn_hidden_size=13696,
+        kv_channels=128,
+        num_attention_heads=32,
+        seq_length=2048,
+        hidden_dropout=0.0,
+        attention_dropout=0.0,
+        layernorm_epsilon=1e-5,
+        rmsnorm=True,
+        apply_residual_connection_post_layernorm=False,
+        post_layer_norm=True,
+        add_bias_linear=False,
+        add_qkv_bias=False,
+        bias_dropout_fusion=True,
+        multi_query_attention=False,
+        multi_query_group_num=1,
+        apply_query_key_layer_scaling=True,
+        attention_softmax_in_fp32=True,
+        fp32_residual_connection=False,
+        quantization_bit=0,
+        pre_seq_len=None,
+        prefix_projection=False,
+        **kwargs
+    ):
+        self.num_layers = num_layers
+        self.vocab_size = padded_vocab_size
+        self.padded_vocab_size = padded_vocab_size
+        self.hidden_size = hidden_size
+        self.ffn_hidden_size = ffn_hidden_size
+        self.kv_channels = kv_channels
+        self.num_attention_heads = num_attention_heads
+        self.seq_length = seq_length
+        self.hidden_dropout = hidden_dropout
+        self.attention_dropout = attention_dropout
+        self.layernorm_epsilon = layernorm_epsilon
+        self.rmsnorm = rmsnorm
+        self.apply_residual_connection_post_layernorm = apply_residual_connection_post_layernorm
+        self.post_layer_norm = post_layer_norm
+        self.add_bias_linear = add_bias_linear
+        self.add_qkv_bias = add_qkv_bias
+        self.bias_dropout_fusion = bias_dropout_fusion
+        self.multi_query_attention = multi_query_attention
+        self.multi_query_group_num = multi_query_group_num
+        self.apply_query_key_layer_scaling = apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = attention_softmax_in_fp32
+        self.fp32_residual_connection = fp32_residual_connection
+        self.quantization_bit = quantization_bit
+        self.pre_seq_len = pre_seq_len
+        self.prefix_projection = prefix_projection
+        super().__init__(**kwargs)

checkpoint-3000/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.31.0"
+}

checkpoint-3000/modeling_chatglm.py

@@ -0,0 +1,1285 @@
+""" PyTorch ChatGLM model. """
+
+import math
+import copy
+import warnings
+import re
+import sys
+
+import torch
+import torch.utils.checkpoint
+import torch.nn.functional as F
+from torch import nn
+from torch.nn import CrossEntropyLoss, LayerNorm
+from torch.nn import CrossEntropyLoss, LayerNorm, MSELoss, BCEWithLogitsLoss
+from torch.nn.utils import skip_init
+from typing import Optional, Tuple, Union, List, Callable, Dict, Any
+
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+from transformers.generation.logits_process import LogitsProcessor
+from transformers.generation.utils import LogitsProcessorList, StoppingCriteriaList, GenerationConfig, ModelOutput
+
+from .configuration_chatglm import ChatGLMConfig
+
+# flags required to enable jit fusion kernels
+
+if sys.platform != 'darwin':
+    torch._C._jit_set_profiling_mode(False)
+    torch._C._jit_set_profiling_executor(False)
+    torch._C._jit_override_can_fuse_on_cpu(True)
+    torch._C._jit_override_can_fuse_on_gpu(True)
+
+logger = logging.get_logger(__name__)
+
+_CHECKPOINT_FOR_DOC = "THUDM/ChatGLM2-6B"
+_CONFIG_FOR_DOC = "ChatGLM6BConfig"
+
+CHATGLM_6B_PRETRAINED_MODEL_ARCHIVE_LIST = [
+    "THUDM/chatglm2-6b",
+    # See all ChatGLM models at https://huggingface.co/models?filter=chatglm
+]
+
+
+def default_init(cls, *args, **kwargs):
+    return cls(*args, **kwargs)
+
+
+class InvalidScoreLogitsProcessor(LogitsProcessor):
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
+        if torch.isnan(scores).any() or torch.isinf(scores).any():
+            scores.zero_()
+            scores[..., 5] = 5e4
+        return scores
+
+
+class PrefixEncoder(torch.nn.Module):
+    """
+    The torch.nn model to encode the prefix
+    Input shape: (batch-size, prefix-length)
+    Output shape: (batch-size, prefix-length, 2*layers*hidden)
+    """
+
+    def __init__(self, config: ChatGLMConfig):
+        super().__init__()
+        self.prefix_projection = config.prefix_projection
+        if self.prefix_projection:
+            # Use a two-layer MLP to encode the prefix
+            kv_size = config.num_layers * config.kv_channels * config.multi_query_group_num * 2
+            self.embedding = torch.nn.Embedding(config.pre_seq_len, kv_size)
+            self.trans = torch.nn.Sequential(
+                torch.nn.Linear(kv_size, config.hidden_size),
+                torch.nn.Tanh(),
+                torch.nn.Linear(config.hidden_size, kv_size)
+            )
+        else:
+            self.embedding = torch.nn.Embedding(config.pre_seq_len,
+                                                config.num_layers * config.kv_channels * config.multi_query_group_num * 2)
+
+    def forward(self, prefix: torch.Tensor):
+        if self.prefix_projection:
+            prefix_tokens = self.embedding(prefix)
+            past_key_values = self.trans(prefix_tokens)
+        else:
+            past_key_values = self.embedding(prefix)
+        return past_key_values
+
+
+def split_tensor_along_last_dim(
+        tensor: torch.Tensor,
+        num_partitions: int,
+        contiguous_split_chunks: bool = False,
+) -> List[torch.Tensor]:
+    """Split a tensor along its last dimension.
+
+    Arguments:
+        tensor: input tensor.
+        num_partitions: number of partitions to split the tensor
+        contiguous_split_chunks: If True, make each chunk contiguous
+                                 in memory.
+
+    Returns:
+        A list of Tensors
+    """
+    # Get the size and dimension.
+    last_dim = tensor.dim() - 1
+    last_dim_size = tensor.size()[last_dim] // num_partitions
+    # Split.
+    tensor_list = torch.split(tensor, last_dim_size, dim=last_dim)
+    # Note: torch.split does not create contiguous tensors by default.
+    if contiguous_split_chunks:
+        return tuple(chunk.contiguous() for chunk in tensor_list)
+
+    return tensor_list
+
+
+class RotaryEmbedding(nn.Module):
+    def __init__(self, dim, original_impl=False, device=None, dtype=None):
+        super().__init__()
+        inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2, device=device).to(dtype=dtype) / dim))
+        self.register_buffer("inv_freq", inv_freq)
+        self.dim = dim
+        self.original_impl = original_impl
+
+    def forward_impl(
+            self, seq_len: int, n_elem: int, dtype: torch.dtype, device: torch.device, base: int = 10000
+    ):
+        """Enhanced Transformer with Rotary Position Embedding.
+
+        Derived from: https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/labml_nn/
+        transformers/rope/__init__.py. MIT License:
+        https://github.com/labmlai/annotated_deep_learning_paper_implementations/blob/master/license.
+        """
+        # $\Theta = {\theta_i = 10000^{\frac{2(i-1)}{d}}, i \in [1, 2, ..., \frac{d}{2}]}$
+        theta = 1.0 / (base ** (torch.arange(0, n_elem, 2, dtype=dtype, device=device) / n_elem))
+
+        # Create position indexes `[0, 1, ..., seq_len - 1]`
+        seq_idx = torch.arange(seq_len, dtype=dtype, device=device)
+
+        # Calculate the product of position index and $\theta_i$
+        idx_theta = torch.outer(seq_idx, theta).float()
+
+        cache = torch.stack([torch.cos(idx_theta), torch.sin(idx_theta)], dim=-1)
+
+        # this is to mimic the behaviour of complex32, else we will get different results
+        if dtype in (torch.float16, torch.bfloat16, torch.int8):
+            cache = cache.bfloat16() if dtype == torch.bfloat16 else cache.half()
+        return cache
+
+    def forward(self, max_seq_len, offset=0):
+        return self.forward_impl(
+            max_seq_len, self.dim, dtype=self.inv_freq.dtype, device=self.inv_freq.device
+        )
+
+
+@torch.jit.script
+def apply_rotary_pos_emb(x: torch.Tensor, rope_cache: torch.Tensor) -> torch.Tensor:
+    # x: [sq, b, np, hn]
+    sq, b, np, hn = x.size(0), x.size(1), x.size(2), x.size(3)
+    rot_dim = rope_cache.shape[-2] * 2
+    x, x_pass = x[..., :rot_dim], x[..., rot_dim:]
+    # truncate to support variable sizes
+    rope_cache = rope_cache[:sq]
+    xshaped = x.reshape(sq, -1, np, rot_dim // 2, 2)
+    rope_cache = rope_cache.view(sq, -1, 1, xshaped.size(3), 2)
+    x_out2 = torch.stack(
+        [
+            xshaped[..., 0] * rope_cache[..., 0] - xshaped[..., 1] * rope_cache[..., 1],
+            xshaped[..., 1] * rope_cache[..., 0] + xshaped[..., 0] * rope_cache[..., 1],
+        ],
+        -1,
+    )
+    x_out2 = x_out2.flatten(3)
+    return torch.cat((x_out2, x_pass), dim=-1)
+
+
+class RMSNorm(torch.nn.Module):
+    def __init__(self, normalized_shape, eps=1e-5, device=None, dtype=None, **kwargs):
+        super().__init__()
+        self.weight = torch.nn.Parameter(torch.empty(normalized_shape, device=device, dtype=dtype))
+        self.eps = eps
+
+    def forward(self, hidden_states: torch.Tensor):
+        input_dtype = hidden_states.dtype
+        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
+
+        return (self.weight * hidden_states).to(input_dtype)
+
+
+class CoreAttention(torch.nn.Module):
+    def __init__(self, config: ChatGLMConfig, layer_number):
+        super(CoreAttention, self).__init__()
+
+        self.apply_query_key_layer_scaling = config.apply_query_key_layer_scaling
+        self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
+        if self.apply_query_key_layer_scaling:
+            self.attention_softmax_in_fp32 = True
+        self.layer_number = max(1, layer_number)
+
+        projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_partition = projection_size
+        self.hidden_size_per_attention_head = projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        coeff = None
+        self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
+        if self.apply_query_key_layer_scaling:
+            coeff = self.layer_number
+            self.norm_factor *= coeff
+        self.coeff = coeff
+
+        self.attention_dropout = torch.nn.Dropout(config.attention_dropout)
+
+    def forward(self, query_layer, key_layer, value_layer, attention_mask):
+        pytorch_major_version = int(torch.__version__.split('.')[0])
+        if pytorch_major_version >= 2:
+            query_layer, key_layer, value_layer = [k.permute(1, 2, 0, 3) for k in [query_layer, key_layer, value_layer]]
+            if attention_mask is None and query_layer.shape[2] == key_layer.shape[2]:
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 is_causal=True)
+            else:
+                if attention_mask is not None:
+                    attention_mask = ~attention_mask
+                context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
+                                                                                 attention_mask)
+            context_layer = context_layer.permute(2, 0, 1, 3)
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.reshape(*new_context_layer_shape)
+        else:
+            # Raw attention scores
+
+            # [b, np, sq, sk]
+            output_size = (query_layer.size(1), query_layer.size(2), query_layer.size(0), key_layer.size(0))
+
+            # [sq, b, np, hn] -> [sq, b * np, hn]
+            query_layer = query_layer.view(output_size[2], output_size[0] * output_size[1], -1)
+            # [sk, b, np, hn] -> [sk, b * np, hn]
+            key_layer = key_layer.view(output_size[3], output_size[0] * output_size[1], -1)
+
+            # preallocting input tensor: [b * np, sq, sk]
+            matmul_input_buffer = torch.empty(
+                output_size[0] * output_size[1], output_size[2], output_size[3], dtype=query_layer.dtype,
+                device=query_layer.device
+            )
+
+            # Raw attention scores. [b * np, sq, sk]
+            matmul_result = torch.baddbmm(
+                matmul_input_buffer,
+                query_layer.transpose(0, 1),  # [b * np, sq, hn]
+                key_layer.transpose(0, 1).transpose(1, 2),  # [b * np, hn, sk]
+                beta=0.0,
+                alpha=(1.0 / self.norm_factor),
+            )
+
+            # change view to [b, np, sq, sk]
+            attention_scores = matmul_result.view(*output_size)
+
+            # ===========================
+            # Attention probs and dropout
+            # ===========================
+
+            # attention scores and attention mask [b, np, sq, sk]
+            if self.attention_softmax_in_fp32:
+                attention_scores = attention_scores.float()
+            if self.coeff is not None:
+                attention_scores = attention_scores * self.coeff
+            if attention_mask is None and attention_scores.shape[2] == attention_scores.shape[3]:
+                attention_mask = torch.ones(output_size[0], 1, output_size[2], output_size[3],
+                                            device=attention_scores.device, dtype=torch.bool)
+                attention_mask.tril_()
+                attention_mask = ~attention_mask
+            if attention_mask is not None:
+                attention_scores = attention_scores.masked_fill(attention_mask, float("-inf"))
+            attention_probs = F.softmax(attention_scores, dim=-1)
+            attention_probs = attention_probs.type_as(value_layer)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.attention_dropout(attention_probs)
+            # =========================
+            # Context layer. [sq, b, hp]
+            # =========================
+
+            # value_layer -> context layer.
+            # [sk, b, np, hn] --> [b, np, sq, hn]
+
+            # context layer shape: [b, np, sq, hn]
+            output_size = (value_layer.size(1), value_layer.size(2), query_layer.size(0), value_layer.size(3))
+            # change view [sk, b * np, hn]
+            value_layer = value_layer.view(value_layer.size(0), output_size[0] * output_size[1], -1)
+            # change view [b * np, sq, sk]
+            attention_probs = attention_probs.view(output_size[0] * output_size[1], output_size[2], -1)
+            # matmul: [b * np, sq, hn]
+            context_layer = torch.bmm(attention_probs, value_layer.transpose(0, 1))
+            # change view [b, np, sq, hn]
+            context_layer = context_layer.view(*output_size)
+            # [b, np, sq, hn] --> [sq, b, np, hn]
+            context_layer = context_layer.permute(2, 0, 1, 3).contiguous()
+            # [sq, b, np, hn] --> [sq, b, hp]
+            new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
+            context_layer = context_layer.view(*new_context_layer_shape)
+
+        return context_layer
+
+
+class SelfAttention(torch.nn.Module):
+    """Parallel self-attention layer abstract class.
+
+    Self-attention layer takes input with size [s, b, h]
+    and returns output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(SelfAttention, self).__init__()
+        self.layer_number = max(1, layer_number)
+
+        self.projection_size = config.kv_channels * config.num_attention_heads
+
+        # Per attention head and per partition values.
+        self.hidden_size_per_attention_head = self.projection_size // config.num_attention_heads
+        self.num_attention_heads_per_partition = config.num_attention_heads
+
+        self.multi_query_attention = config.multi_query_attention
+        self.qkv_hidden_size = 3 * self.projection_size
+        if self.multi_query_attention:
+            self.num_multi_query_groups_per_partition = config.multi_query_group_num
+            self.qkv_hidden_size = (
+                    self.projection_size + 2 * self.hidden_size_per_attention_head * config.multi_query_group_num
+            )
+        self.query_key_value = nn.Linear(config.hidden_size, self.qkv_hidden_size,
+                                         bias=config.add_bias_linear or config.add_qkv_bias,
+                                         device=device, **_config_to_kwargs(config)
+                                         )
+
+        self.core_attention = CoreAttention(config, self.layer_number)
+
+        # Output.
+        self.dense = nn.Linear(self.projection_size, config.hidden_size, bias=config.add_bias_linear,
+                               device=device, **_config_to_kwargs(config)
+                               )
+
+    def _allocate_memory(self, inference_max_sequence_len, batch_size, device=None, dtype=None):
+        if self.multi_query_attention:
+            num_attention_heads = self.num_multi_query_groups_per_partition
+        else:
+            num_attention_heads = self.num_attention_heads_per_partition
+        return torch.empty(
+            inference_max_sequence_len,
+            batch_size,
+            num_attention_heads,
+            self.hidden_size_per_attention_head,
+            dtype=dtype,
+            device=device,
+        )
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True
+    ):
+        # hidden_states: [sq, b, h]
+
+        # =================================================
+        # Pre-allocate memory for key-values for inference.
+        # =================================================
+        # =====================
+        # Query, Key, and Value
+        # =====================
+
+        # Attention heads [sq, b, h] --> [sq, b, (np * 3 * hn)]
+        mixed_x_layer = self.query_key_value(hidden_states)
+
+        if self.multi_query_attention:
+            (query_layer, key_layer, value_layer) = mixed_x_layer.split(
+                [
+                    self.num_attention_heads_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                    self.num_multi_query_groups_per_partition * self.hidden_size_per_attention_head,
+                ],
+                dim=-1,
+            )
+            query_layer = query_layer.view(
+                query_layer.size()[:-1] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            key_layer = key_layer.view(
+                key_layer.size()[:-1] + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.view(
+                value_layer.size()[:-1]
+                + (self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head)
+            )
+        else:
+            new_tensor_shape = mixed_x_layer.size()[:-1] + \
+                               (self.num_attention_heads_per_partition,
+                                3 * self.hidden_size_per_attention_head)
+            mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
+
+            # [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
+            (query_layer, key_layer, value_layer) = split_tensor_along_last_dim(mixed_x_layer, 3)
+
+        # apply relative positional encoding (rotary embedding)
+        if rotary_pos_emb is not None:
+            query_layer = apply_rotary_pos_emb(query_layer, rotary_pos_emb)
+            key_layer = apply_rotary_pos_emb(key_layer, rotary_pos_emb)
+
+        # adjust key and value for inference
+        if kv_cache is not None:
+            cache_k, cache_v = kv_cache
+            key_layer = torch.cat((cache_k, key_layer), dim=0)
+            value_layer = torch.cat((cache_v, value_layer), dim=0)
+        if use_cache:
+            kv_cache = (key_layer, value_layer)
+        else:
+            kv_cache = None
+
+        if self.multi_query_attention:
+            key_layer = key_layer.unsqueeze(-2)
+            key_layer = key_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            key_layer = key_layer.contiguous().view(
+                key_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+            value_layer = value_layer.unsqueeze(-2)
+            value_layer = value_layer.expand(
+                -1, -1, -1, self.num_attention_heads_per_partition // self.num_multi_query_groups_per_partition, -1
+            )
+            value_layer = value_layer.contiguous().view(
+                value_layer.size()[:2] + (self.num_attention_heads_per_partition, self.hidden_size_per_attention_head)
+            )
+
+        # ==================================
+        # core attention computation
+        # ==================================
+
+        context_layer = self.core_attention(query_layer, key_layer, value_layer, attention_mask)
+
+        # =================
+        # Output. [sq, b, h]
+        # =================
+
+        output = self.dense(context_layer)
+
+        return output, kv_cache
+
+
+def _config_to_kwargs(args):
+    common_kwargs = {
+        "dtype": args.torch_dtype,
+    }
+    return common_kwargs
+
+
+class MLP(torch.nn.Module):
+    """MLP.
+
+    MLP will take the input with h hidden state, project it to 4*h
+    hidden dimension, perform nonlinear transformation, and project the
+    state back into h hidden dimension.
+    """
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(MLP, self).__init__()
+
+        self.add_bias = config.add_bias_linear
+
+        # Project to 4h. If using swiglu double the output width, see https://arxiv.org/pdf/2002.05202.pdf
+        self.dense_h_to_4h = nn.Linear(
+            config.hidden_size,
+            config.ffn_hidden_size * 2,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+        def swiglu(x):
+            x = torch.chunk(x, 2, dim=-1)
+            return F.silu(x[0]) * x[1]
+
+        self.activation_func = swiglu
+
+        # Project back to h.
+        self.dense_4h_to_h = nn.Linear(
+            config.ffn_hidden_size,
+            config.hidden_size,
+            bias=self.add_bias,
+            device=device,
+            **_config_to_kwargs(config)
+        )
+
+    def forward(self, hidden_states):
+        # [s, b, 4hp]
+        intermediate_parallel = self.dense_h_to_4h(hidden_states)
+        intermediate_parallel = self.activation_func(intermediate_parallel)
+        # [s, b, h]
+        output = self.dense_4h_to_h(intermediate_parallel)
+        return output
+
+
+class GLMBlock(torch.nn.Module):
+    """A single transformer layer.
+
+    Transformer layer takes input with size [s, b, h] and returns an
+    output of the same size.
+    """
+
+    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
+        super(GLMBlock, self).__init__()
+        self.layer_number = layer_number
+
+        self.apply_residual_connection_post_layernorm = config.apply_residual_connection_post_layernorm
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+        LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+        # Layernorm on the input data.
+        self.input_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                             dtype=config.torch_dtype)
+
+        # Self attention.
+        self.self_attention = SelfAttention(config, layer_number, device=device)
+        self.hidden_dropout = config.hidden_dropout
+
+        # Layernorm on the attention output
+        self.post_attention_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                      dtype=config.torch_dtype)
+
+        # MLP
+        self.mlp = MLP(config, device=device)
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_cache=None, use_cache=True,
+    ):
+        # hidden_states: [s, b, h]
+
+        # Layer norm at the beginning of the transformer layer.
+        layernorm_output = self.input_layernorm(hidden_states)
+        # Self attention.
+        attention_output, kv_cache = self.self_attention(
+            layernorm_output,
+            attention_mask,
+            rotary_pos_emb,
+            kv_cache=kv_cache,
+            use_cache=use_cache
+        )
+
+        # Residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = hidden_states
+
+        layernorm_input = torch.nn.functional.dropout(attention_output, p=self.hidden_dropout, training=self.training)
+        layernorm_input = residual + layernorm_input
+
+        # Layer norm post the self attention.
+        layernorm_output = self.post_attention_layernorm(layernorm_input)
+
+        # MLP.
+        mlp_output = self.mlp(layernorm_output)
+
+        # Second residual connection.
+        if self.apply_residual_connection_post_layernorm:
+            residual = layernorm_output
+        else:
+            residual = layernorm_input
+
+        output = torch.nn.functional.dropout(mlp_output, p=self.hidden_dropout, training=self.training)
+        output = residual + output
+
+        return output, kv_cache
+
+
+class GLMTransformer(torch.nn.Module):
+    """Transformer class."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(GLMTransformer, self).__init__()
+
+        self.fp32_residual_connection = config.fp32_residual_connection
+        self.post_layer_norm = config.post_layer_norm
+
+        # Number of layers.
+        self.num_layers = config.num_layers
+
+        # Transformer layers.
+        def build_layer(layer_number):
+            return GLMBlock(config, layer_number, device=device)
+
+        self.layers = torch.nn.ModuleList([build_layer(i + 1) for i in range(self.num_layers)])
+
+        if self.post_layer_norm:
+            LayerNormFunc = RMSNorm if config.rmsnorm else LayerNorm
+            # Final layer norm before output.
+            self.final_layernorm = LayerNormFunc(config.hidden_size, eps=config.layernorm_epsilon, device=device,
+                                                 dtype=config.torch_dtype)
+
+        self.gradient_checkpointing = False
+
+    def _get_layer(self, layer_number):
+        return self.layers[layer_number]
+
+    def forward(
+            self, hidden_states, attention_mask, rotary_pos_emb, kv_caches=None,
+            use_cache: Optional[bool] = True,
+            output_hidden_states: Optional[bool] = False,
+    ):
+        if not kv_caches:
+            kv_caches = [None for _ in range(self.num_layers)]
+        presents = () if use_cache else None
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning_once(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        all_self_attentions = None
+        all_hidden_states = () if output_hidden_states else None
+        for index in range(self.num_layers):
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            layer = self._get_layer(index)
+            if self.gradient_checkpointing and self.training:
+                layer_ret = torch.utils.checkpoint.checkpoint(
+                    layer,
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_caches[index],
+                    use_cache
+                )
+            else:
+                layer_ret = layer(
+                    hidden_states,
+                    attention_mask,
+                    rotary_pos_emb,
+                    kv_cache=kv_caches[index],
+                    use_cache=use_cache
+                )
+            hidden_states, kv_cache = layer_ret
+            if use_cache:
+                presents = presents + (kv_cache,)
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        # Final layer norm.
+        if self.post_layer_norm:
+            hidden_states = self.final_layernorm(hidden_states)
+
+        return hidden_states, presents, all_hidden_states, all_self_attentions
+
+
+class ChatGLMPreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and
+    a simple interface for downloading and loading pretrained models.
+    """
+
+    is_parallelizable = False
+    supports_gradient_checkpointing = True
+    config_class = ChatGLMConfig
+    base_model_prefix = "transformer"
+    _no_split_modules = ["GLMBlock"]
+
+    def _init_weights(self, module: nn.Module):
+        """Initialize the weights."""
+        return
+
+    def get_masks(self, input_ids, past_key_values, padding_mask=None):
+        batch_size, seq_length = input_ids.shape
+        full_attention_mask = torch.ones(batch_size, seq_length, seq_length, device=input_ids.device)
+        full_attention_mask.tril_()
+        past_length = 0
+        if past_key_values:
+            past_length = past_key_values[0][0].shape[0]
+        if past_length:
+            full_attention_mask = torch.cat((torch.ones(batch_size, seq_length, past_length,
+                                                        device=input_ids.device), full_attention_mask), dim=-1)
+        if padding_mask is not None:
+            full_attention_mask = full_attention_mask * padding_mask.unsqueeze(1)
+        if not past_length and padding_mask is not None:
+            full_attention_mask -= padding_mask.unsqueeze(-1) - 1
+        full_attention_mask = (full_attention_mask < 0.5).bool()
+        full_attention_mask.unsqueeze_(1)
+        return full_attention_mask
+
+    def get_position_ids(self, input_ids, device):
+        batch_size, seq_length = input_ids.shape
+        position_ids = torch.arange(seq_length, dtype=torch.long, device=device).unsqueeze(0).repeat(batch_size, 1)
+        return position_ids
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, GLMTransformer):
+            module.gradient_checkpointing = value
+
+
+class Embedding(torch.nn.Module):
+    """Language model embeddings."""
+
+    def __init__(self, config: ChatGLMConfig, device=None):
+        super(Embedding, self).__init__()
+
+        self.hidden_size = config.hidden_size
+        # Word embeddings (parallel).
+        self.word_embeddings = nn.Embedding(
+            config.padded_vocab_size,
+            self.hidden_size,
+            dtype=config.torch_dtype,
+            device=device
+        )
+        self.fp32_residual_connection = config.fp32_residual_connection
+
+    def forward(self, input_ids):
+        # Embeddings.
+        words_embeddings = self.word_embeddings(input_ids)
+        embeddings = words_embeddings
+        # Data format change to avoid explicit tranposes : [b s h] --> [s b h].
+        embeddings = embeddings.transpose(0, 1).contiguous()
+        # If the input flag for fp32 residual connection is set, convert for float.
+        if self.fp32_residual_connection:
+            embeddings = embeddings.float()
+        return embeddings
+
+
+class ChatGLMModel(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, device=None, empty_init=True):
+        super().__init__(config)
+        if empty_init:
+            init_method = skip_init
+        else:
+            init_method = default_init
+        init_kwargs = {}
+        if device is not None:
+            init_kwargs["device"] = device
+        self.embedding = init_method(Embedding, config, **init_kwargs)
+        self.num_layers = config.num_layers
+        self.multi_query_group_num = config.multi_query_group_num
+        self.kv_channels = config.kv_channels
+
+        # Rotary positional embeddings
+        self.seq_length = config.seq_length
+        rotary_dim = (
+            config.hidden_size // config.num_attention_heads if config.kv_channels is None else config.kv_channels
+        )
+
+        self.rotary_pos_emb = RotaryEmbedding(rotary_dim // 2, original_impl=config.original_rope, device=device,
+                                              dtype=config.torch_dtype)
+        self.encoder = init_method(GLMTransformer, config, **init_kwargs)
+        self.output_layer = init_method(nn.Linear, config.hidden_size, config.padded_vocab_size, bias=False,
+                                        dtype=config.torch_dtype, **init_kwargs)
+        self.pre_seq_len = config.pre_seq_len
+        self.prefix_projection = config.prefix_projection
+        if self.pre_seq_len is not None:
+            for param in self.parameters():
+                param.requires_grad = False
+            self.prefix_tokens = torch.arange(self.pre_seq_len).long()
+            self.prefix_encoder = PrefixEncoder(config)
+            self.dropout = torch.nn.Dropout(0.1)
+
+    def get_input_embeddings(self):
+        return self.embedding.word_embeddings
+
+    def get_prompt(self, batch_size, device, dtype=torch.half):
+        prefix_tokens = self.prefix_tokens.unsqueeze(0).expand(batch_size, -1).to(device)
+        past_key_values = self.prefix_encoder(prefix_tokens).type(dtype)
+        past_key_values = past_key_values.view(
+            batch_size,
+            self.pre_seq_len,
+            self.num_layers * 2,
+            self.multi_query_group_num,
+            self.kv_channels
+        )
+        # seq_len, b, nh, hidden_size
+        past_key_values = self.dropout(past_key_values)
+        past_key_values = past_key_values.permute([2, 1, 0, 3, 4]).split(2)
+        return past_key_values
+
+    def forward(
+            self,
+            input_ids,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.BoolTensor] = None,
+            full_attention_mask: Optional[torch.BoolTensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ):
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        batch_size, seq_length = input_ids.shape
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embedding(input_ids)
+
+        if self.pre_seq_len is not None:
+            if past_key_values is None:
+                past_key_values = self.get_prompt(batch_size=batch_size, device=input_ids.device,
+                                                  dtype=inputs_embeds.dtype)
+            if attention_mask is not None:
+                attention_mask = torch.cat([attention_mask.new_ones((batch_size, self.pre_seq_len)),
+                                            attention_mask], dim=-1)
+
+        if full_attention_mask is None:
+            if (attention_mask is not None and not attention_mask.all()) or (past_key_values and seq_length != 1):
+                full_attention_mask = self.get_masks(input_ids, past_key_values, padding_mask=attention_mask)
+
+        # Rotary positional embeddings
+        rotary_pos_emb = self.rotary_pos_emb(self.seq_length)
+        if position_ids is not None:
+            rotary_pos_emb = rotary_pos_emb[position_ids]
+        else:
+            rotary_pos_emb = rotary_pos_emb[None, :seq_length]
+        rotary_pos_emb = rotary_pos_emb.transpose(0, 1).contiguous()
+
+        # Run encoder.
+        hidden_states, presents, all_hidden_states, all_self_attentions = self.encoder(
+            inputs_embeds, full_attention_mask, rotary_pos_emb=rotary_pos_emb,
+            kv_caches=past_key_values, use_cache=use_cache, output_hidden_states=output_hidden_states
+        )
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, presents, all_hidden_states, all_self_attentions] if v is not None)
+
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=presents,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+        )
+
+    def quantize(self, weight_bit_width: int):
+        from .quantization import quantize
+        quantize(self.encoder, weight_bit_width)
+        return self
+
+
+class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.max_sequence_length = config.max_length
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+        self.config = config
+        self.quantized = False
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def _update_model_kwargs_for_generation(
+            self,
+            outputs: ModelOutput,
+            model_kwargs: Dict[str, Any],
+            is_encoder_decoder: bool = False,
+            standardize_cache_format: bool = False,
+    ) -> Dict[str, Any]:
+        # update past_key_values
+        model_kwargs["past_key_values"] = self._extract_past_from_model_output(
+            outputs, standardize_cache_format=standardize_cache_format
+        )
+
+        # update attention mask
+        if "attention_mask" in model_kwargs:
+            attention_mask = model_kwargs["attention_mask"]
+            model_kwargs["attention_mask"] = torch.cat(
+                [attention_mask, attention_mask.new_ones((attention_mask.shape[0], 1))], dim=-1
+            )
+
+        # update position ids
+        if "position_ids" in model_kwargs:
+            position_ids = model_kwargs["position_ids"]
+            new_position_id = position_ids[..., -1:].clone()
+            new_position_id += 1
+            model_kwargs["position_ids"] = torch.cat(
+                [position_ids, new_position_id], dim=-1
+            )
+
+        model_kwargs["is_first_forward"] = False
+        return model_kwargs
+
+    def prepare_inputs_for_generation(
+            self,
+            input_ids: torch.LongTensor,
+            past_key_values: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            is_first_forward: bool = True,
+            **kwargs
+    ) -> dict:
+        # only last token for input_ids if past is not None
+        if position_ids is None:
+            position_ids = self.get_position_ids(input_ids, device=input_ids.device)
+        if not is_first_forward:
+            if past_key_values is not None:
+                position_ids = position_ids[..., -1:]
+                input_ids = input_ids[:, -1:]
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past_key_values,
+            "position_ids": position_ids,
+            "attention_mask": attention_mask,
+            "return_last_logit": True,
+            "use_cache": use_cache
+        }
+
+    def forward(
+            self,
+            input_ids: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.Tensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[torch.FloatTensor]] = None,
+            inputs_embeds: Optional[torch.Tensor] = None,
+            labels: Optional[torch.Tensor] = None,
+            use_cache: Optional[bool] = None,
+            output_attentions: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+            return_last_logit: Optional[bool] = False,
+    ):
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        if return_last_logit:
+            hidden_states = hidden_states[-1:]
+        lm_logits = self.transformer.output_layer(hidden_states)
+        lm_logits = lm_logits.transpose(0, 1).contiguous()
+
+        loss = None
+        if labels is not None:
+            lm_logits = lm_logits.to(torch.float32)
+
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss(ignore_index=-100)
+            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+
+            lm_logits = lm_logits.to(hidden_states.dtype)
+            loss = loss.to(hidden_states.dtype)
+
+        if not return_dict:
+            output = (lm_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+    @staticmethod
+    def _reorder_cache(
+            past: Tuple[Tuple[torch.Tensor, torch.Tensor], ...], beam_idx: torch.LongTensor
+    ) -> Tuple[Tuple[torch.Tensor, torch.Tensor], ...]:
+        """
+        This function is used to re-order the `past_key_values` cache if [`~PreTrainedModel.beam_search`] or
+        [`~PreTrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
+        beam_idx at every generation step.
+
+        Output shares the same memory storage as `past`.
+        """
+        return tuple(
+            (
+                layer_past[0].index_select(1, beam_idx.to(layer_past[0].device)),
+                layer_past[1].index_select(1, beam_idx.to(layer_past[1].device)),
+            )
+            for layer_past in past
+        )
+
+    def process_response(self, response):
+        response = response.strip()
+        response = response.replace("[[训练时间]]", "2023年")
+        return response
+
+    def build_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        prompt = tokenizer.build_prompt(query, history=history)
+        inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    def build_stream_inputs(self, tokenizer, query: str, history: List[Tuple[str, str]] = None):
+        if history:
+            prompt = "\n\n[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            input_ids = tokenizer.encode(prompt, add_special_tokens=False)
+            input_ids = input_ids[1:]
+            inputs = tokenizer.batch_encode_plus([(input_ids, None)], return_tensors="pt", add_special_tokens=False)
+        else:
+            prompt = "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+            inputs = tokenizer([prompt], return_tensors="pt")
+        inputs = inputs.to(self.device)
+        return inputs
+
+    @torch.inference_mode()
+    def chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, max_length: int = 8192, num_beams=1,
+             do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "num_beams": num_beams, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        inputs = self.build_inputs(tokenizer, query, history=history)
+        outputs = self.generate(**inputs, **gen_kwargs)
+        outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+        response = tokenizer.decode(outputs)
+        response = self.process_response(response)
+        history = history + [(query, response)]
+        return response, history
+
+    @torch.inference_mode()
+    def stream_chat(self, tokenizer, query: str, history: List[Tuple[str, str]] = None, past_key_values=None,
+                    max_length: int = 8192, do_sample=True, top_p=0.8, temperature=0.8, logits_processor=None,
+                    return_past_key_values=False, **kwargs):
+        if history is None:
+            history = []
+        if logits_processor is None:
+            logits_processor = LogitsProcessorList()
+        logits_processor.append(InvalidScoreLogitsProcessor())
+        gen_kwargs = {"max_length": max_length, "do_sample": do_sample, "top_p": top_p,
+                      "temperature": temperature, "logits_processor": logits_processor, **kwargs}
+        if past_key_values is None and not return_past_key_values:
+            inputs = self.build_inputs(tokenizer, query, history=history)
+        else:
+            inputs = self.build_stream_inputs(tokenizer, query, history=history)
+        if past_key_values is not None:
+            past_length = past_key_values[0][0].shape[0]
+            if self.transformer.pre_seq_len is not None:
+                past_length -= self.transformer.pre_seq_len
+            inputs.position_ids += past_length
+            attention_mask = inputs.attention_mask
+            attention_mask = torch.cat((attention_mask.new_ones(1, past_length), attention_mask), dim=1)
+            inputs['attention_mask'] = attention_mask
+        for outputs in self.stream_generate(**inputs, past_key_values=past_key_values,
+                                            return_past_key_values=return_past_key_values, **gen_kwargs):
+            if return_past_key_values:
+                outputs, past_key_values = outputs
+            outputs = outputs.tolist()[0][len(inputs["input_ids"][0]):]
+            response = tokenizer.decode(outputs)
+            if response and response[-1] != "�":
+                response = self.process_response(response)
+                new_history = history + [(query, response)]
+                if return_past_key_values:
+                    yield response, new_history, past_key_values
+                else:
+                    yield response, new_history
+
+    @torch.inference_mode()
+    def stream_generate(
+            self,
+            input_ids,
+            generation_config: Optional[GenerationConfig] = None,
+            logits_processor: Optional[LogitsProcessorList] = None,
+            stopping_criteria: Optional[StoppingCriteriaList] = None,
+            prefix_allowed_tokens_fn: Optional[Callable[[int, torch.Tensor], List[int]]] = None,
+            return_past_key_values=False,
+            **kwargs,
+    ):
+        batch_size, input_ids_seq_length = input_ids.shape[0], input_ids.shape[-1]
+
+        if generation_config is None:
+            generation_config = self.generation_config
+        generation_config = copy.deepcopy(generation_config)
+        model_kwargs = generation_config.update(**kwargs)
+        model_kwargs["use_cache"] = generation_config.use_cache
+        bos_token_id, eos_token_id = generation_config.bos_token_id, generation_config.eos_token_id
+
+        if isinstance(eos_token_id, int):
+            eos_token_id = [eos_token_id]
+
+        has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
+        if has_default_max_length and generation_config.max_new_tokens is None:
+            warnings.warn(
+                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
+                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
+                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                UserWarning,
+            )
+        elif generation_config.max_new_tokens is not None:
+            generation_config.max_length = generation_config.max_new_tokens + input_ids_seq_length
+            if not has_default_max_length:
+                logger.warn(
+                    f"Both `max_new_tokens` (={generation_config.max_new_tokens}) and `max_length`(="
+                    f"{generation_config.max_length}) seem to have been set. `max_new_tokens` will take precedence. "
+                    "Please refer to the documentation for more information. "
+                    "(https://huggingface.co/docs/transformers/main/en/main_classes/text_generation)",
+                    UserWarning,
+                )
+
+        if input_ids_seq_length >= generation_config.max_length:
+            input_ids_string = "decoder_input_ids" if self.config.is_encoder_decoder else "input_ids"
+            logger.warning(
+                f"Input length of {input_ids_string} is {input_ids_seq_length}, but `max_length` is set to"
+                f" {generation_config.max_length}. This can lead to unexpected behavior. You should consider"
+                " increasing `max_new_tokens`."
+            )
+
+        # 2. Set generation parameters if not already defined
+        logits_processor = logits_processor if logits_processor is not None else LogitsProcessorList()
+        stopping_criteria = stopping_criteria if stopping_criteria is not None else StoppingCriteriaList()
+
+        logits_processor = self._get_logits_processor(
+            generation_config=generation_config,
+            input_ids_seq_length=input_ids_seq_length,
+            encoder_input_ids=input_ids,
+            prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
+            logits_processor=logits_processor,
+        )
+
+        stopping_criteria = self._get_stopping_criteria(
+            generation_config=generation_config, stopping_criteria=stopping_criteria
+        )
+        logits_warper = self._get_logits_warper(generation_config)
+
+        unfinished_sequences = input_ids.new(input_ids.shape[0]).fill_(1)
+        scores = None
+        while True:
+            model_inputs = self.prepare_inputs_for_generation(input_ids, **model_kwargs)
+            # forward pass to get next token
+            outputs = self(
+                **model_inputs,
+                return_dict=True,
+                output_attentions=False,
+                output_hidden_states=False,
+            )
+
+            next_token_logits = outputs.logits[:, -1, :]
+
+            # pre-process distribution
+            next_token_scores = logits_processor(input_ids, next_token_logits)
+            next_token_scores = logits_warper(input_ids, next_token_scores)
+
+            # sample
+            probs = nn.functional.softmax(next_token_scores, dim=-1)
+            if generation_config.do_sample:
+                next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
+            else:
+                next_tokens = torch.argmax(probs, dim=-1)
+
+            # update generated ids, model inputs, and length for next step
+            input_ids = torch.cat([input_ids, next_tokens[:, None]], dim=-1)
+            model_kwargs = self._update_model_kwargs_for_generation(
+                outputs, model_kwargs, is_encoder_decoder=self.config.is_encoder_decoder
+            )
+            unfinished_sequences = unfinished_sequences.mul((sum(next_tokens != i for i in eos_token_id)).long())
+            if return_past_key_values:
+                yield input_ids, outputs.past_key_values
+            else:
+                yield input_ids
+            # stop when each sentence is finished, or if we exceed the maximum length
+            if unfinished_sequences.max() == 0 or stopping_criteria(input_ids, scores):
+                break
+
+    def quantize(self, bits: int, empty_init=False, device=None, **kwargs):
+        if bits == 0:
+            return
+
+        from .quantization import quantize
+
+        if self.quantized:
+            logger.info("Already quantized.")
+            return self
+
+        self.quantized = True
+
+        self.config.quantization_bit = bits
+
+        self.transformer.encoder = quantize(self.transformer.encoder, bits, empty_init=empty_init, device=device,
+                                            **kwargs)
+        return self
+
+
+class ChatGLMForSequenceClassification(ChatGLMPreTrainedModel):
+    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
+        super().__init__(config)
+
+        self.num_labels = config.num_labels
+        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
+
+        self.classifier_head = nn.Linear(config.hidden_size, config.num_labels, bias=True, dtype=torch.half)
+        if config.classifier_dropout is not None:
+            self.dropout = nn.Dropout(config.classifier_dropout)
+        else:
+            self.dropout = None
+        self.config = config
+
+        if self.config.quantization_bit:
+            self.quantize(self.config.quantization_bit, empty_init=True)
+
+    def forward(
+            self,
+            input_ids: Optional[torch.LongTensor] = None,
+            position_ids: Optional[torch.LongTensor] = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            full_attention_mask: Optional[torch.Tensor] = None,
+            past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+            inputs_embeds: Optional[torch.LongTensor] = None,
+            labels: Optional[torch.LongTensor] = None,
+            use_cache: Optional[bool] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor, ...], SequenceClassifierOutputWithPast]:
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            full_attention_mask=full_attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        pooled_hidden_states = hidden_states[-1]
+        if self.dropout is not None:
+            pooled_hidden_states = self.dropout(pooled_hidden_states)
+        logits = self.classifier_head(pooled_hidden_states)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze().float(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits.float(), labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(logits.view(-1, self.num_labels).float(), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits.float(), labels.view(-1, self.num_labels))
+
+        if not return_dict:
+            output = (logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )

checkpoint-3000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d933d739be3ea081921f3f2720b05f67648f7469fcd8090da4997203ccd8d0ee
+size 14681455

checkpoint-3000/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c027dc9dc576448d290ff5857990857beba3d336cdaed8ff78183b6eaaeb9a8
+size 7340861

checkpoint-3000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cff56eb2132f83374842f7d3d336ebe853f1a9cd3e56b4675981d6fadecb6c5b
+size 14575

checkpoint-3000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:789a1605f05260342a605be992efe3dec445d76b75c62bdef618345fc0105cc0
+size 627

checkpoint-3000/special_tokens_map.json

@@ -0,0 +1 @@
+{}

checkpoint-3000/tokenization_chatglm.py

@@ -0,0 +1,257 @@
+import os
+import torch
+from typing import List, Optional, Union, Dict
+from sentencepiece import SentencePieceProcessor
+from transformers import PreTrainedTokenizer
+from transformers.utils import logging, PaddingStrategy
+from transformers.tokenization_utils_base import EncodedInput, BatchEncoding
+
+
+class SPTokenizer:
+    def __init__(self, model_path: str):
+        # reload tokenizer
+        assert os.path.isfile(model_path), model_path
+        self.sp_model = SentencePieceProcessor(model_file=model_path)
+
+        # BOS / EOS token IDs
+        self.n_words: int = self.sp_model.vocab_size()
+        self.bos_id: int = self.sp_model.bos_id()
+        self.eos_id: int = self.sp_model.eos_id()
+        self.pad_id: int = self.sp_model.unk_id()
+        assert self.sp_model.vocab_size() == self.sp_model.get_piece_size()
+
+        special_tokens = ["[MASK]", "[gMASK]", "[sMASK]", "sop", "eop"]
+        self.special_tokens = {}
+        self.index_special_tokens = {}
+        for token in special_tokens:
+            self.special_tokens[token] = self.n_words
+            self.index_special_tokens[self.n_words] = token
+            self.n_words += 1
+
+    def tokenize(self, s: str):
+        return self.sp_model.EncodeAsPieces(s)
+
+    def encode(self, s: str, bos: bool = False, eos: bool = False) -> List[int]:
+        assert type(s) is str
+        t = self.sp_model.encode(s)
+        if bos:
+            t = [self.bos_id] + t
+        if eos:
+            t = t + [self.eos_id]
+        return t
+
+    def decode(self, t: List[int]) -> str:
+        return self.sp_model.decode(t)
+
+    def decode_tokens(self, tokens: List[str]) -> str:
+        text = self.sp_model.DecodePieces(tokens)
+        return text
+
+    def convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        return self.sp_model.PieceToId(token)
+
+    def convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        if index in self.index_special_tokens or index in [self.eos_id, self.bos_id, self.pad_id] or index < 0:
+            return ""
+        return self.sp_model.IdToPiece(index)
+
+
+class ChatGLMTokenizer(PreTrainedTokenizer):
+    vocab_files_names = {"vocab_file": "tokenizer.model"}
+
+    model_input_names = ["input_ids", "attention_mask", "position_ids"]
+
+    def __init__(self, vocab_file, padding_side="left", clean_up_tokenization_spaces=False, **kwargs):
+        self.name = "GLMTokenizer"
+
+        self.vocab_file = vocab_file
+        self.tokenizer = SPTokenizer(vocab_file)
+        self.special_tokens = {
+            "<bos>": self.tokenizer.bos_id,
+            "<eos>": self.tokenizer.eos_id,
+            "<pad>": self.tokenizer.pad_id
+        }
+        super().__init__(padding_side=padding_side, clean_up_tokenization_spaces=clean_up_tokenization_spaces, **kwargs)
+
+    def get_command(self, token):
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        assert token in self.tokenizer.special_tokens, f"{token} is not a special token for {self.name}"
+        return self.tokenizer.special_tokens[token]
+
+    @property
+    def unk_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token(self) -> str:
+        return "<unk>"
+
+    @property
+    def pad_token_id(self):
+        return self.get_command("<pad>")
+
+    @property
+    def eos_token(self) -> str:
+        return "</s>"
+
+    @property
+    def eos_token_id(self):
+        return self.get_command("<eos>")
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_words
+
+    def get_vocab(self):
+        """ Returns vocab as a dict """
+        vocab = {self._convert_id_to_token(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text, **kwargs):
+        return self.tokenizer.tokenize(text)
+
+    def _convert_token_to_id(self, token):
+        """ Converts a token (str) in an id using the vocab. """
+        return self.tokenizer.convert_token_to_id(token)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        return self.tokenizer.convert_id_to_token(index)
+
+    def convert_tokens_to_string(self, tokens: List[str]) -> str:
+        return self.tokenizer.decode_tokens(tokens)
+
+    def save_vocabulary(self, save_directory, filename_prefix=None):
+        """
+        Save the vocabulary and special tokens file to a directory.
+
+        Args:
+            save_directory (`str`):
+                The directory in which to save the vocabulary.
+            filename_prefix (`str`, *optional*):
+                An optional prefix to add to the named of the saved files.
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        if os.path.isdir(save_directory):
+            vocab_file = os.path.join(
+                save_directory, self.vocab_files_names["vocab_file"]
+            )
+        else:
+            vocab_file = save_directory
+
+        with open(self.vocab_file, 'rb') as fin:
+            proto_str = fin.read()
+
+        with open(vocab_file, "wb") as writer:
+            writer.write(proto_str)
+
+        return (vocab_file,)
+
+    def get_prefix_tokens(self):
+        prefix_tokens = [self.get_command("[gMASK]"), self.get_command("sop")]
+        return prefix_tokens
+
+    def build_prompt(self, query, history=None):
+        if history is None:
+            history = []
+        prompt = ""
+        for i, (old_query, response) in enumerate(history):
+            prompt += "[Round {}]\n\n问：{}\n\n答：{}\n\n".format(i + 1, old_query, response)
+        prompt += "[Round {}]\n\n问：{}\n\n答：".format(len(history) + 1, query)
+        return prompt
+
+    def build_inputs_with_special_tokens(
+            self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. A BERT sequence has the following format:
+
+        - single sequence: `[CLS] X [SEP]`
+        - pair of sequences: `[CLS] A [SEP] B [SEP]`
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        prefix_tokens = self.get_prefix_tokens()
+        token_ids_0 = prefix_tokens + token_ids_0
+        if token_ids_1 is not None:
+            token_ids_0 = token_ids_0 + token_ids_1 + [self.get_command("<eos>")]
+        return token_ids_0
+
+    def _pad(
+            self,
+            encoded_inputs: Union[Dict[str, EncodedInput], BatchEncoding],
+            max_length: Optional[int] = None,
+            padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
+            pad_to_multiple_of: Optional[int] = None,
+            return_attention_mask: Optional[bool] = None,
+    ) -> dict:
+        """
+        Pad encoded inputs (on left/right and up to predefined length or max length in the batch)
+
+        Args:
+            encoded_inputs:
+                Dictionary of tokenized inputs (`List[int]`) or batch of tokenized inputs (`List[List[int]]`).
+            max_length: maximum length of the returned list and optionally padding length (see below).
+                Will truncate by taking into account the special tokens.
+            padding_strategy: PaddingStrategy to use for padding.
+
+                - PaddingStrategy.LONGEST Pad to the longest sequence in the batch
+                - PaddingStrategy.MAX_LENGTH: Pad to the max length (default)
+                - PaddingStrategy.DO_NOT_PAD: Do not pad
+                The tokenizer padding sides are defined in self.padding_side:
+
+                    - 'left': pads on the left of the sequences
+                    - 'right': pads on the right of the sequences
+            pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
+                This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
+                `>= 7.5` (Volta).
+            return_attention_mask:
+                (optional) Set to False to avoid returning attention mask (default: set to model specifics)
+        """
+        # Load from model defaults
+        assert self.padding_side == "left"
+
+        required_input = encoded_inputs[self.model_input_names[0]]
+        seq_length = len(required_input)
+
+        if padding_strategy == PaddingStrategy.LONGEST:
+            max_length = len(required_input)
+
+        if max_length is not None and pad_to_multiple_of is not None and (max_length % pad_to_multiple_of != 0):
+            max_length = ((max_length // pad_to_multiple_of) + 1) * pad_to_multiple_of
+
+        needs_to_be_padded = padding_strategy != PaddingStrategy.DO_NOT_PAD and len(required_input) != max_length
+
+        # Initialize attention mask if not present.
+        if "attention_mask" not in encoded_inputs:
+            encoded_inputs["attention_mask"] = [1] * seq_length
+
+        if "position_ids" not in encoded_inputs:
+            encoded_inputs["position_ids"] = list(range(seq_length))
+
+        if needs_to_be_padded:
+            difference = max_length - len(required_input)
+
+            if "attention_mask" in encoded_inputs:
+                encoded_inputs["attention_mask"] = [0] * difference + encoded_inputs["attention_mask"]
+            if "position_ids" in encoded_inputs:
+                encoded_inputs["position_ids"] = [0] * difference + encoded_inputs["position_ids"]
+            encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input
+
+        return encoded_inputs

checkpoint-3000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7dc4c393423b76e4373e5157ddc34803a0189ba96b21ddbb40269d31468a6f2
+size 1018370

checkpoint-3000/tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_chatglm.ChatGLMTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "padding_side": "left",
+  "remove_space": false,
+  "tokenizer_class": "ChatGLMTokenizer"
+}

checkpoint-3000/trainer_state.json

@@ -0,0 +1,1816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2770786672516105,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019933333333333334,
+      "loss": 2.6644,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019866666666666668,
+      "loss": 1.7151,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0198,
+      "loss": 1.6228,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019733333333333335,
+      "loss": 1.401,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019666666666666666,
+      "loss": 1.6172,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0196,
+      "loss": 1.4695,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019533333333333333,
+      "loss": 1.5137,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019466666666666667,
+      "loss": 1.5425,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0194,
+      "loss": 1.4272,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019333333333333334,
+      "loss": 1.3727,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019266666666666668,
+      "loss": 1.3114,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0192,
+      "loss": 1.4758,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019133333333333332,
+      "loss": 1.5219,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.01906666666666667,
+      "loss": 1.376,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019,
+      "loss": 1.4257,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.018933333333333333,
+      "loss": 1.3474,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018866666666666667,
+      "loss": 1.2929,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0188,
+      "loss": 1.3208,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018733333333333334,
+      "loss": 1.3381,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018666666666666668,
+      "loss": 1.3644,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018600000000000002,
+      "loss": 1.2932,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018533333333333332,
+      "loss": 1.4092,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018466666666666666,
+      "loss": 1.3006,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0184,
+      "loss": 1.4572,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018333333333333333,
+      "loss": 1.2789,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018266666666666667,
+      "loss": 1.4444,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0182,
+      "loss": 1.4511,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01813333333333333,
+      "loss": 1.3541,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01806666666666667,
+      "loss": 1.3228,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.018000000000000002,
+      "loss": 1.3185,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017933333333333332,
+      "loss": 1.199,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017866666666666666,
+      "loss": 1.3417,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0178,
+      "loss": 1.4251,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017733333333333334,
+      "loss": 1.3574,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017666666666666667,
+      "loss": 1.2547,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0176,
+      "loss": 1.2651,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017533333333333335,
+      "loss": 1.3414,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017466666666666665,
+      "loss": 1.3322,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0174,
+      "loss": 1.4147,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017333333333333336,
+      "loss": 1.2813,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017266666666666666,
+      "loss": 1.3687,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0172,
+      "loss": 1.5593,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017133333333333334,
+      "loss": 1.3073,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017066666666666667,
+      "loss": 1.2359,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017,
+      "loss": 1.2474,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.016933333333333335,
+      "loss": 1.3874,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.01686666666666667,
+      "loss": 1.3203,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0168,
+      "loss": 1.2875,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016733333333333333,
+      "loss": 1.2767,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016666666666666666,
+      "loss": 1.3017,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0166,
+      "loss": 1.2321,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016533333333333334,
+      "loss": 1.1719,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016466666666666668,
+      "loss": 1.2552,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016399999999999998,
+      "loss": 1.3816,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01633333333333333,
+      "loss": 1.2956,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01626666666666667,
+      "loss": 1.2061,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016200000000000003,
+      "loss": 1.2086,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016133333333333333,
+      "loss": 1.1633,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016066666666666667,
+      "loss": 1.2638,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.016,
+      "loss": 1.3441,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015933333333333334,
+      "loss": 1.2924,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015866666666666668,
+      "loss": 1.1818,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0158,
+      "loss": 1.3918,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015733333333333332,
+      "loss": 1.2232,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015666666666666666,
+      "loss": 1.2472,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015600000000000001,
+      "loss": 1.2398,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015533333333333333,
+      "loss": 1.3649,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015466666666666667,
+      "loss": 1.2302,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0154,
+      "loss": 1.2053,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015333333333333334,
+      "loss": 1.2974,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015266666666666666,
+      "loss": 1.3036,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0152,
+      "loss": 1.3162,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015133333333333334,
+      "loss": 1.2567,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015066666666666666,
+      "loss": 1.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015,
+      "loss": 1.2692,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014933333333333335,
+      "loss": 1.1332,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014866666666666667,
+      "loss": 1.2949,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0148,
+      "loss": 1.2703,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014733333333333334,
+      "loss": 1.3891,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014666666666666666,
+      "loss": 1.3594,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0146,
+      "loss": 1.166,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014533333333333334,
+      "loss": 1.3256,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014466666666666668,
+      "loss": 1.2669,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0144,
+      "loss": 1.241,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014333333333333333,
+      "loss": 1.2591,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014266666666666667,
+      "loss": 1.238,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014199999999999999,
+      "loss": 1.3583,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014133333333333333,
+      "loss": 1.164,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014066666666666668,
+      "loss": 1.2367,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013999999999999999,
+      "loss": 1.1864,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013933333333333334,
+      "loss": 1.2259,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013866666666666668,
+      "loss": 1.2129,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0138,
+      "loss": 1.2085,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013733333333333334,
+      "loss": 1.2316,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013666666666666667,
+      "loss": 1.2721,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013600000000000001,
+      "loss": 1.2428,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013533333333333333,
+      "loss": 1.2126,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013466666666666667,
+      "loss": 1.1583,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0134,
+      "loss": 1.2776,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013333333333333332,
+      "loss": 1.2317,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013266666666666666,
+      "loss": 1.2047,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013200000000000002,
+      "loss": 1.202,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013133333333333332,
+      "loss": 1.1877,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013066666666666667,
+      "loss": 1.3071,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013000000000000001,
+      "loss": 1.2976,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012933333333333333,
+      "loss": 1.4156,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012866666666666667,
+      "loss": 1.173,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0128,
+      "loss": 1.2263,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012733333333333334,
+      "loss": 1.3235,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012666666666666666,
+      "loss": 1.1923,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0126,
+      "loss": 1.2879,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012533333333333334,
+      "loss": 1.196,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012466666666666666,
+      "loss": 1.2605,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0124,
+      "loss": 1.2303,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012333333333333335,
+      "loss": 1.2381,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012266666666666665,
+      "loss": 1.2709,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0122,
+      "loss": 1.2121,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012133333333333335,
+      "loss": 1.3713,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012066666666666668,
+      "loss": 1.2923,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012,
+      "loss": 1.2947,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011933333333333334,
+      "loss": 1.1538,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011866666666666668,
+      "loss": 1.1312,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0118,
+      "loss": 1.1807,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011733333333333333,
+      "loss": 1.2729,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011666666666666667,
+      "loss": 1.21,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0116,
+      "loss": 1.1986,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011533333333333333,
+      "loss": 1.2003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011466666666666667,
+      "loss": 1.1773,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011399999999999999,
+      "loss": 1.3241,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011333333333333332,
+      "loss": 1.2157,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011266666666666668,
+      "loss": 1.2549,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011200000000000002,
+      "loss": 1.3245,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011133333333333334,
+      "loss": 1.2109,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011066666666666667,
+      "loss": 1.1979,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011000000000000001,
+      "loss": 1.2804,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010933333333333333,
+      "loss": 1.2655,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010866666666666667,
+      "loss": 1.1264,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0108,
+      "loss": 1.2949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010733333333333333,
+      "loss": 1.2038,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010666666666666666,
+      "loss": 1.2514,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0106,
+      "loss": 1.1692,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010533333333333332,
+      "loss": 1.1947,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010466666666666666,
+      "loss": 1.3294,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010400000000000001,
+      "loss": 1.2169,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010333333333333335,
+      "loss": 1.3113,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010266666666666667,
+      "loss": 1.1322,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0102,
+      "loss": 1.4228,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010133333333333334,
+      "loss": 1.2384,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010066666666666666,
+      "loss": 1.2107,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.01,
+      "loss": 1.2655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009933333333333334,
+      "loss": 1.2991,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009866666666666668,
+      "loss": 1.324,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0098,
+      "loss": 1.3443,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009733333333333333,
+      "loss": 1.1389,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009666666666666667,
+      "loss": 1.2308,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0096,
+      "loss": 1.1847,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009533333333333335,
+      "loss": 1.3154,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009466666666666667,
+      "loss": 1.233,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0094,
+      "loss": 1.147,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009333333333333334,
+      "loss": 1.1824,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009266666666666666,
+      "loss": 1.2093,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0092,
+      "loss": 1.2111,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009133333333333334,
+      "loss": 1.0596,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009066666666666666,
+      "loss": 1.3025,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009000000000000001,
+      "loss": 1.1726,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008933333333333333,
+      "loss": 1.2078,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008866666666666667,
+      "loss": 1.2652,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0088,
+      "loss": 1.2033,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008733333333333333,
+      "loss": 1.1598,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008666666666666668,
+      "loss": 1.1904,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0086,
+      "loss": 1.242,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008533333333333334,
+      "loss": 1.3042,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008466666666666667,
+      "loss": 1.3653,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0084,
+      "loss": 1.1784,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008333333333333333,
+      "loss": 1.2306,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008266666666666667,
+      "loss": 1.2139,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008199999999999999,
+      "loss": 1.1891,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008133333333333334,
+      "loss": 1.2619,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008066666666666666,
+      "loss": 1.0873,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008,
+      "loss": 1.2537,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007933333333333334,
+      "loss": 1.2575,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007866666666666666,
+      "loss": 1.1043,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0078000000000000005,
+      "loss": 1.2063,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007733333333333333,
+      "loss": 1.1602,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007666666666666667,
+      "loss": 1.1474,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0076,
+      "loss": 1.1482,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007533333333333333,
+      "loss": 1.2124,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074666666666666675,
+      "loss": 1.195,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074,
+      "loss": 1.1426,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007333333333333333,
+      "loss": 1.2067,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007266666666666667,
+      "loss": 1.1649,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0072,
+      "loss": 1.0978,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0071333333333333335,
+      "loss": 1.2298,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007066666666666666,
+      "loss": 1.195,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006999999999999999,
+      "loss": 1.2032,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006933333333333334,
+      "loss": 1.1134,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006866666666666667,
+      "loss": 1.2925,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0068000000000000005,
+      "loss": 1.1389,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006733333333333333,
+      "loss": 1.1952,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006666666666666666,
+      "loss": 1.0672,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006600000000000001,
+      "loss": 1.2243,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006533333333333334,
+      "loss": 1.218,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006466666666666667,
+      "loss": 1.1988,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0064,
+      "loss": 1.1776,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006333333333333333,
+      "loss": 1.1506,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006266666666666667,
+      "loss": 1.0386,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0062,
+      "loss": 1.2601,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006133333333333333,
+      "loss": 1.058,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006066666666666667,
+      "loss": 1.2243,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006,
+      "loss": 1.2445,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.005933333333333334,
+      "loss": 1.2567,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005866666666666667,
+      "loss": 1.1815,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0058,
+      "loss": 1.3061,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005733333333333333,
+      "loss": 1.3064,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005666666666666666,
+      "loss": 1.2739,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005600000000000001,
+      "loss": 1.2754,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005533333333333334,
+      "loss": 1.1678,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0054666666666666665,
+      "loss": 1.191,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0054,
+      "loss": 1.2127,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005333333333333333,
+      "loss": 1.3057,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005266666666666666,
+      "loss": 1.1092,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005200000000000001,
+      "loss": 1.1615,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0051333333333333335,
+      "loss": 1.1769,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005066666666666667,
+      "loss": 1.1396,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005,
+      "loss": 1.1663,
+      "step": 2250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004933333333333334,
+      "loss": 1.0931,
+      "step": 2260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004866666666666667,
+      "loss": 1.1143,
+      "step": 2270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0048,
+      "loss": 1.1891,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004733333333333333,
+      "loss": 1.1636,
+      "step": 2290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004666666666666667,
+      "loss": 1.1777,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0046,
+      "loss": 1.1933,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004533333333333333,
+      "loss": 1.1606,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0044666666666666665,
+      "loss": 1.2233,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0044,
+      "loss": 1.2237,
+      "step": 2340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004333333333333334,
+      "loss": 1.2323,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004266666666666667,
+      "loss": 1.0837,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0042,
+      "loss": 1.1411,
+      "step": 2370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0041333333333333335,
+      "loss": 1.0772,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004066666666666667,
+      "loss": 1.1952,
+      "step": 2390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004,
+      "loss": 1.1267,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.003933333333333333,
+      "loss": 1.2142,
+      "step": 2410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0038666666666666667,
+      "loss": 1.2375,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0038,
+      "loss": 1.1964,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0037333333333333337,
+      "loss": 1.1305,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0036666666666666666,
+      "loss": 1.1478,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0036,
+      "loss": 1.1542,
+      "step": 2460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003533333333333333,
+      "loss": 1.1168,
+      "step": 2470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003466666666666667,
+      "loss": 1.1185,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0034000000000000002,
+      "loss": 1.0708,
+      "step": 2490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003333333333333333,
+      "loss": 1.1487,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003266666666666667,
+      "loss": 1.1814,
+      "step": 2510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0032,
+      "loss": 1.1583,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0031333333333333335,
+      "loss": 1.2117,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0030666666666666663,
+      "loss": 1.1998,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.003,
+      "loss": 1.2355,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0029333333333333334,
+      "loss": 1.2694,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0028666666666666667,
+      "loss": 1.1819,
+      "step": 2570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0028000000000000004,
+      "loss": 1.1469,
+      "step": 2580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0027333333333333333,
+      "loss": 1.1726,
+      "step": 2590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0026666666666666666,
+      "loss": 1.0332,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0026000000000000003,
+      "loss": 1.2277,
+      "step": 2610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0025333333333333336,
+      "loss": 1.1335,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.002466666666666667,
+      "loss": 1.0854,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0024,
+      "loss": 1.1181,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0023333333333333335,
+      "loss": 1.1004,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0022666666666666664,
+      "loss": 1.1311,
+      "step": 2660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0022,
+      "loss": 1.0965,
+      "step": 2670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0021333333333333334,
+      "loss": 1.2944,
+      "step": 2680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0020666666666666667,
+      "loss": 1.1267,
+      "step": 2690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.002,
+      "loss": 1.0006,
+      "step": 2700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019333333333333333,
+      "loss": 1.1332,
+      "step": 2710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0018666666666666669,
+      "loss": 1.133,
+      "step": 2720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0018,
+      "loss": 1.1696,
+      "step": 2730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017333333333333335,
+      "loss": 1.2066,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0016666666666666666,
+      "loss": 1.1698,
+      "step": 2750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0016,
+      "loss": 1.1531,
+      "step": 2760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0015333333333333332,
+      "loss": 1.333,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0014666666666666667,
+      "loss": 1.0968,
+      "step": 2780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0014000000000000002,
+      "loss": 1.2056,
+      "step": 2790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0013333333333333333,
+      "loss": 1.1381,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0012666666666666668,
+      "loss": 1.1355,
+      "step": 2810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0012,
+      "loss": 1.289,
+      "step": 2820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0011333333333333332,
+      "loss": 1.16,
+      "step": 2830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0010666666666666667,
+      "loss": 1.1275,
+      "step": 2840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001,
+      "loss": 1.0971,
+      "step": 2850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009333333333333334,
+      "loss": 1.0893,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008666666666666667,
+      "loss": 1.1557,
+      "step": 2870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008,
+      "loss": 1.1966,
+      "step": 2880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007333333333333333,
+      "loss": 1.2223,
+      "step": 2890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 1.1097,
+      "step": 2900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0006,
+      "loss": 1.1612,
+      "step": 2910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005333333333333334,
+      "loss": 1.2289,
+      "step": 2920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004666666666666667,
+      "loss": 1.222,
+      "step": 2930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004,
+      "loss": 1.1519,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 1.3762,
+      "step": 2950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 1.2506,
+      "step": 2960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002,
+      "loss": 1.2153,
+      "step": 2970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.1327,
+      "step": 2980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.2242,
+      "step": 2990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0,
+      "loss": 1.1963,
+      "step": 3000
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 1,
+  "total_flos": 1.73594313916416e+17,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70ac9eb43a2d446e07cde6bbbb21250fe0373a4093de76a8aa1f7223e3836bcd
+size 4155

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 0.28,
+    "train_loss": 1.2429857851664226,
+    "train_runtime": 27386.6257,
+    "train_samples": 173235,
+    "train_samples_per_second": 1.753,
+    "train_steps_per_second": 0.11
+}

trainer_state.json

@@ -0,0 +1,1825 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2770786672516105,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019933333333333334,
+      "loss": 2.6644,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019866666666666668,
+      "loss": 1.7151,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0198,
+      "loss": 1.6228,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019733333333333335,
+      "loss": 1.401,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.019666666666666666,
+      "loss": 1.6172,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0196,
+      "loss": 1.4695,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019533333333333333,
+      "loss": 1.5137,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019466666666666667,
+      "loss": 1.5425,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0194,
+      "loss": 1.4272,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019333333333333334,
+      "loss": 1.3727,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019266666666666668,
+      "loss": 1.3114,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0192,
+      "loss": 1.4758,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019133333333333332,
+      "loss": 1.5219,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.01906666666666667,
+      "loss": 1.376,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.019,
+      "loss": 1.4257,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.018933333333333333,
+      "loss": 1.3474,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018866666666666667,
+      "loss": 1.2929,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0188,
+      "loss": 1.3208,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018733333333333334,
+      "loss": 1.3381,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018666666666666668,
+      "loss": 1.3644,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018600000000000002,
+      "loss": 1.2932,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018533333333333332,
+      "loss": 1.4092,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018466666666666666,
+      "loss": 1.3006,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0184,
+      "loss": 1.4572,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018333333333333333,
+      "loss": 1.2789,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.018266666666666667,
+      "loss": 1.4444,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0182,
+      "loss": 1.4511,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01813333333333333,
+      "loss": 1.3541,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.01806666666666667,
+      "loss": 1.3228,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.018000000000000002,
+      "loss": 1.3185,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017933333333333332,
+      "loss": 1.199,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017866666666666666,
+      "loss": 1.3417,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0178,
+      "loss": 1.4251,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017733333333333334,
+      "loss": 1.3574,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017666666666666667,
+      "loss": 1.2547,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0176,
+      "loss": 1.2651,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.017533333333333335,
+      "loss": 1.3414,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017466666666666665,
+      "loss": 1.3322,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0174,
+      "loss": 1.4147,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017333333333333336,
+      "loss": 1.2813,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017266666666666666,
+      "loss": 1.3687,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0172,
+      "loss": 1.5593,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017133333333333334,
+      "loss": 1.3073,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017066666666666667,
+      "loss": 1.2359,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.017,
+      "loss": 1.2474,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.016933333333333335,
+      "loss": 1.3874,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.01686666666666667,
+      "loss": 1.3203,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0168,
+      "loss": 1.2875,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016733333333333333,
+      "loss": 1.2767,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016666666666666666,
+      "loss": 1.3017,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0166,
+      "loss": 1.2321,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016533333333333334,
+      "loss": 1.1719,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016466666666666668,
+      "loss": 1.2552,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016399999999999998,
+      "loss": 1.3816,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01633333333333333,
+      "loss": 1.2956,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.01626666666666667,
+      "loss": 1.2061,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016200000000000003,
+      "loss": 1.2086,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016133333333333333,
+      "loss": 1.1633,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.016066666666666667,
+      "loss": 1.2638,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.016,
+      "loss": 1.3441,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015933333333333334,
+      "loss": 1.2924,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015866666666666668,
+      "loss": 1.1818,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0158,
+      "loss": 1.3918,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015733333333333332,
+      "loss": 1.2232,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015666666666666666,
+      "loss": 1.2472,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015600000000000001,
+      "loss": 1.2398,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015533333333333333,
+      "loss": 1.3649,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015466666666666667,
+      "loss": 1.2302,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0154,
+      "loss": 1.2053,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.015333333333333334,
+      "loss": 1.2974,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015266666666666666,
+      "loss": 1.3036,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0152,
+      "loss": 1.3162,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015133333333333334,
+      "loss": 1.2567,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015066666666666666,
+      "loss": 1.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.015,
+      "loss": 1.2692,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014933333333333335,
+      "loss": 1.1332,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014866666666666667,
+      "loss": 1.2949,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0148,
+      "loss": 1.2703,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014733333333333334,
+      "loss": 1.3891,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.014666666666666666,
+      "loss": 1.3594,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0146,
+      "loss": 1.166,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014533333333333334,
+      "loss": 1.3256,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014466666666666668,
+      "loss": 1.2669,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0144,
+      "loss": 1.241,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014333333333333333,
+      "loss": 1.2591,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014266666666666667,
+      "loss": 1.238,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014199999999999999,
+      "loss": 1.3583,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014133333333333333,
+      "loss": 1.164,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.014066666666666668,
+      "loss": 1.2367,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013999999999999999,
+      "loss": 1.1864,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013933333333333334,
+      "loss": 1.2259,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.013866666666666668,
+      "loss": 1.2129,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0138,
+      "loss": 1.2085,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013733333333333334,
+      "loss": 1.2316,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013666666666666667,
+      "loss": 1.2721,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013600000000000001,
+      "loss": 1.2428,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013533333333333333,
+      "loss": 1.2126,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013466666666666667,
+      "loss": 1.1583,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0134,
+      "loss": 1.2776,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013333333333333332,
+      "loss": 1.2317,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013266666666666666,
+      "loss": 1.2047,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.013200000000000002,
+      "loss": 1.202,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013133333333333332,
+      "loss": 1.1877,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013066666666666667,
+      "loss": 1.3071,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.013000000000000001,
+      "loss": 1.2976,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012933333333333333,
+      "loss": 1.4156,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012866666666666667,
+      "loss": 1.173,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0128,
+      "loss": 1.2263,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012733333333333334,
+      "loss": 1.3235,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012666666666666666,
+      "loss": 1.1923,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0126,
+      "loss": 1.2879,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012533333333333334,
+      "loss": 1.196,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.012466666666666666,
+      "loss": 1.2605,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0124,
+      "loss": 1.2303,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012333333333333335,
+      "loss": 1.2381,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012266666666666665,
+      "loss": 1.2709,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0122,
+      "loss": 1.2121,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012133333333333335,
+      "loss": 1.3713,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012066666666666668,
+      "loss": 1.2923,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.012,
+      "loss": 1.2947,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011933333333333334,
+      "loss": 1.1538,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011866666666666668,
+      "loss": 1.1312,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0118,
+      "loss": 1.1807,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.011733333333333333,
+      "loss": 1.2729,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011666666666666667,
+      "loss": 1.21,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0116,
+      "loss": 1.1986,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011533333333333333,
+      "loss": 1.2003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011466666666666667,
+      "loss": 1.1773,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011399999999999999,
+      "loss": 1.3241,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011333333333333332,
+      "loss": 1.2157,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011266666666666668,
+      "loss": 1.2549,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011200000000000002,
+      "loss": 1.3245,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011133333333333334,
+      "loss": 1.2109,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011066666666666667,
+      "loss": 1.1979,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.011000000000000001,
+      "loss": 1.2804,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010933333333333333,
+      "loss": 1.2655,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010866666666666667,
+      "loss": 1.1264,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0108,
+      "loss": 1.2949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010733333333333333,
+      "loss": 1.2038,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010666666666666666,
+      "loss": 1.2514,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0106,
+      "loss": 1.1692,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010533333333333332,
+      "loss": 1.1947,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010466666666666666,
+      "loss": 1.3294,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010400000000000001,
+      "loss": 1.2169,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010333333333333335,
+      "loss": 1.3113,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.010266666666666667,
+      "loss": 1.1322,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0102,
+      "loss": 1.4228,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010133333333333334,
+      "loss": 1.2384,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.010066666666666666,
+      "loss": 1.2107,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.01,
+      "loss": 1.2655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009933333333333334,
+      "loss": 1.2991,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009866666666666668,
+      "loss": 1.324,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0098,
+      "loss": 1.3443,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009733333333333333,
+      "loss": 1.1389,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.009666666666666667,
+      "loss": 1.2308,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0096,
+      "loss": 1.1847,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009533333333333335,
+      "loss": 1.3154,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009466666666666667,
+      "loss": 1.233,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0094,
+      "loss": 1.147,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009333333333333334,
+      "loss": 1.1824,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009266666666666666,
+      "loss": 1.2093,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0092,
+      "loss": 1.2111,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009133333333333334,
+      "loss": 1.0596,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009066666666666666,
+      "loss": 1.3025,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.009000000000000001,
+      "loss": 1.1726,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008933333333333333,
+      "loss": 1.2078,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.008866666666666667,
+      "loss": 1.2652,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0088,
+      "loss": 1.2033,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008733333333333333,
+      "loss": 1.1598,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008666666666666668,
+      "loss": 1.1904,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0086,
+      "loss": 1.242,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008533333333333334,
+      "loss": 1.3042,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008466666666666667,
+      "loss": 1.3653,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0084,
+      "loss": 1.1784,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008333333333333333,
+      "loss": 1.2306,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008266666666666667,
+      "loss": 1.2139,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008199999999999999,
+      "loss": 1.1891,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.008133333333333334,
+      "loss": 1.2619,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008066666666666666,
+      "loss": 1.0873,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.008,
+      "loss": 1.2537,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007933333333333334,
+      "loss": 1.2575,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007866666666666666,
+      "loss": 1.1043,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0078000000000000005,
+      "loss": 1.2063,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007733333333333333,
+      "loss": 1.1602,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007666666666666667,
+      "loss": 1.1474,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0076,
+      "loss": 1.1482,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.007533333333333333,
+      "loss": 1.2124,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074666666666666675,
+      "loss": 1.195,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0074,
+      "loss": 1.1426,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007333333333333333,
+      "loss": 1.2067,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007266666666666667,
+      "loss": 1.1649,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0072,
+      "loss": 1.0978,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0071333333333333335,
+      "loss": 1.2298,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.007066666666666666,
+      "loss": 1.195,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006999999999999999,
+      "loss": 1.2032,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006933333333333334,
+      "loss": 1.1134,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006866666666666667,
+      "loss": 1.2925,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0068000000000000005,
+      "loss": 1.1389,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006733333333333333,
+      "loss": 1.1952,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.006666666666666666,
+      "loss": 1.0672,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006600000000000001,
+      "loss": 1.2243,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006533333333333334,
+      "loss": 1.218,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006466666666666667,
+      "loss": 1.1988,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0064,
+      "loss": 1.1776,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006333333333333333,
+      "loss": 1.1506,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006266666666666667,
+      "loss": 1.0386,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0062,
+      "loss": 1.2601,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006133333333333333,
+      "loss": 1.058,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006066666666666667,
+      "loss": 1.2243,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.006,
+      "loss": 1.2445,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.005933333333333334,
+      "loss": 1.2567,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005866666666666667,
+      "loss": 1.1815,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0058,
+      "loss": 1.3061,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005733333333333333,
+      "loss": 1.3064,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005666666666666666,
+      "loss": 1.2739,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005600000000000001,
+      "loss": 1.2754,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005533333333333334,
+      "loss": 1.1678,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0054666666666666665,
+      "loss": 1.191,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0054,
+      "loss": 1.2127,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005333333333333333,
+      "loss": 1.3057,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.005266666666666666,
+      "loss": 1.1092,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005200000000000001,
+      "loss": 1.1615,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0051333333333333335,
+      "loss": 1.1769,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005066666666666667,
+      "loss": 1.1396,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.005,
+      "loss": 1.1663,
+      "step": 2250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004933333333333334,
+      "loss": 1.0931,
+      "step": 2260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004866666666666667,
+      "loss": 1.1143,
+      "step": 2270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0048,
+      "loss": 1.1891,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004733333333333333,
+      "loss": 1.1636,
+      "step": 2290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004666666666666667,
+      "loss": 1.1777,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0046,
+      "loss": 1.1933,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.004533333333333333,
+      "loss": 1.1606,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0044666666666666665,
+      "loss": 1.2233,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0044,
+      "loss": 1.2237,
+      "step": 2340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004333333333333334,
+      "loss": 1.2323,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004266666666666667,
+      "loss": 1.0837,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0042,
+      "loss": 1.1411,
+      "step": 2370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0041333333333333335,
+      "loss": 1.0772,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004066666666666667,
+      "loss": 1.1952,
+      "step": 2390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.004,
+      "loss": 1.1267,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.003933333333333333,
+      "loss": 1.2142,
+      "step": 2410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0038666666666666667,
+      "loss": 1.2375,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0038,
+      "loss": 1.1964,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0037333333333333337,
+      "loss": 1.1305,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0036666666666666666,
+      "loss": 1.1478,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0036,
+      "loss": 1.1542,
+      "step": 2460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003533333333333333,
+      "loss": 1.1168,
+      "step": 2470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003466666666666667,
+      "loss": 1.1185,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0034000000000000002,
+      "loss": 1.0708,
+      "step": 2490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003333333333333333,
+      "loss": 1.1487,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.003266666666666667,
+      "loss": 1.1814,
+      "step": 2510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0032,
+      "loss": 1.1583,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0031333333333333335,
+      "loss": 1.2117,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0030666666666666663,
+      "loss": 1.1998,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.003,
+      "loss": 1.2355,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0029333333333333334,
+      "loss": 1.2694,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0028666666666666667,
+      "loss": 1.1819,
+      "step": 2570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0028000000000000004,
+      "loss": 1.1469,
+      "step": 2580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0027333333333333333,
+      "loss": 1.1726,
+      "step": 2590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0026666666666666666,
+      "loss": 1.0332,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0026000000000000003,
+      "loss": 1.2277,
+      "step": 2610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0025333333333333336,
+      "loss": 1.1335,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.002466666666666667,
+      "loss": 1.0854,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0024,
+      "loss": 1.1181,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0023333333333333335,
+      "loss": 1.1004,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0022666666666666664,
+      "loss": 1.1311,
+      "step": 2660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0022,
+      "loss": 1.0965,
+      "step": 2670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0021333333333333334,
+      "loss": 1.2944,
+      "step": 2680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0020666666666666667,
+      "loss": 1.1267,
+      "step": 2690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.002,
+      "loss": 1.0006,
+      "step": 2700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019333333333333333,
+      "loss": 1.1332,
+      "step": 2710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0018666666666666669,
+      "loss": 1.133,
+      "step": 2720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0018,
+      "loss": 1.1696,
+      "step": 2730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017333333333333335,
+      "loss": 1.2066,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0016666666666666666,
+      "loss": 1.1698,
+      "step": 2750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0016,
+      "loss": 1.1531,
+      "step": 2760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0015333333333333332,
+      "loss": 1.333,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0014666666666666667,
+      "loss": 1.0968,
+      "step": 2780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0014000000000000002,
+      "loss": 1.2056,
+      "step": 2790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0013333333333333333,
+      "loss": 1.1381,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0012666666666666668,
+      "loss": 1.1355,
+      "step": 2810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0012,
+      "loss": 1.289,
+      "step": 2820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0011333333333333332,
+      "loss": 1.16,
+      "step": 2830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0010666666666666667,
+      "loss": 1.1275,
+      "step": 2840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001,
+      "loss": 1.0971,
+      "step": 2850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009333333333333334,
+      "loss": 1.0893,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008666666666666667,
+      "loss": 1.1557,
+      "step": 2870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008,
+      "loss": 1.1966,
+      "step": 2880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007333333333333333,
+      "loss": 1.2223,
+      "step": 2890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 1.1097,
+      "step": 2900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0006,
+      "loss": 1.1612,
+      "step": 2910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005333333333333334,
+      "loss": 1.2289,
+      "step": 2920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004666666666666667,
+      "loss": 1.222,
+      "step": 2930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004,
+      "loss": 1.1519,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 1.3762,
+      "step": 2950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 1.2506,
+      "step": 2960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002,
+      "loss": 1.2153,
+      "step": 2970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.1327,
+      "step": 2980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.2242,
+      "step": 2990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0,
+      "loss": 1.1963,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "step": 3000,
+      "total_flos": 1.73594313916416e+17,
+      "train_loss": 1.2429857851664226,
+      "train_runtime": 27386.6257,
+      "train_samples_per_second": 1.753,
+      "train_steps_per_second": 0.11
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 1,
+  "total_flos": 1.73594313916416e+17,
+  "trial_name": null,
+  "trial_params": null
+}