Commit
•
8d146e9
1
Parent(s):
627bc8d
Spicyboros-3.1 + Yi-34B-Llama
Browse files- README.md +55 -0
- adapter_config.json +28 -0
- adapter_model.bin +3 -0
- added_tokens.json +4 -0
- config.json +40 -0
- configuration_yi.py +121 -0
- special_tokens_map.json +24 -0
- tokenizer.model +3 -0
- tokenizer_config.json +65 -0
README.md
ADDED
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- generated_from_trainer
|
4 |
+
model-index:
|
5 |
+
- name: airo-lora-out
|
6 |
+
results: []
|
7 |
+
---
|
8 |
+
|
9 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
10 |
+
should probably proofread and complete it, then remove this comment. -->
|
11 |
+
|
12 |
+
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
13 |
+
# airo-lora-out
|
14 |
+
|
15 |
+
This model was trained from scratch on the None dataset.
|
16 |
+
|
17 |
+
## Model description
|
18 |
+
|
19 |
+
More information needed
|
20 |
+
|
21 |
+
## Intended uses & limitations
|
22 |
+
|
23 |
+
More information needed
|
24 |
+
|
25 |
+
## Training and evaluation data
|
26 |
+
|
27 |
+
More information needed
|
28 |
+
|
29 |
+
## Training procedure
|
30 |
+
|
31 |
+
### Training hyperparameters
|
32 |
+
|
33 |
+
The following hyperparameters were used during training:
|
34 |
+
- learning_rate: 0.0001
|
35 |
+
- train_batch_size: 2
|
36 |
+
- eval_batch_size: 2
|
37 |
+
- seed: 42
|
38 |
+
- distributed_type: multi-GPU
|
39 |
+
- num_devices: 2
|
40 |
+
- total_train_batch_size: 4
|
41 |
+
- total_eval_batch_size: 4
|
42 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
43 |
+
- lr_scheduler_type: constant
|
44 |
+
- num_epochs: 1
|
45 |
+
|
46 |
+
### Training results
|
47 |
+
|
48 |
+
|
49 |
+
|
50 |
+
### Framework versions
|
51 |
+
|
52 |
+
- Transformers 4.34.1
|
53 |
+
- Pytorch 2.0.1+cu118
|
54 |
+
- Datasets 2.14.6
|
55 |
+
- Tokenizers 0.14.1
|
adapter_config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"alpha_pattern": {},
|
3 |
+
"auto_mapping": null,
|
4 |
+
"base_model_name_or_path": "/users/ubuntu/models/Yi-34B-Llama",
|
5 |
+
"bias": "none",
|
6 |
+
"fan_in_fan_out": null,
|
7 |
+
"inference_mode": true,
|
8 |
+
"init_lora_weights": true,
|
9 |
+
"layers_pattern": null,
|
10 |
+
"layers_to_transform": null,
|
11 |
+
"lora_alpha": 16,
|
12 |
+
"lora_dropout": 0.07,
|
13 |
+
"modules_to_save": null,
|
14 |
+
"peft_type": "LORA",
|
15 |
+
"r": 64,
|
16 |
+
"rank_pattern": {},
|
17 |
+
"revision": null,
|
18 |
+
"target_modules": [
|
19 |
+
"down_proj",
|
20 |
+
"up_proj",
|
21 |
+
"o_proj",
|
22 |
+
"gate_proj",
|
23 |
+
"k_proj",
|
24 |
+
"q_proj",
|
25 |
+
"v_proj"
|
26 |
+
],
|
27 |
+
"task_type": "CAUSAL_LM"
|
28 |
+
}
|
adapter_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:38e1a702e8e2ecb0690c1b3d373c8da2ec30cb70a5a41ef3080f783897807e92
|
3 |
+
size 1966383405
|
added_tokens.json
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"</s>": 64001,
|
3 |
+
"<s>": 64000
|
4 |
+
}
|
config.json
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/users/ubuntu/models/Yi-34B-Llama",
|
3 |
+
"architectures": [
|
4 |
+
"LlamaForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_bias": false,
|
7 |
+
"bos_token_id": 1,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 7168,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 20480,
|
13 |
+
"max_position_embeddings": 4096,
|
14 |
+
"model_type": "llama",
|
15 |
+
"num_attention_heads": 56,
|
16 |
+
"num_hidden_layers": 60,
|
17 |
+
"num_key_value_heads": 8,
|
18 |
+
"pad_token_id": 0,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"quantization_config": {
|
21 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
22 |
+
"bnb_4bit_quant_type": "nf4",
|
23 |
+
"bnb_4bit_use_double_quant": true,
|
24 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
25 |
+
"llm_int8_has_fp16_weight": false,
|
26 |
+
"llm_int8_skip_modules": null,
|
27 |
+
"llm_int8_threshold": 6.0,
|
28 |
+
"load_in_4bit": true,
|
29 |
+
"load_in_8bit": false,
|
30 |
+
"quant_method": "bitsandbytes"
|
31 |
+
},
|
32 |
+
"rms_norm_eps": 1e-05,
|
33 |
+
"rope_scaling": null,
|
34 |
+
"rope_theta": 5000000.0,
|
35 |
+
"tie_word_embeddings": false,
|
36 |
+
"torch_dtype": "bfloat16",
|
37 |
+
"transformers_version": "4.34.1",
|
38 |
+
"use_cache": false,
|
39 |
+
"vocab_size": 64002
|
40 |
+
}
|
configuration_yi.py
ADDED
@@ -0,0 +1,121 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
""" Yi model configuration"""
|
2 |
+
from transformers.configuration_utils import PretrainedConfig
|
3 |
+
from transformers.utils import logging
|
4 |
+
|
5 |
+
logger = logging.get_logger(__name__)
|
6 |
+
|
7 |
+
Yi_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
|
8 |
+
|
9 |
+
|
10 |
+
class YiConfig(PretrainedConfig):
|
11 |
+
r"""
|
12 |
+
This is the configuration class to store the configuration of a [`YiModel`]. It is used to instantiate an Yi
|
13 |
+
model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
|
14 |
+
defaults will yield a similar configuration to that of the Yi model.
|
15 |
+
|
16 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
17 |
+
documentation from [`PretrainedConfig`] for more information.
|
18 |
+
|
19 |
+
|
20 |
+
Args:
|
21 |
+
vocab_size (`int`, *optional*, defaults to 64000):
|
22 |
+
Vocabulary size of the Yi model. Defines the number of different tokens that can be represented by the
|
23 |
+
`inputs_ids` passed when calling [`YiModel`]
|
24 |
+
hidden_size (`int`, *optional*, defaults to 4096):
|
25 |
+
Dimension of the hidden representations.
|
26 |
+
intermediate_size (`int`, *optional*, defaults to 11008):
|
27 |
+
Dimension of the MLP representations.
|
28 |
+
num_hidden_layers (`int`, *optional*, defaults to 32):
|
29 |
+
Number of hidden layers in the Transformer encoder.
|
30 |
+
num_attention_heads (`int`, *optional*, defaults to 32):
|
31 |
+
Number of attention heads for each attention layer in the Transformer encoder.
|
32 |
+
num_key_value_heads (`int`, *optional*):
|
33 |
+
This is the number of key_value heads that should be used to implement Grouped Query Attention. If
|
34 |
+
`num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
|
35 |
+
`num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
|
36 |
+
converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
|
37 |
+
by meanpooling all the original heads within that group. For more details checkout [this
|
38 |
+
paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
|
39 |
+
`num_attention_heads`.
|
40 |
+
hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
|
41 |
+
The non-linear activation function (function or string) in the decoder.
|
42 |
+
max_position_embeddings (`int`, *optional*, defaults to 4096):
|
43 |
+
The maximum sequence length that this model might ever be used with. Typically set this to something large
|
44 |
+
just in case (e.g., 512 or 1024 or 2048 or 4096).
|
45 |
+
initializer_range (`float`, *optional*, defaults to 0.02):
|
46 |
+
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
|
47 |
+
rms_norm_eps (`float`, *optional*, defaults to 1e-5):
|
48 |
+
The epsilon used by the rms normalization layers.
|
49 |
+
use_cache (`bool`, *optional*, defaults to `True`):
|
50 |
+
Whether or not the model should return the last key/values attentions (not used by all models). Only
|
51 |
+
relevant if `config.is_decoder=True`.
|
52 |
+
tie_word_embeddings(`bool`, *optional*, defaults to `False`):
|
53 |
+
Whether to tie weight embeddings
|
54 |
+
output_attentions (`bool`, *optional*, defaults to `False`):
|
55 |
+
Whether or not to output attentions.
|
56 |
+
rope_theta (`float`, *optional*, defaults to 5000000.0):
|
57 |
+
The base period of the RoPE embeddings.
|
58 |
+
Example:
|
59 |
+
|
60 |
+
```python
|
61 |
+
>>> from transformers import YiModel, YiConfig
|
62 |
+
|
63 |
+
>>> # Initializing a Yi style configuration
|
64 |
+
>>> configuration = YiConfig()
|
65 |
+
|
66 |
+
>>> # Initializing a model from the Yi style configuration
|
67 |
+
>>> model = YiModel(configuration)
|
68 |
+
|
69 |
+
>>> # Accessing the model configuration
|
70 |
+
>>> configuration = model.config
|
71 |
+
```"""
|
72 |
+
model_type = "Yi"
|
73 |
+
keys_to_ignore_at_inference = ["past_key_values"]
|
74 |
+
|
75 |
+
def __init__(
|
76 |
+
self,
|
77 |
+
vocab_size=64000,
|
78 |
+
hidden_size=4096,
|
79 |
+
intermediate_size=11008,
|
80 |
+
num_hidden_layers=32,
|
81 |
+
num_attention_heads=32,
|
82 |
+
num_key_value_heads=4,
|
83 |
+
hidden_act="silu",
|
84 |
+
max_position_embeddings=4096,
|
85 |
+
initializer_range=0.02,
|
86 |
+
rms_norm_eps=1e-5,
|
87 |
+
use_cache=True,
|
88 |
+
pad_token_id=0,
|
89 |
+
bos_token_id=1,
|
90 |
+
eos_token_id=2,
|
91 |
+
tie_word_embeddings=False,
|
92 |
+
output_attentions=False,
|
93 |
+
rope_theta=5000000.0,
|
94 |
+
**kwargs,
|
95 |
+
):
|
96 |
+
self.vocab_size = vocab_size
|
97 |
+
self.max_position_embeddings = max_position_embeddings
|
98 |
+
self.hidden_size = hidden_size
|
99 |
+
self.intermediate_size = intermediate_size
|
100 |
+
self.num_hidden_layers = num_hidden_layers
|
101 |
+
self.num_attention_heads = num_attention_heads
|
102 |
+
|
103 |
+
# for backward compatibility
|
104 |
+
if num_key_value_heads is None:
|
105 |
+
num_key_value_heads = num_attention_heads
|
106 |
+
|
107 |
+
self.num_key_value_heads = num_key_value_heads
|
108 |
+
self.hidden_act = hidden_act
|
109 |
+
self.initializer_range = initializer_range
|
110 |
+
self.rms_norm_eps = rms_norm_eps
|
111 |
+
self.use_cache = use_cache
|
112 |
+
self.output_attentions = output_attentions
|
113 |
+
self.rope_theta = rope_theta
|
114 |
+
|
115 |
+
super().__init__(
|
116 |
+
pad_token_id=pad_token_id,
|
117 |
+
bos_token_id=bos_token_id,
|
118 |
+
eos_token_id=eos_token_id,
|
119 |
+
tie_word_embeddings=tie_word_embeddings,
|
120 |
+
**kwargs,
|
121 |
+
)
|
special_tokens_map.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<|startoftext|>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": false,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "<|endoftext|>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": "</s>",
|
17 |
+
"unk_token": {
|
18 |
+
"content": "<unk>",
|
19 |
+
"lstrip": false,
|
20 |
+
"normalized": false,
|
21 |
+
"rstrip": false,
|
22 |
+
"single_word": false
|
23 |
+
}
|
24 |
+
}
|
tokenizer.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:386c49cf943d71aa110361135338c50e38beeff0a66593480421f37b319e1a39
|
3 |
+
size 1033105
|
tokenizer_config.json
ADDED
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": false,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"added_tokens_decoder": {
|
5 |
+
"0": {
|
6 |
+
"content": "<unk>",
|
7 |
+
"lstrip": false,
|
8 |
+
"normalized": false,
|
9 |
+
"rstrip": false,
|
10 |
+
"single_word": false,
|
11 |
+
"special": true
|
12 |
+
},
|
13 |
+
"1": {
|
14 |
+
"content": "<|startoftext|>",
|
15 |
+
"lstrip": false,
|
16 |
+
"normalized": false,
|
17 |
+
"rstrip": false,
|
18 |
+
"single_word": false,
|
19 |
+
"special": true
|
20 |
+
},
|
21 |
+
"2": {
|
22 |
+
"content": "<|endoftext|>",
|
23 |
+
"lstrip": false,
|
24 |
+
"normalized": false,
|
25 |
+
"rstrip": false,
|
26 |
+
"single_word": false,
|
27 |
+
"special": true
|
28 |
+
},
|
29 |
+
"64000": {
|
30 |
+
"content": "<s>",
|
31 |
+
"lstrip": false,
|
32 |
+
"normalized": false,
|
33 |
+
"rstrip": false,
|
34 |
+
"single_word": false,
|
35 |
+
"special": true
|
36 |
+
},
|
37 |
+
"64001": {
|
38 |
+
"content": "</s>",
|
39 |
+
"lstrip": false,
|
40 |
+
"normalized": false,
|
41 |
+
"rstrip": false,
|
42 |
+
"single_word": false,
|
43 |
+
"special": true
|
44 |
+
}
|
45 |
+
},
|
46 |
+
"auto_map": {
|
47 |
+
"AutoTokenizer": [
|
48 |
+
"tokenization_yi.YiTokenizer",
|
49 |
+
null
|
50 |
+
]
|
51 |
+
},
|
52 |
+
"bos_token": "<|startoftext|>",
|
53 |
+
"clean_up_tokenization_spaces": false,
|
54 |
+
"eos_token": "<|endoftext|>",
|
55 |
+
"legacy": true,
|
56 |
+
"model_max_length": 4096,
|
57 |
+
"pad_token": "</s>",
|
58 |
+
"sp_model_kwargs": {},
|
59 |
+
"spaces_between_special_tokens": false,
|
60 |
+
"tokenizer_class": "LlamaTokenizer",
|
61 |
+
"trust_remote_code": false,
|
62 |
+
"unk_token": "<unk>",
|
63 |
+
"use_default_system_prompt": true,
|
64 |
+
"use_fast": true
|
65 |
+
}
|