add readme update for showing how this model was created, what it looks like in the header, and that it takes ~80s to merge 5 models

README.md

@@ -1,3 +1,692 @@
 ---
 license: unknown
 ---
+
+## Merging models like lego blocks using ddare and ties
+
+If you want to fine-tune, here's an example Unsloth fine tuning guide for:
+[Alpaca + TinyLlama + RoPE Scaling full example.ipynb](https://colab.research.google.com/drive/1AZghoNBQaMDgWJpi4RbffGM1h6raLUj9?usp=sharing#scrollTo=LjY75GoYUCB8)
+
+## How do I generate my own model merges?
+
+The code below merges the following HuggingFace TinyLlama models:
+
+- TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
+- Doctor-Shotgun/TinyLlama-1.1B-32k-Instruct
+- Doctor-Shotgun/TinyLlama-1.1B-32k
+- Tensoic/TinyLlama-1.1B-3T-openhermes
+- Josephgflowers/TinyLlama-3T-Cinder-v1.3
+
+```python3
+import transformers
+import torch
+import logging
+from ddare.merge import merge_tensors
+from ddare.tensor import dare_ties_sparsification, relative_norm, divide_tensor_into_sets
+from ddare.util import get_device
+import re
+from typing import Dict, Tuple, List
+
+logging.basicConfig(level=logging.INFO)
+log = logging.getLogger(__name__)
+
+
+def get_models(
+    models: List[str],
+    trust_remote_code: bool,
+):
+    config = {
+        'torch_dtype': torch.float16,
+        'low_cpu_mem_usage': False,
+        'trust_remote_code': trust_remote_code,
+    }
+    loaded_models = []
+    num_models = len(models)
+    for midx, model_path in enumerate(models):
+        log.info(
+            f"loading model={midx}/{num_models} "
+            f"model={model_path} "
+        )
+        loaded_models.append(
+            transformers.AutoModelForCausalLM.from_pretrained(
+                model_path,
+                **config
+            )
+        )
+    return loaded_models
+
+
+def pm(
+    model,
+):
+    keys = model.state_dict().keys()
+    log.info(f"model keys={len(keys)}")
+    for i, k in enumerate(keys):
+        tensor = model.state_dict()[k]
+        log.info(
+            f"{i:3d} {k} shape={tensor.shape} "
+            f"type={tensor.dtype} dev={tensor.device} "
+            f"contig={tensor.is_contiguous()}")
+
+
+def run_text_test(
+    model,
+    model_path,
+    device: str,
+    question: str,
+):
+    base_model = model.to(device)
+    log.info(
+        f"loading model={model_path}"
+    )
+    tokenizer = transformers.AutoTokenizer.from_pretrained(
+        model_path,
+        torch_dtype=torch.float16)
+
+    inputs = tokenizer(
+        question,
+        return_tensors="pt"
+    ).to("cuda")
+    with torch.backends.cuda.sdp_kernel(
+        enable_flash=True,
+        enable_math=False,
+        enable_mem_efficient=False
+    ):
+        outputs = base_model.generate(**inputs)
+    log.info(tokenizer.decode(outputs[0], skip_special_tokens=True))
+    base_model = base_model.to("cpu")
+
+
+def get_layer_type(
+    key: str
+) -> Tuple[int, str]:
+    matcher = re.compile(r"model.layers.(\d+).(.+)")
+    m = matcher.match(key)
+    if m is None:
+        if "model.norm.weight" == key:
+            return -1, "norm"
+        if "model.embed_tokens.weight" == key:
+            return -1, "embed"
+        if "lm_head.weight" == key:
+            return -1, "head"
+        log.info(f"Unknown key {key}")
+        return -1, "unknown"
+    return int(m.group(1)), m.group(2)
+
+
+def merge_model_with_ties(
+    models: List[str],
+    model_dst: str,
+    trust_remote_code: bool = True
+):
+    models = get_models(
+        models=models,
+        trust_remote_code=trust_remote_code,
+    )
+    config = {}
+    result_dict: Dict[str, torch.Tensor] = {}
+    device = get_device()
+    keys = models[0].state_dict().keys()
+    num_keys = len(keys)
+    for k in keys:
+        block, layer_type = get_layer_type(k)
+        m0: torch.Tensor = models[0].state_dict()[k]
+        result = m0.clone()
+        sets = divide_tensor_into_sets(tensor=m0, n_sets=4)
+
+        # get the src layers to merge
+        m = [
+            models[1].state_dict()[k],
+            models[2].state_dict()[k],
+            models[3].state_dict()[k],
+        ]
+
+        # build a ratio
+        ratio = {
+            'to_q': 0.0,
+            'to_k': 0.0,
+            'to_v': 0.0,
+        }.get(layer_type, .5)
+
+        norm_ratio = 0.68
+        log.info(
+            f"model={k} {num_keys} shape={m0.shape} "
+            f"dtype={m0.dtype} {m0.device} "
+            f"raio={ratio} "
+            f"contig={m0.is_contiguous()} "
+            f"norm={norm_ratio}")
+
+        # for all tensors
+        for i, tensor in enumerate(m):
+            if layer_type == "to_k":
+                # Get to_q key
+                q_base = models[0].state_dict()[k.replace("to_k", "to_q")]
+                q_merge = models[i].state_dict()[k.replace("to_k", "to_q")]
+                scale = relative_norm(q_merge, q_base)
+                tensor = tensor.to(device) / scale
+                del scale
+            elif layer_type == "to_q":
+                scale = relative_norm(tensor, m0)
+                tensor = tensor.to(device) * scale
+                del scale
+            slice_mask = (
+                sets == i
+            ).bool()
+            new_tensor = dare_ties_sparsification(
+                model_a_param=m0,
+                model_b_param=tensor,
+                drop_rate=norm_ratio,
+                ties="sum",
+                rescale="off",
+                device=device,
+                **config)
+            new_tensor = merge_tensors("slerp", m0, tensor, ratio)
+            result = torch.where(slice_mask, new_tensor, result)
+            del new_tensor, slice_mask
+
+        result_dict[k] = result
+    # end of merge
+
+    log.info(
+        f"{config} - done merge saving to file: {model_dst}"
+    )
+    out_model = (
+        transformers.AutoModelForCausalLM.from_pretrained(
+            model_dst,
+            **config
+        )
+    )
+    out_model.state_dict = lambda: result_dict
+    out_model.save_pretrained(model_dst)
+
+
+def run():
+    log.info("start")
+    model_src = "TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T"
+    model_dst = "matlok/tinyllama-cinder-openhermes-32k"
+    config = {
+        'torch_dtype': torch.float16,
+        'low_cpu_mem_usage': False,
+        'trust_remote_code': True,
+    }
+    models = [
+        model_src,
+        "Doctor-Shotgun/TinyLlama-1.1B-32k-Instruct",
+        "Doctor-Shotgun/TinyLlama-1.1B-32k",
+        "Tensoic/TinyLlama-1.1B-3T-openhermes",
+        "Josephgflowers/TinyLlama-3T-Cinder-v1.3",
+    ]
+    merge_model_with_ties(
+        models=models,
+        model_dst=model_dst
+    )
+    log.info(f"loading newly-created file: {model_dst}")
+    model = transformers.AutoModelForCausalLM.from_pretrained(
+        model_dst,
+        **config
+    )
+    pm(model=model)
+    log.info(f"done loading new model: {model} file: {model_dst}")
+
+
+if __name__ == "__main__":
+    run()
+```
+
+
+### Logs
+
+Here's hte logs
+
+```
+Total VRAM 12282 MB, total RAM 85434 MB
+Set vram state to: NORMAL_VRAM
+Device: cuda:0 NVIDIA GeForce RTX 4070 Ti : native
+VAE dtype: torch.bfloat16
+INFO:__main__:start
+INFO:__main__:loading model=0/5 model=TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
+INFO:__main__:loading model=1/5 model=Doctor-Shotgun/TinyLlama-1.1B-32k-Instruct
+INFO:__main__:loading model=2/5 model=Doctor-Shotgun/TinyLlama-1.1B-32k
+INFO:__main__:loading model=3/5 model=Tensoic/TinyLlama-1.1B-3T-openhermes
+INFO:__main__:loading model=4/5 model=Josephgflowers/TinyLlama-3T-Cinder-v1.3
+INFO:__main__:model=model.embed_tokens.weight 201 shape=torch.Size([32000, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.0.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.1.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.2.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.3.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.4.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.5.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.6.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.7.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.8.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.9.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.10.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.11.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.12.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.13.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.14.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.15.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.16.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.17.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.18.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.19.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.20.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.self_attn.q_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.self_attn.k_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.self_attn.v_proj.weight 201 shape=torch.Size([256, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.self_attn.o_proj.weight 201 shape=torch.Size([2048, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.mlp.gate_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.mlp.up_proj.weight 201 shape=torch.Size([5632, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.mlp.down_proj.weight 201 shape=torch.Size([2048, 5632]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.input_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.layers.21.post_attention_layernorm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=model.norm.weight 201 shape=torch.Size([2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:model=lm_head.weight 201 shape=torch.Size([32000, 2048]) dtype=torch.float16 cpu raio=0.5 contig=True norm=0.68
+INFO:__main__:{} - done merge saving to file: matlok/tinyllama-cinder-openhermes-32k
+config.json: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 724/724 [00:00<00:00, 6.15MB/s]
+model.safetensors: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.20G/2.20G [00:57<00:00, 38.0MB/s]
+generation_config.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 133/133 [00:00<00:00, 1.82MB/s]
+INFO:__main__:loading newly-created file: matlok/tinyllama-cinder-openhermes-32k
+INFO:__main__:model keys=201
+INFO:__main__:  0 model.embed_tokens.weight shape=torch.Size([32000, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  1 model.layers.0.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  2 model.layers.0.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  3 model.layers.0.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  4 model.layers.0.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  5 model.layers.0.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  6 model.layers.0.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  7 model.layers.0.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  8 model.layers.0.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:  9 model.layers.0.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 10 model.layers.1.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 11 model.layers.1.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 12 model.layers.1.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 13 model.layers.1.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 14 model.layers.1.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 15 model.layers.1.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 16 model.layers.1.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 17 model.layers.1.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 18 model.layers.1.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 19 model.layers.2.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 20 model.layers.2.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 21 model.layers.2.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 22 model.layers.2.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 23 model.layers.2.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 24 model.layers.2.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 25 model.layers.2.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 26 model.layers.2.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 27 model.layers.2.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 28 model.layers.3.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 29 model.layers.3.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 30 model.layers.3.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 31 model.layers.3.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 32 model.layers.3.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 33 model.layers.3.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 34 model.layers.3.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 35 model.layers.3.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 36 model.layers.3.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 37 model.layers.4.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 38 model.layers.4.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 39 model.layers.4.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 40 model.layers.4.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 41 model.layers.4.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 42 model.layers.4.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 43 model.layers.4.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 44 model.layers.4.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 45 model.layers.4.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 46 model.layers.5.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 47 model.layers.5.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 48 model.layers.5.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 49 model.layers.5.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 50 model.layers.5.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 51 model.layers.5.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 52 model.layers.5.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 53 model.layers.5.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 54 model.layers.5.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 55 model.layers.6.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 56 model.layers.6.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 57 model.layers.6.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 58 model.layers.6.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 59 model.layers.6.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 60 model.layers.6.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 61 model.layers.6.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 62 model.layers.6.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 63 model.layers.6.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 64 model.layers.7.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 65 model.layers.7.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 66 model.layers.7.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 67 model.layers.7.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 68 model.layers.7.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 69 model.layers.7.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 70 model.layers.7.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 71 model.layers.7.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 72 model.layers.7.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 73 model.layers.8.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 74 model.layers.8.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 75 model.layers.8.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 76 model.layers.8.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 77 model.layers.8.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 78 model.layers.8.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 79 model.layers.8.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 80 model.layers.8.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 81 model.layers.8.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 82 model.layers.9.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 83 model.layers.9.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 84 model.layers.9.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 85 model.layers.9.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 86 model.layers.9.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 87 model.layers.9.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 88 model.layers.9.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 89 model.layers.9.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 90 model.layers.9.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 91 model.layers.10.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 92 model.layers.10.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 93 model.layers.10.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 94 model.layers.10.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 95 model.layers.10.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 96 model.layers.10.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 97 model.layers.10.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 98 model.layers.10.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__: 99 model.layers.10.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:100 model.layers.11.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:101 model.layers.11.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:102 model.layers.11.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:103 model.layers.11.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:104 model.layers.11.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:105 model.layers.11.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:106 model.layers.11.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:107 model.layers.11.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:108 model.layers.11.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:109 model.layers.12.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:110 model.layers.12.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:111 model.layers.12.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:112 model.layers.12.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:113 model.layers.12.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:114 model.layers.12.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:115 model.layers.12.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:116 model.layers.12.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:117 model.layers.12.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:118 model.layers.13.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:119 model.layers.13.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:120 model.layers.13.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:121 model.layers.13.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:122 model.layers.13.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:123 model.layers.13.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:124 model.layers.13.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:125 model.layers.13.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:126 model.layers.13.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:127 model.layers.14.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:128 model.layers.14.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:129 model.layers.14.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:130 model.layers.14.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:131 model.layers.14.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:132 model.layers.14.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:133 model.layers.14.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:134 model.layers.14.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:135 model.layers.14.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:136 model.layers.15.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:137 model.layers.15.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:138 model.layers.15.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:139 model.layers.15.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:140 model.layers.15.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:141 model.layers.15.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:142 model.layers.15.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:143 model.layers.15.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:144 model.layers.15.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:145 model.layers.16.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:146 model.layers.16.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:147 model.layers.16.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:148 model.layers.16.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:149 model.layers.16.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:150 model.layers.16.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:151 model.layers.16.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:152 model.layers.16.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:153 model.layers.16.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:154 model.layers.17.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:155 model.layers.17.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:156 model.layers.17.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:157 model.layers.17.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:158 model.layers.17.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:159 model.layers.17.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:160 model.layers.17.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:161 model.layers.17.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:162 model.layers.17.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:163 model.layers.18.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:164 model.layers.18.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:165 model.layers.18.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:166 model.layers.18.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:167 model.layers.18.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:168 model.layers.18.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:169 model.layers.18.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:170 model.layers.18.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:171 model.layers.18.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:172 model.layers.19.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:173 model.layers.19.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:174 model.layers.19.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:175 model.layers.19.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:176 model.layers.19.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:177 model.layers.19.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:178 model.layers.19.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:179 model.layers.19.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:180 model.layers.19.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:181 model.layers.20.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:182 model.layers.20.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:183 model.layers.20.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:184 model.layers.20.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:185 model.layers.20.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:186 model.layers.20.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:187 model.layers.20.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:188 model.layers.20.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:189 model.layers.20.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:190 model.layers.21.self_attn.q_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:191 model.layers.21.self_attn.k_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:192 model.layers.21.self_attn.v_proj.weight shape=torch.Size([256, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:193 model.layers.21.self_attn.o_proj.weight shape=torch.Size([2048, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:194 model.layers.21.mlp.gate_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:195 model.layers.21.mlp.up_proj.weight shape=torch.Size([5632, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:196 model.layers.21.mlp.down_proj.weight shape=torch.Size([2048, 5632]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:197 model.layers.21.input_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:198 model.layers.21.post_attention_layernorm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:199 model.norm.weight shape=torch.Size([2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:200 lm_head.weight shape=torch.Size([32000, 2048]) type=torch.float16 dev=cpu contig=True
+INFO:__main__:done loading new model: LlamaForCausalLM(
+  (model): LlamaModel(
+    (embed_tokens): Embedding(32000, 2048)
+    (layers): ModuleList(
+      (0-21): 22 x LlamaDecoderLayer(
+        (self_attn): LlamaSdpaAttention(
+          (q_proj): Linear(in_features=2048, out_features=2048, bias=False)
+          (k_proj): Linear(in_features=2048, out_features=256, bias=False)
+          (v_proj): Linear(in_features=2048, out_features=256, bias=False)
+          (o_proj): Linear(in_features=2048, out_features=2048, bias=False)
+          (rotary_emb): LlamaRotaryEmbedding()
+        )
+        (mlp): LlamaMLP(
+          (gate_proj): Linear(in_features=2048, out_features=5632, bias=False)
+          (up_proj): Linear(in_features=2048, out_features=5632, bias=False)
+          (down_proj): Linear(in_features=5632, out_features=2048, bias=False)
+          (act_fn): SiLU()
+        )
+        (input_layernorm): LlamaRMSNorm()
+        (post_attention_layernorm): LlamaRMSNorm()
+      )
+    )
+    (norm): LlamaRMSNorm()
+  )
+  (lm_head): Linear(in_features=2048, out_features=32000, bias=False)
+) file: matlok/tinyllama-cinder-openhermes-32k
+
+real	1m18.070s
+user	2m10.228s
+sys	0m14.040s
+```
+
+Note: code sample above was modified from [this very helpful GitHub gist](https://gist.github.com/maldevide/08829eada04ad9bd78e46c1a3787d42b)