Upload 43 files

config.json

@@ -0,0 +1,24 @@
+{
+  "_name_or_path": "huggyllama/llama-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "max_sequence_length": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.28.0",
+  "use_cache": true,
+  "vocab_size": 32000
+}

finetune.py

@@ -0,0 +1,283 @@
+import os
+import sys
+from typing import List
+
+import fire
+import torch
+import transformers
+from datasets import load_dataset
+
+"""
+Unused imports:
+import torch.nn as nn
+import bitsandbytes as bnb
+"""
+
+from peft import (
+    LoraConfig,
+    get_peft_model,
+    get_peft_model_state_dict,
+    prepare_model_for_int8_training,
+    set_peft_model_state_dict,
+)
+from transformers import LlamaForCausalLM, LlamaTokenizer
+
+from utils.prompter import Prompter
+
+
+def train(
+    # model/data params
+    base_model: str = "./hf_ckpt",  # the only required argument
+    data_path: str = "ayuan0324/ocean_only",
+    output_dir: str = "./lora-alpaca",
+    # training hyperparams
+    batch_size: int = 128,
+    micro_batch_size: int = 4,
+    num_epochs: int = 3,
+    learning_rate: float = 1e-4,
+    cutoff_len: int = 512,
+    val_set_size: int = 2000,
+    # lora hyperparams
+    lora_r: int = 8,
+    lora_alpha: int = 16,
+    lora_dropout: float = 0.05,
+    lora_target_modules: List[str] = [
+        "q_proj",
+        "v_proj",
+    ],
+    # llm hyperparams
+    train_on_inputs: bool = True,  # if False, masks out inputs in loss
+    add_eos_token: bool = False,
+    group_by_length: bool = False,  # faster, but produces an odd training loss curve
+    # wandb params
+    wandb_project: str = "",
+    wandb_run_name: str = "",
+    wandb_watch: str = "",  # options: false | gradients | all
+    wandb_log_model: str = "",  # options: false | true
+    resume_from_checkpoint: str = None,  # either training checkpoint or final adapter
+    prompt_template_name: str = "alpaca",  # The prompt template to use, will default to alpaca.
+):
+    if int(os.environ.get("LOCAL_RANK", 0)) == 0:
+        print(
+            f"Training Alpaca-LoRA model with params:\n"
+            f"base_model: {base_model}\n"
+            f"data_path: {data_path}\n"
+            f"output_dir: {output_dir}\n"
+            f"batch_size: {batch_size}\n"
+            f"micro_batch_size: {micro_batch_size}\n"
+            f"num_epochs: {num_epochs}\n"
+            f"learning_rate: {learning_rate}\n"
+            f"cutoff_len: {cutoff_len}\n"
+            f"val_set_size: {val_set_size}\n"
+            f"lora_r: {lora_r}\n"
+            f"lora_alpha: {lora_alpha}\n"
+            f"lora_dropout: {lora_dropout}\n"
+            f"lora_target_modules: {lora_target_modules}\n"
+            f"train_on_inputs: {train_on_inputs}\n"
+            f"add_eos_token: {add_eos_token}\n"
+            f"group_by_length: {group_by_length}\n"
+            f"wandb_project: {wandb_project}\n"
+            f"wandb_run_name: {wandb_run_name}\n"
+            f"wandb_watch: {wandb_watch}\n"
+            f"wandb_log_model: {wandb_log_model}\n"
+            f"resume_from_checkpoint: {resume_from_checkpoint or False}\n"
+            f"prompt template: {prompt_template_name}\n"
+        )
+    assert (
+        base_model
+    ), "Please specify a --base_model, e.g. --base_model='huggyllama/llama-7b'"
+    gradient_accumulation_steps = batch_size // micro_batch_size
+
+    prompter = Prompter(prompt_template_name)
+
+    device_map = "auto"
+    world_size = int(os.environ.get("WORLD_SIZE", 1))
+    ddp = world_size != 1
+    if ddp:
+        device_map = {"": int(os.environ.get("LOCAL_RANK") or 0)}
+        gradient_accumulation_steps = gradient_accumulation_steps // world_size
+
+    # Check if parameter passed or if set within environ
+    use_wandb = len(wandb_project) > 0 or (
+        "WANDB_PROJECT" in os.environ and len(os.environ["WANDB_PROJECT"]) > 0
+    )
+    # Only overwrite environ if wandb param passed
+    if len(wandb_project) > 0:
+        os.environ["WANDB_PROJECT"] = wandb_project
+    if len(wandb_watch) > 0:
+        os.environ["WANDB_WATCH"] = wandb_watch
+    if len(wandb_log_model) > 0:
+        os.environ["WANDB_LOG_MODEL"] = wandb_log_model
+
+    model = LlamaForCausalLM.from_pretrained(
+        base_model,
+        load_in_8bit=True,
+        torch_dtype=torch.float16,
+        device_map=device_map,
+    )
+
+    tokenizer = LlamaTokenizer.from_pretrained(base_model)
+
+    tokenizer.pad_token_id = (
+        0  # unk. we want this to be different from the eos token
+    )
+    tokenizer.padding_side = "left"  # Allow batched inference
+
+    def tokenize(prompt, add_eos_token=True):
+        # there's probably a way to do this with the tokenizer settings
+        # but again, gotta move fast
+        result = tokenizer(
+            prompt,
+            truncation=True,
+            max_length=cutoff_len,
+            padding=False,
+            return_tensors=None,
+        )
+        if (
+            result["input_ids"][-1] != tokenizer.eos_token_id
+            and len(result["input_ids"]) < cutoff_len
+            and add_eos_token
+        ):
+            result["input_ids"].append(tokenizer.eos_token_id)
+            result["attention_mask"].append(1)
+
+        result["labels"] = result["input_ids"].copy()
+
+        return result
+
+    def generate_and_tokenize_prompt(data_point):
+        full_prompt = prompter.generate_prompt(
+            data_point["instruction"],
+            data_point["input"],
+            data_point["output"],
+        )
+        tokenized_full_prompt = tokenize(full_prompt)
+        if not train_on_inputs:
+            user_prompt = prompter.generate_prompt(
+                data_point["instruction"], data_point["input"]
+            )
+            tokenized_user_prompt = tokenize(
+                user_prompt, add_eos_token=add_eos_token
+            )
+            user_prompt_len = len(tokenized_user_prompt["input_ids"])
+
+            if add_eos_token:
+                user_prompt_len -= 1
+
+            tokenized_full_prompt["labels"] = [
+                -100
+            ] * user_prompt_len + tokenized_full_prompt["labels"][
+                user_prompt_len:
+            ]  # could be sped up, probably
+        return tokenized_full_prompt
+
+    model = prepare_model_for_int8_training(model)
+
+    config = LoraConfig(
+        r=lora_r,
+        lora_alpha=lora_alpha,
+        target_modules=lora_target_modules,
+        lora_dropout=lora_dropout,
+        bias="none",
+        task_type="CAUSAL_LM",
+    )
+    model = get_peft_model(model, config)
+
+    if data_path.endswith(".json") or data_path.endswith(".jsonl"):
+        data = load_dataset("json", data_files=data_path)
+    else:
+        data = load_dataset(data_path)
+
+    if resume_from_checkpoint:
+        # Check the available weights and load them
+        checkpoint_name = os.path.join(
+            resume_from_checkpoint, "pytorch_model.bin"
+        )  # Full checkpoint
+        if not os.path.exists(checkpoint_name):
+            checkpoint_name = os.path.join(
+                resume_from_checkpoint, "adapter_model.bin"
+            )  # only LoRA model - LoRA config above has to fit
+            resume_from_checkpoint = (
+                False  # So the trainer won't try loading its state
+            )
+        # The two files above have a different name depending on how they were saved, but are actually the same.
+        if os.path.exists(checkpoint_name):
+            print(f"Restarting from {checkpoint_name}")
+            adapters_weights = torch.load(checkpoint_name)
+            set_peft_model_state_dict(model, adapters_weights)
+        else:
+            print(f"Checkpoint {checkpoint_name} not found")
+
+    model.print_trainable_parameters()  # Be more transparent about the % of trainable params.
+
+    if val_set_size > 0:
+        train_val = data["train"].train_test_split(
+            test_size=val_set_size, shuffle=True, seed=42
+        )
+        train_data = (
+            train_val["train"].shuffle().map(generate_and_tokenize_prompt)
+        )
+        val_data = (
+            train_val["test"].shuffle().map(generate_and_tokenize_prompt)
+        )
+    else:
+        train_data = data["train"].shuffle().map(generate_and_tokenize_prompt)
+        val_data = None
+
+    if not ddp and torch.cuda.device_count() > 1:
+        # keeps Trainer from trying its own DataParallelism when more than 1 gpu is available
+        model.is_parallelizable = True
+        model.model_parallel = True
+
+    trainer = transformers.Trainer(
+        model=model,
+        train_dataset=train_data,
+        eval_dataset=val_data,
+        args=transformers.TrainingArguments(
+            per_device_train_batch_size=micro_batch_size,
+            gradient_accumulation_steps=gradient_accumulation_steps,
+            warmup_steps=100,
+            num_train_epochs=num_epochs,
+            learning_rate=learning_rate,
+            fp16=True,
+            logging_steps=10,
+            optim="adamw_torch",
+            evaluation_strategy="steps" if val_set_size > 0 else "no",
+            save_strategy="steps",
+            eval_steps=200 if val_set_size > 0 else None,
+            save_steps=200,
+            output_dir=output_dir,
+            save_total_limit=3,
+            load_best_model_at_end=True if val_set_size > 0 else False,
+            ddp_find_unused_parameters=False if ddp else None,
+            group_by_length=group_by_length,
+            report_to="wandb" if use_wandb else None,
+            run_name=wandb_run_name if use_wandb else None,
+        ),
+        data_collator=transformers.DataCollatorForSeq2Seq(
+            tokenizer, pad_to_multiple_of=8, return_tensors="pt", padding=True
+        ),
+    )
+    model.config.use_cache = False
+
+    old_state_dict = model.state_dict
+    model.state_dict = (
+        lambda self, *_, **__: get_peft_model_state_dict(
+            self, old_state_dict()
+        )
+    ).__get__(model, type(model))
+
+    if torch.__version__ >= "2" and sys.platform != "win32":
+        model = torch.compile(model)
+
+    trainer.train(resume_from_checkpoint=resume_from_checkpoint)
+
+    model.save_pretrained(output_dir)
+
+    print(
+        "\n If there's a warning about missing keys above, please disregard :)"
+    )
+
+
+if __name__ == "__main__":
+    fire.Fire(train)

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.28.0"
+}

pytorch_model-00001-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90fc3335deafc1d0cd82699db4fafbfa22e5fc5c52696e05ad1930e6446c11fa
+size 396364479

pytorch_model-00002-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96ce010ee58182b60924d585564c8b7d8bfb5e84f0f1fafc0176b36b8f49e5c0
+size 371215393

pytorch_model-00003-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45e329af0cfdee73b8dbee049a3825a60305360629a80ca20dfd5ffbc1480235
+size 371215986

pytorch_model-00004-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce8a1e871b4c94010a63b9deb286e5ff7b02b0b3d83c3221ced303c66bf0d8b6
+size 371215986

pytorch_model-00005-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac9fff71fd60f865e2ca9572ea35fc9312c138a751db424a26b4f4f044504d2e
+size 371215986

pytorch_model-00006-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc56fa3fcf2556087de8deb1c6ef2f2e9740cc2c5005b8644d7167d4543883dc
+size 314575888

pytorch_model-00007-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c7486ef5ab3126ebce1f16eff9e50357409fe0e87a54bcd705bbbd47fbf508f
+size 314592882

pytorch_model-00008-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b93fead1f4c2dbd9566dfd20fce2cc19620238a6dbace3d7595e6f61354e0920
+size 314592882

pytorch_model-00009-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b25592d4d48d2cc5249b8df5602d95c090cf7269d5b829d330f7efccaaaf33d
+size 371215393

pytorch_model-00010-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a5b8de0bb5870624f744740cb6407a461fb46c2171ee68e96de631c43d21712
+size 371215986

pytorch_model-00011-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:776350ef96be9db69e0a3965ef154c47c49c812ff782d2d3e869746d3bfae493
+size 371215986

pytorch_model-00012-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66e6c471713784f1d67472d3c3e6adf0c4f54b074e787272ab630b07855f8b01
+size 371215986

pytorch_model-00013-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73d31049bc510f3a4dd1b99722ce9616083fde95dac580c500bb45f4fab6aeb5
+size 314575888

pytorch_model-00014-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cda02e4999cbb38ed962e2b6c886bc4a9d9e1f02bcc6eb250b46253d52e4179
+size 314592882

pytorch_model-00015-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb2f69ed87a5538da3ae130996cdea24fda0d44bf9f156f9aaec26ee8e595ad7
+size 314592882

pytorch_model-00016-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9f429e7148088e3300073e08d5c4604d5b1fb08f50979aeaa8a08aadfaada2b
+size 371215393

pytorch_model-00017-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c84d5406ba562b112c54ade3f7d23b9dd4f3e9b6b0b14381e91eb5d44d9fa579
+size 371215986

pytorch_model-00018-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35e64c30f73093d649aeb9128d6172dec4c5a943bb53bdb0b5cfbd6002b3d368
+size 371215986

pytorch_model-00019-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5cce86ebb65f3de5ae89c5ebc240d77edf64aae65db519d46b34a1212e076d9
+size 371215986

pytorch_model-00020-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4eb9082ff795c22ecb5f47833525e7a6f9b6cbcaa95d7d6d0c5dad89544aaa61
+size 314575888

pytorch_model-00021-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f2dace08ad6da9a5a2a8492e6e43e3534420a296b4d222f003fb3a19b4f601f
+size 314592882

pytorch_model-00022-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73716429a131afa2787a0f23d5ca68265279cf407bdd3965da989d55d86cba8e
+size 314592882

pytorch_model-00023-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e18ed7f91b6f34e1d17ac67f1deda6893f340c9b8de18d34895f95df39c964ae
+size 371215393

pytorch_model-00024-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f7a0695f9766fe5a501238b769c09a7ae65fdf9927c3709a2af258972d90873
+size 371215986

pytorch_model-00025-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:724cde5b763d3296daf126f1995fed529291cff8db7b21e19df71ffa7088480f
+size 371215986

pytorch_model-00026-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6cb5e18deac94f237ff77d4df32da3d910198a4a5e1e6dade3cd1992abade7d
+size 371215986

pytorch_model-00027-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef4761a5d3fec5ffa3833a222c24f984c0e594e82f9fcb547270e007cc18186
+size 314575888

pytorch_model-00028-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45816163d157fd97f91a92ff2d35475a26d054624fce208df76f4143dfbf46a5
+size 314592882

pytorch_model-00029-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb6d11c146f4e5d71bd553d86fba93799cf3bd9a4ed7f73b095719b33b16df2c
+size 314592882

pytorch_model-00030-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8e7bd467e6f8955591475bc755017a5da85c0f5990dce959921cfb95e0e781b
+size 371215393

pytorch_model-00031-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c396a1a4c0c475beced1220877157fe9eaace0f9837ff806cc716032baa41b73
+size 371215986

pytorch_model-00032-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79970ec11ba37086fe77fc441c44c5c7355958d720d5264f8c10ca4e31f801e2
+size 371215986

pytorch_model-00033-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1774033715bee46e83b1df9cbdf735eb863acbfaa8366506f5bde2284826ad2
+size 371215986

pytorch_model-00034-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d79092fd6b36c131b49d1af7c7a11148a3d2e5e751a188008cfa2156f1ad28d
+size 314575888

pytorch_model-00035-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce14b273fc3e7b48133a2dc19a67aeb4b5e833ae112ac48d71caae66ab7c34b9
+size 314592882

pytorch_model-00036-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10fbea1247f48991cc3e930843a3c568db3cd8416dcb52044c91200da80d123f
+size 314592882

pytorch_model-00037-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fc157263f2901696a39292c3bde890cded5696b3becbad0a7b89539bc1c4efc
+size 371215393

pytorch_model-00038-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38d4728a76a0b2f6c1db125d74cb8150fb240602588cd1e52d37de63d886808f
+size 304114977

pytorch_model-00039-of-00039.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3350fb71c2bc1e88d77d9685716fd2759e879137814a4120d98265a43f405919
+size 262144938

pytorch_model.bin.index.json

@@ -0,0 +1,330 @@
+{
+  "metadata": {
+    "total_size": 13476839424
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00039-of-00039.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00039.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00039.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00039.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00039.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00039.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00039.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00039.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00002-of-00039.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00039.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00013-of-00039.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00039.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00014-of-00039.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00039.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00015-of-00039.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00039.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00016-of-00039.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00017-of-00039.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00018-of-00039.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00018-of-00039.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00039.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00019-of-00039.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00020-of-00039.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00020-of-00039.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00039.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00021-of-00039.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00022-of-00039.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00022-of-00039.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00024-of-00039.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00023-of-00039.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00003-of-00039.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00039.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00024-of-00039.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00039.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00025-of-00039.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00026-of-00039.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00026-of-00039.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00027-of-00039.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00027-of-00039.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00028-of-00039.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00028-of-00039.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00039.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00029-of-00039.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00039.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00030-of-00039.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00031-of-00039.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00032-of-00039.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00032-of-00039.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00039.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00033-of-00039.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00034-of-00039.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00034-of-00039.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00039.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00035-of-00039.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00004-of-00039.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00039.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00005-of-00039.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00036-of-00039.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00036-of-00039.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00038-of-00039.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00037-of-00039.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00039.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00006-of-00039.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00039.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00007-of-00039.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00039.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00008-of-00039.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00039.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00009-of-00039.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00010-of-00039.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00039.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00011-of-00039.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00039.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00012-of-00039.bin",
+    "model.norm.weight": "pytorch_model-00038-of-00039.bin"
+  }
+}