a

README.md

@@ -0,0 +1,156 @@
+---
+license: apache-2.0
+datasets:
+- berkeley-nest/Nectar
+language:
+- en
+---
+# Starling-RM-34B
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+Starling-RM-34B is a reward model trained from [Yi-34B-Chat](https://huggingface.co/01-ai/Yi-34B-Chat). Following the method of training reward model in [the instructGPT paper](https://arxiv.org/abs/2203.02155), we remove the last layer of Yi-34B-Chat, 
+and concatenate a linear layer that outputs scalar for any pair of input prompt and response. We train the reward model with preference dataset [berkeley-nest/Nectar](https://huggingface.co/datasets/berkeley-nest/Nectar), 
+with the K-wise maximum likelihood estimator proposed in [this paper](https://arxiv.org/abs/2301.11270). The reward model outputs a scalar for any given prompt and response. A response that is more helpful and 
+less harmful will get the highest reward score. Note that since the preference dataset [berkeley-nest/Nectar](https://huggingface.co/datasets/berkeley-nest/Nectar) is based on GPT-4 preference, the reward model is likely to be biased
+towards GPT-4's own preference, including longer responses and certain response format. 
+
+For more detailed discussions, please check out our [blog post](https://starling.cs.berkeley.edu), and stay tuned for our upcoming code and paper!
+
+
+- **Developed by:** Banghua Zhu * , Evan Frick * , Tianhao Wu * , Hanlin Zhu and Jiantao Jiao.
+- **Model type:** Reward Model for RLHF
+- **License:** Apache-2.0 license under the condition that the dataset is not used to compete with OpenAI
+- **Finetuned from model:** [Yi-34B-Chat](https://huggingface.co/01-ai/Yi-34B-Chat)
+
+
+### Model Sources
+
+<!-- Provide the basic links for the model. -->
+
+- **Blog:** https://starling.cs.berkeley.edu/
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+Please use the following code for inference with the reward model.
+
+```python
+import os
+import torch
+from torch import nn
+from transformers import AutoTokenizer, LlamaPreTrainedModel,LlamaModel
+import math
+
+## Define the reward model function class
+
+class LlamaForSequenceClassification(LlamaPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.transformer = LlamaModel(config)
+        self.v_head = nn.Linear(config.hidden_size, 1, bias=False)
+        self.PAD_ID = 0
+        # Initialize weights and apply final processing
+        self.post_init()
+    
+    def get_device(self):
+        return self.transformer.device
+
+    def forward(
+          self,
+          input_ids=None,
+          past_key_values=None,
+          attention_mask=None,
+          position_ids=None,
+      ):
+          transformer_outputs = self.transformer(
+              input_ids,
+              attention_mask=attention_mask,
+              position_ids=position_ids,
+              output_hidden_states=True,
+          )
+          hidden_states = transformer_outputs.hidden_states[-1]
+          scores = []
+          rewards = self.v_head(hidden_states).squeeze(-1)
+          bs = int(input_ids.shape[0])
+          for i in range(bs):
+              c_inds = (input_ids[i] == self.PAD_ID).nonzero()
+              c_ind = c_inds[0].item() if len(c_inds) > 0 else input_ids.shape[1]
+              scores.append(rewards[i, c_ind - 1])
+          scores = torch.stack(scores)
+          return {"scores": scores}
+
+## Load the model and tokenizer
+
+reward_model = LlamaForSequenceClassification.from_pretrained("berkeley-nest/Starling-RM-34B",torch_dtype=torch.bfloat16)
+reward_tokenizer = AutoTokenizer.from_pretrained("01-ai/Yi-34B-Chat")
+reward_tokenizer.truncation_side = "left"
+
+reward_model.eval().requires_grad_(False)
+
+## Define the reward function
+reward_device = "cpu"
+reward_batch_size = 1
+
+def get_reward(samples):
+    """samples: List[str]"""
+    input_ids = []
+    attention_masks = []
+    encodings_dict = reward_tokenizer(
+        samples,
+        truncation=True,
+        max_length=2048,
+        padding="max_length",
+        return_tensors="pt",
+    ).to(reward_device)
+    input_ids = encodings_dict["input_ids"]
+    attention_masks = encodings_dict["attention_mask"]
+    mbs = reward_batch_size
+    out = []
+    for i in range(math.ceil(len(samples) / mbs)):
+        rewards = reward_model(input_ids=input_ids[i * mbs : (i + 1) * mbs], attention_mask=attention_masks[i * mbs : (i + 1) * mbs])
+        out.extend(rewards["scores"])
+    return torch.hstack(out)
+
+## Inference over test prompts with Yi chat template
+
+test_sample = ["<|im_start|>user\nHello!<|im_end|>\n<|im_start|>assistant\nHi, how can I help you?<|im_end|>"] 
+reward_for_test_sample = get_reward(test_sample)
+print(reward_for_test_sample)
+```
+
+## Metrics
+
+### Accuracy Metrics
+| Model                | Human Preference | Truth Preference | Safety Preference | Average      |
+|----------------------|------------------|------------------|-------------------|--------------|
+| Starling-RM-7B-alpha |            0.762 |            0.684 |             0.767 |        0.738 |
+| Starling-RM-34B      |        **0.807** |        **0.712** |         **0.782** |    **0.767** |
+|                      |                  |                  |                   |              |
+
+Starling-RM-34B improves over Starling-RM-7B-alpha in *every metric* we benchmarked.  Accuracy is measured as the rate in which the better response receives a higher score than the worse response.  In the case of more than 2 responses, the accuracy is the average of the accuracy for each possible pairing.
+
+The Human Preference benchmark is measured with [LMSYS's Chatbot Arena Conversations](https://huggingface.co/datasets/lmsys/chatbot_arena_conversations), where the winning model's response is considered the better response.
+
+The Truth Preference benchmark is measured with the [Truthful QA](https://huggingface.co/datasets/truthful_qa) dataset, where we expect reward(best_answer) >= reward(correct_answer) > reward(incorrect_answer).
+
+The Safety Preference benchmark is measured with [PKU's Safe-RLHF](https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF) dataset.  On datapoints with one safe response and one unsafe response, we expect the safe response to always have higher reward.
+
+For all benchmarks, we subsample down to <= 1000 prompts.
+
+## License
+The dataset, model and online demo is a research preview intended for non-commercial use only, subject to the [Yi License](https://huggingface.co/01-ai/Yi-34B-Chat/blob/main/LICENSE), OpenAI data generation [Terms of Use](https://openai.com/policies/terms-of-use), and [ShareGPT Privacy Practices](https://chrome.google.com/webstore/detail/sharegpt-share-your-chatg/daiacboceoaocpibfodeljbdfacokfjb). Please contact us if you find any potential violation.
+
+## Acknowledgment
+We would like to thank Nexusflow.ai for the assistance with compute which made these efforts possible.  We would like to thank the [LMSYS Organization](https://lmsys.org/) for their support of [lmsys-chat-1M](https://huggingface.co/datasets/lmsys/lmsys-chat-1m) dataset, evaluation and online demo. We would like to thank the open source community for their efforts in providing the datasets and base models we used to develope the project, including but not limited to Anthropic, Llama, Mistral, Hugging Face H4, LMSYS, OpenChat, OpenBMB, Flan and ShareGPT.
+
+## Citation
+```
+@misc{starling2023,
+    title = {Starling-7B: Improving LLM Helpfulness & Harmlessness with RLAIF},
+    url = {},
+    author = {Zhu, Banghua and Frick, Evan and Wu, Tianhao and Zhu, Hanlin and Jiao, Jiantao},
+    month = {November},
+    year = {2023}
+}
+```

config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "01-ai/Yi-34B-Chat",
+  "architectures": [
+    "LLMForSequenceRegression"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 7168,
+  "initializer_range": 0.02,
+  "intermediate_size": 20480,
+  "max_position_embeddings": 4096,
+  "model_type": "llama",
+  "num_attention_heads": 56,
+  "num_hidden_layers": 60,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 5000000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.36.2",
+  "use_cache": true,
+  "vocab_size": 64000
+}

model-00001-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eba6fa1ce7dc04dba78c9e07d2eb531cbc6153bd4c6ae19d170073b0de17dc93
+size 4793130952

model-00002-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7d14a86e98e04226e0c054358e95b775fce6ac821db50670b7eff4527a7ae2c
+size 4756459944

model-00003-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8f0dfde985253434e93e487c2b6c6b7da08699cea9ebf501994da2b3cd776c
+size 4991370392

model-00004-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20cb53e057ee9e17517eacc1ee6cb15f88715f8f131696bf93af8344ffe5b744
+size 4756459984

model-00005-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07a173669e36d91c15ff306c5f25a5656029b4f5d9a2e3d7a34b59abb8e82e95
+size 4756459976

model-00006-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57990ae417ee713f4ef6d98628481b1e95ac62d22001d6718cfe94f96061f93f
+size 4991370416

model-00007-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb0b69915d6c372a248c151aac308490f0c0381d9afbac3258aa7949d0d43126
+size 4756459984

model-00008-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9fbd5f339e4c14107c003465e39cc98ebc82906690b948cd28e1d6288575d04
+size 4756459976

model-00009-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:048482b3ca43830321075ae4d2438a087ae29470db0b2bab949540b2e4e48f06
+size 4991370416

model-00010-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f128281ca82fdf3dd18c9da87e0a63122084753bdb93f568d83eb69de732ffa
+size 4756459984

model-00011-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc23008f8b71647e7285656c1674f87ddabeaf2366a36f34bc2701f1365688c0
+size 4756459976

model-00012-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f5739ed330784ca4e8264766d7b43cdd4266a80b1cedb5c1df67a12a08660a4
+size 4991370416

model-00013-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:128cc9aa4d6f386ffe5ddf4e34c7bb03822a60d9cb19b0b31c85c36a8b7b0779
+size 4756459984

model-00014-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fff6e97920a37f27c49da76e2493dd356d9168121f8c22b1f8d0b78bfea7efc5
+size 4756459976

model-00015-of-00015.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab40380562961c4b386f31c266e61edb022d477463a0093ddff465397977e180
+size 293659192

model.safetensors.index.json

@@ -0,0 +1,550 @@
+{
+  "metadata": {
+    "total_size": 67860344832
+  },
+  "weight_map": {
+    "transformer.embed_tokens.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.input_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.mlp.down_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.mlp.gate_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.mlp.up_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.post_attention_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.self_attn.k_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.self_attn.o_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.self_attn.q_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.0.self_attn.v_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.input_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.mlp.down_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.mlp.gate_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.mlp.up_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.post_attention_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.self_attn.k_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.self_attn.o_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.self_attn.q_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.1.self_attn.v_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.10.input_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.mlp.down_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.mlp.gate_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.mlp.up_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.post_attention_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.self_attn.k_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.self_attn.o_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.self_attn.q_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.10.self_attn.v_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.input_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.mlp.down_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.mlp.gate_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.mlp.up_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.post_attention_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.self_attn.k_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.self_attn.o_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.self_attn.q_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.11.self_attn.v_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.12.input_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.12.mlp.down_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.12.mlp.gate_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.12.mlp.up_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.12.post_attention_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.12.self_attn.k_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.12.self_attn.o_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.12.self_attn.q_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.12.self_attn.v_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.13.input_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.mlp.down_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.mlp.gate_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.mlp.up_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.post_attention_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.self_attn.k_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.self_attn.o_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.self_attn.q_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.13.self_attn.v_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.input_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.mlp.down_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.mlp.gate_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.mlp.up_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.post_attention_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.self_attn.k_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.self_attn.o_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.self_attn.q_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.14.self_attn.v_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.input_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.mlp.down_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.mlp.gate_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.mlp.up_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.post_attention_layernorm.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.self_attn.k_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.self_attn.o_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.self_attn.q_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.15.self_attn.v_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.16.input_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.16.mlp.down_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.16.mlp.gate_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.16.mlp.up_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.16.post_attention_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.16.self_attn.k_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.16.self_attn.o_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.16.self_attn.q_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.16.self_attn.v_proj.weight": "model-00004-of-00015.safetensors",
+    "transformer.layers.17.input_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.mlp.down_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.mlp.gate_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.mlp.up_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.post_attention_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.self_attn.k_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.self_attn.o_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.self_attn.q_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.17.self_attn.v_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.input_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.mlp.down_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.mlp.gate_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.mlp.up_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.post_attention_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.self_attn.k_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.self_attn.o_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.self_attn.q_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.18.self_attn.v_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.input_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.mlp.down_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.mlp.gate_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.mlp.up_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.post_attention_layernorm.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.self_attn.k_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.self_attn.o_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.self_attn.q_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.19.self_attn.v_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.2.input_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.mlp.down_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.mlp.gate_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.mlp.up_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.post_attention_layernorm.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.self_attn.k_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.self_attn.o_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.self_attn.q_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.2.self_attn.v_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.20.input_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.20.mlp.down_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.20.mlp.gate_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.20.mlp.up_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.20.post_attention_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.20.self_attn.k_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.20.self_attn.o_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.20.self_attn.q_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.20.self_attn.v_proj.weight": "model-00005-of-00015.safetensors",
+    "transformer.layers.21.input_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.mlp.down_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.mlp.gate_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.mlp.up_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.post_attention_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.self_attn.k_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.self_attn.o_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.self_attn.q_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.21.self_attn.v_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.input_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.mlp.down_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.mlp.gate_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.mlp.up_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.post_attention_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.self_attn.k_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.self_attn.o_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.self_attn.q_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.22.self_attn.v_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.input_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.mlp.down_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.mlp.gate_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.mlp.up_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.post_attention_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.self_attn.k_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.self_attn.o_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.self_attn.q_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.23.self_attn.v_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.input_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.mlp.down_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.mlp.gate_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.mlp.up_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.post_attention_layernorm.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.self_attn.k_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.self_attn.o_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.self_attn.q_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.24.self_attn.v_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.25.input_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.25.mlp.down_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.25.mlp.gate_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.25.mlp.up_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.25.post_attention_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.25.self_attn.k_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.25.self_attn.o_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.25.self_attn.q_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.25.self_attn.v_proj.weight": "model-00006-of-00015.safetensors",
+    "transformer.layers.26.input_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.mlp.down_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.mlp.gate_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.mlp.up_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.post_attention_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.self_attn.k_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.self_attn.o_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.self_attn.q_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.26.self_attn.v_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.input_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.mlp.down_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.mlp.gate_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.mlp.up_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.post_attention_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.self_attn.k_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.self_attn.o_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.self_attn.q_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.27.self_attn.v_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.input_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.mlp.down_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.mlp.gate_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.mlp.up_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.post_attention_layernorm.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.self_attn.k_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.self_attn.o_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.self_attn.q_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.28.self_attn.v_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.29.input_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.29.mlp.down_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.29.mlp.gate_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.29.mlp.up_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.29.post_attention_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.29.self_attn.k_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.29.self_attn.o_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.29.self_attn.q_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.29.self_attn.v_proj.weight": "model-00007-of-00015.safetensors",
+    "transformer.layers.3.input_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.3.mlp.down_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.3.mlp.gate_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.3.mlp.up_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.3.post_attention_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.3.self_attn.k_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.3.self_attn.o_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.3.self_attn.q_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.3.self_attn.v_proj.weight": "model-00001-of-00015.safetensors",
+    "transformer.layers.30.input_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.mlp.down_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.mlp.gate_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.mlp.up_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.post_attention_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.self_attn.k_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.self_attn.o_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.self_attn.q_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.30.self_attn.v_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.input_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.mlp.down_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.mlp.gate_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.mlp.up_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.post_attention_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.self_attn.k_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.self_attn.o_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.self_attn.q_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.31.self_attn.v_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.input_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.mlp.down_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.mlp.gate_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.mlp.up_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.post_attention_layernorm.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.self_attn.k_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.self_attn.o_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.self_attn.q_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.32.self_attn.v_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.input_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.33.mlp.down_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.33.mlp.gate_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.mlp.up_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.post_attention_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.33.self_attn.k_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.self_attn.o_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.self_attn.q_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.33.self_attn.v_proj.weight": "model-00008-of-00015.safetensors",
+    "transformer.layers.34.input_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.mlp.down_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.mlp.gate_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.mlp.up_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.post_attention_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.self_attn.k_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.self_attn.o_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.self_attn.q_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.34.self_attn.v_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.input_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.mlp.down_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.mlp.gate_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.mlp.up_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.post_attention_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.self_attn.k_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.self_attn.o_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.self_attn.q_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.35.self_attn.v_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.input_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.mlp.down_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.mlp.gate_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.mlp.up_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.post_attention_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.self_attn.k_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.self_attn.o_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.self_attn.q_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.36.self_attn.v_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.input_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.mlp.down_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.mlp.gate_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.mlp.up_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.post_attention_layernorm.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.self_attn.k_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.self_attn.o_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.self_attn.q_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.37.self_attn.v_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.38.input_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.38.mlp.down_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.38.mlp.gate_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.38.mlp.up_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.38.post_attention_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.38.self_attn.k_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.38.self_attn.o_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.38.self_attn.q_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.38.self_attn.v_proj.weight": "model-00009-of-00015.safetensors",
+    "transformer.layers.39.input_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.mlp.down_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.mlp.gate_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.mlp.up_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.post_attention_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.self_attn.k_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.self_attn.o_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.self_attn.q_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.39.self_attn.v_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.4.input_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.mlp.down_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.mlp.gate_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.mlp.up_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.post_attention_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.self_attn.k_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.self_attn.o_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.self_attn.q_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.4.self_attn.v_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.40.input_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.mlp.down_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.mlp.gate_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.mlp.up_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.post_attention_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.self_attn.k_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.self_attn.o_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.self_attn.q_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.40.self_attn.v_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.input_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.mlp.down_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.mlp.gate_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.mlp.up_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.post_attention_layernorm.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.self_attn.k_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.self_attn.o_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.self_attn.q_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.41.self_attn.v_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.42.input_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.42.mlp.down_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.42.mlp.gate_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.42.mlp.up_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.42.post_attention_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.42.self_attn.k_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.42.self_attn.o_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.42.self_attn.q_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.42.self_attn.v_proj.weight": "model-00010-of-00015.safetensors",
+    "transformer.layers.43.input_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.mlp.down_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.mlp.gate_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.mlp.up_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.post_attention_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.self_attn.k_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.self_attn.o_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.self_attn.q_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.43.self_attn.v_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.input_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.mlp.down_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.mlp.gate_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.mlp.up_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.post_attention_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.self_attn.k_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.self_attn.o_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.self_attn.q_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.44.self_attn.v_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.input_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.mlp.down_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.mlp.gate_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.mlp.up_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.post_attention_layernorm.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.self_attn.k_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.self_attn.o_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.self_attn.q_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.45.self_attn.v_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.input_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.46.mlp.down_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.46.mlp.gate_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.mlp.up_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.post_attention_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.46.self_attn.k_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.self_attn.o_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.self_attn.q_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.46.self_attn.v_proj.weight": "model-00011-of-00015.safetensors",
+    "transformer.layers.47.input_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.mlp.down_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.mlp.gate_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.mlp.up_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.post_attention_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.self_attn.k_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.self_attn.o_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.self_attn.q_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.47.self_attn.v_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.input_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.mlp.down_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.mlp.gate_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.mlp.up_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.post_attention_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.self_attn.k_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.self_attn.o_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.self_attn.q_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.48.self_attn.v_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.input_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.mlp.down_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.mlp.gate_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.mlp.up_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.post_attention_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.self_attn.k_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.self_attn.o_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.self_attn.q_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.49.self_attn.v_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.5.input_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.mlp.down_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.mlp.gate_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.mlp.up_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.post_attention_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.self_attn.k_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.self_attn.o_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.self_attn.q_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.5.self_attn.v_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.50.input_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.mlp.down_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.mlp.gate_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.mlp.up_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.post_attention_layernorm.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.self_attn.k_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.self_attn.o_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.self_attn.q_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.50.self_attn.v_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.51.input_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.51.mlp.down_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.51.mlp.gate_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.51.mlp.up_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.51.post_attention_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.51.self_attn.k_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.51.self_attn.o_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.51.self_attn.q_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.51.self_attn.v_proj.weight": "model-00012-of-00015.safetensors",
+    "transformer.layers.52.input_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.mlp.down_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.mlp.gate_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.mlp.up_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.post_attention_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.self_attn.k_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.self_attn.o_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.self_attn.q_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.52.self_attn.v_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.input_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.mlp.down_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.mlp.gate_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.mlp.up_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.post_attention_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.self_attn.k_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.self_attn.o_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.self_attn.q_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.53.self_attn.v_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.input_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.mlp.down_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.mlp.gate_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.mlp.up_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.post_attention_layernorm.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.self_attn.k_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.self_attn.o_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.self_attn.q_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.54.self_attn.v_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.55.input_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.55.mlp.down_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.55.mlp.gate_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.55.mlp.up_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.55.post_attention_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.55.self_attn.k_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.55.self_attn.o_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.55.self_attn.q_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.55.self_attn.v_proj.weight": "model-00013-of-00015.safetensors",
+    "transformer.layers.56.input_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.mlp.down_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.mlp.gate_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.mlp.up_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.post_attention_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.self_attn.k_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.self_attn.o_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.self_attn.q_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.56.self_attn.v_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.input_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.mlp.down_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.mlp.gate_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.mlp.up_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.post_attention_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.self_attn.k_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.self_attn.o_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.self_attn.q_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.57.self_attn.v_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.input_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.mlp.down_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.mlp.gate_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.mlp.up_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.post_attention_layernorm.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.self_attn.k_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.self_attn.o_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.self_attn.q_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.58.self_attn.v_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.input_layernorm.weight": "model-00015-of-00015.safetensors",
+    "transformer.layers.59.mlp.down_proj.weight": "model-00015-of-00015.safetensors",
+    "transformer.layers.59.mlp.gate_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.mlp.up_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.post_attention_layernorm.weight": "model-00015-of-00015.safetensors",
+    "transformer.layers.59.self_attn.k_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.self_attn.o_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.self_attn.q_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.59.self_attn.v_proj.weight": "model-00014-of-00015.safetensors",
+    "transformer.layers.6.input_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.mlp.down_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.mlp.gate_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.mlp.up_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.post_attention_layernorm.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.self_attn.k_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.self_attn.o_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.self_attn.q_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.6.self_attn.v_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.input_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.7.mlp.down_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.7.mlp.gate_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.mlp.up_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.post_attention_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.7.self_attn.k_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.self_attn.o_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.self_attn.q_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.7.self_attn.v_proj.weight": "model-00002-of-00015.safetensors",
+    "transformer.layers.8.input_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.mlp.down_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.mlp.gate_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.mlp.up_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.post_attention_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.self_attn.k_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.self_attn.o_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.self_attn.q_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.8.self_attn.v_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.input_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.mlp.down_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.mlp.gate_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.mlp.up_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.post_attention_layernorm.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.self_attn.k_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.self_attn.o_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.self_attn.q_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.layers.9.self_attn.v_proj.weight": "model-00003-of-00015.safetensors",
+    "transformer.norm.weight": "model-00015-of-00015.safetensors",
+    "v_head.weight": "model-00015-of-00015.safetensors"
+  }
+}