Upload IQuestCoderForCausalLM

README.md

@@ -0,0 +1,199 @@
+---
+library_name: transformers
+tags: []
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## 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. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]

config.json

@@ -0,0 +1,49 @@
+{
+  "architectures": [
+    "IQuestCoderForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_iquestcoder.IQuestCoderConfig",
+    "AutoModel": "modeling_iquestcoder.IQuestCoderModel",
+    "AutoModelForCausalLM": "modeling_iquestcoder.IQuestCoderForCausalLM",
+    "AutoModelForQuestionAnswering": "modeling_iquestcoder.IQuestCoderForQuestionAnswering",
+    "AutoModelForSequenceClassification": "modeling_iquestcoder.IQuestCoderForSequenceClassification",
+    "AutoModelForTokenClassification": "modeling_iquestcoder.IQuestCoderForTokenClassification"
+  },
+  "bos_token_id": 1,
+  "clip_qkv": null,
+  "dtype": "bfloat16",
+  "eos_token_id": [
+    2,
+    75864,
+    75869
+  ],
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 5120,
+  "initializer_range": 0.02,
+  "intermediate_size": 27648,
+  "max_position_embeddings": 131072,
+  "max_window_layers": 0,
+  "mlp_bias": false,
+  "model_type": "iquestcoder",
+  "num_attention_heads": 40,
+  "num_hidden_layers": 80,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 500000.0,
+    "rope_type": "default"
+  },
+  "rope_theta": 500000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.4.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 76800
+}

configuration_iquestcoder.py

@@ -0,0 +1,182 @@
+"""IQuestCoder model configuration."""
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
+class IQuestCoderConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`IQuestCoderModel`]. It is used to instantiate
+    an IQuestCoder model according to the specified arguments, defining the model architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 76800):
+            Vocabulary size of the IQuestCoder model. Defines the number of different tokens that can be represented
+            by the `inputs_ids` passed when calling [`IQuestCoderModel`].
+        hidden_size (`int`, *optional*, defaults to 5120):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 27648):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 80):
+            Number of hidden layers in the Transformer decoder.
+        num_attention_heads (`int`, *optional*, defaults to 40):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        num_key_value_heads (`int`, *optional*, defaults to 8):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention (GQA).
+            If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA).
+            If `num_key_value_heads=1`, the model will use Multi Query Attention (MQA).
+        head_dim (`int`, *optional*, defaults to 128):
+            The dimension of each attention head. If not specified, defaults to `hidden_size // num_attention_heads`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 16384):
+            The maximum sequence length that this model might ever be used with.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-05):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models).
+        pad_token_id (`int`, *optional*):
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            End of stream token id.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings.
+        rope_theta (`float`, *optional*, defaults to 500000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Supports various RoPE scaling
+            types including "linear", "dynamic", "yarn", "longrope", etc.
+        attention_bias (`bool`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        mlp_bias (`bool`, *optional*, defaults to `False`):
+            Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers.
+        clip_qkv (`float`, *optional*):
+            If set, clip the query, key, and value tensors to this value. Borrowed from OLMo for training stability.
+        use_sliding_window (`bool`, *optional*, defaults to `False`):
+            Whether to use sliding window attention. Borrowed from Qwen2.
+        sliding_window (`int`, *optional*):
+            The sliding window size. Only effective when `use_sliding_window=True`.
+        max_window_layers (`int`, *optional*, defaults to 0):
+            The number of layers that don't use sliding window attention. Borrowed from Qwen2.
+
+    Example:
+        ```python
+        >>> from configuration_iquestcoder import IQuestCoderConfig
+        >>> from modeling_iquestcoder import IQuestCoderModel
+
+        >>> # Initializing a IQuestCoder configuration
+        >>> configuration = IQuestCoderConfig()
+
+        >>> # Initializing a model from the configuration
+        >>> model = IQuestCoderModel(configuration)
+
+        >>> # Accessing the model configuration
+        >>> configuration = model.config
+        ```
+    """
+
+    model_type = "iquestcoder"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=76800,
+        hidden_size=5120,
+        intermediate_size=27648,
+        num_hidden_layers=80,
+        num_attention_heads=40,
+        num_key_value_heads=8,
+        head_dim=128,
+        hidden_act="silu",
+        max_position_embeddings=16384,
+        initializer_range=0.02,
+        rms_norm_eps=1e-5,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        rope_theta=500000.0,
+        rope_scaling=None,
+        attention_bias=False,
+        attention_dropout=0.0,
+        mlp_bias=False,
+        # IQuestCoder specific (borrowed from OLMo)
+        clip_qkv=None,
+        # IQuestCoder specific (borrowed from Qwen2)
+        use_sliding_window=False,
+        sliding_window=None,
+        max_window_layers=0,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_key_value_heads = num_key_value_heads
+        self.head_dim = head_dim
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self.attention_bias = attention_bias
+        self.attention_dropout = attention_dropout
+        self.mlp_bias = mlp_bias
+        # IQuestCoder specific
+        self.clip_qkv = clip_qkv
+        self.use_sliding_window = use_sliding_window
+        self.sliding_window = sliding_window
+        self.max_window_layers = max_window_layers
+
+        # Validate rope_scaling configuration
+        self._rope_scaling_validation()
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    def _rope_scaling_validation(self):
+        """Validate the `rope_scaling` configuration."""
+        if self.rope_scaling is None:
+            return
+
+        if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) < 1:
+            raise ValueError(
+                "`rope_scaling` must be a dictionary with a minimum of one field, `type` or `rope_type`."
+            )
+        
+        rope_scaling_type = self.rope_scaling.get("type", None) or self.rope_scaling.get("rope_type", None)
+        if rope_scaling_type is None:
+            raise ValueError(
+                "`rope_scaling` must have a `type` or `rope_type` field."
+            )
+        
+        valid_rope_types = ["linear", "dynamic", "yarn", "longrope", "llama3"]
+        if rope_scaling_type not in valid_rope_types:
+            raise ValueError(
+                f"`rope_scaling`'s type field must be one of {valid_rope_types}, got {rope_scaling_type}"
+            )
+
+
+__all__ = ["IQuestCoderConfig"]
+

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": [
+    2,
+    75864,
+    75869
+  ],
+  "transformers_version": "5.4.0"
+}

model-00001-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5cb27125a0d2e92efb0da9bd5e5adfc07576584d9ae6e3c08862bade2028157
+size 9940683472

model-00002-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:beac6ac0449817fc06be0afcc463af40ee684c30fb31aa7f03a7e86c1752aff1
+size 9751972104

model-00003-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52aa28b75354215caf9e550743d1689490f2b59fa9f36f4d6a44271ddceb09d9
+size 9751972128

model-00004-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c54c907d10c4013be78863b5aa31452b0aab126c41008e235e0191487a8b6e83
+size 9751972128

model-00005-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:092cb4820ed355ec25699540e2fb113420a13663523a8d761a2c8f4fc0b04a21
+size 9751972128

model-00006-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b55bad18cb00658c91d56a9255f09007ed616205a9069baa512ff4ed806164cd
+size 9751972128

model-00007-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7ba083cc57b5266a6c2096da402a63a39f728e1559630ad9e9fbb5ac1e43f53
+size 9751972128

model-00008-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6ce6415e3c3c51c0ed5a4eb2600bd574b271f8acd853e2f5c2efd358f887a20
+size 9751972128

model-00009-of-00009.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10fd4c90568d7812df44075ca0dbaf63ae581c44f7e1e7dd4b7541b66f9b19cb
+size 1384163144

model.safetensors.index.json

@@ -0,0 +1,731 @@
+{
+  "metadata": {
+    "total_parameters": 39794283520,
+    "total_size": 79588567040
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00001-of-00009.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00009.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00009.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00009.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00009.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.42.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.43.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.44.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.45.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.46.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.mlp.up_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.47.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.48.input_layernorm.weight": "model-00005-of-00009.safetensors",
+    "model.layers.48.mlp.down_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.48.mlp.gate_proj.weight": "model-00005-of-00009.safetensors",
+    "model.layers.48.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.48.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.48.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.48.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.48.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.48.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.49.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.50.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.50.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.51.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.52.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.53.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.54.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.55.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.56.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.mlp.up_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.57.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.58.input_layernorm.weight": "model-00006-of-00009.safetensors",
+    "model.layers.58.mlp.down_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.58.mlp.gate_proj.weight": "model-00006-of-00009.safetensors",
+    "model.layers.58.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.58.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.58.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.58.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.58.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.58.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.59.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.60.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.60.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.61.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.62.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.63.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.64.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.65.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.66.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.mlp.up_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.67.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.68.input_layernorm.weight": "model-00007-of-00009.safetensors",
+    "model.layers.68.mlp.down_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.68.mlp.gate_proj.weight": "model-00007-of-00009.safetensors",
+    "model.layers.68.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.68.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.68.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.68.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.68.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.68.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.69.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.70.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.70.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.71.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.72.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.73.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.74.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.75.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.76.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.mlp.up_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.77.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.78.input_layernorm.weight": "model-00008-of-00009.safetensors",
+    "model.layers.78.mlp.down_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.78.mlp.gate_proj.weight": "model-00008-of-00009.safetensors",
+    "model.layers.78.mlp.up_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.78.post_attention_layernorm.weight": "model-00009-of-00009.safetensors",
+    "model.layers.78.self_attn.k_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.78.self_attn.o_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.78.self_attn.q_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.78.self_attn.v_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.input_layernorm.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.mlp.down_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.mlp.gate_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.mlp.up_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.post_attention_layernorm.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.self_attn.k_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.self_attn.o_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.self_attn.q_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.79.self_attn.v_proj.weight": "model-00009-of-00009.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00009.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00009.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
+    "model.norm.weight": "model-00009-of-00009.safetensors"
+  }
+}

modeling_iquestcoder.py

@@ -0,0 +1,1068 @@
+"""
+Modified MIT License 
+
+Software Copyright© 2025 IQuest Research
+
+Our only modification is that, if the Software (or any derivative works
+thereof) is used for any of your commercial products or services, you shall
+prominently display "IQuest Coder" on the user interface of such product or
+service.
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+"""
+
+from typing import Callable, List, Optional, Tuple, Union
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from transformers.activations import ACT2FN
+from transformers.cache_utils import Cache, DynamicCache, SlidingWindowCache, StaticCache
+from transformers.generation import GenerationMixin
+from transformers.modeling_attn_mask_utils import AttentionMaskConverter
+from transformers.modeling_flash_attention_utils import FlashAttentionKwargs
+from transformers.modeling_layers import GradientCheckpointingLayer
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    QuestionAnsweringModelOutput,
+    SequenceClassifierOutputWithPast,
+    TokenClassifierOutput,
+)
+from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
+from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
+from transformers.processing_utils import Unpack
+from transformers.utils import (
+    auto_docstring,
+    can_return_tuple,
+    is_torch_flex_attn_available,
+    logging,
+)
+
+from .configuration_iquestcoder import IQuestCoderConfig
+
+
+if is_torch_flex_attn_available():
+    from torch.nn.attention.flex_attention import BlockMask
+    from transformers.integrations.flex_attention import make_flex_block_causal_mask
+
+
+logger = logging.get_logger(__name__)
+
+
+# =============================================================================
+# Helper Functions
+# =============================================================================
+
+def rotate_half(x: torch.Tensor) -> torch.Tensor:
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def apply_rotary_pos_emb(
+    q: torch.Tensor,
+    k: torch.Tensor,
+    cos: torch.Tensor,
+    sin: torch.Tensor,
+    position_ids: Optional[torch.Tensor] = None,
+    unsqueeze_dim: int = 1,
+) -> Tuple[torch.Tensor, torch.Tensor]:
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q: The query tensor.
+        k: The key tensor.
+        cos: The cosine part of the rotary embedding.
+        sin: The sine part of the rotary embedding.
+        position_ids: Deprecated and unused.
+        unsqueeze_dim: The dimension along which to unsqueeze cos and sin.
+
+    Returns:
+        Tuple of query and key tensors rotated using the Rotary Position Embedding.
+    """
+    # Borrowed from OLMo: preserve original dtypes for numerical stability
+    q_dtype, k_dtype = q.dtype, k.dtype
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed.to(q_dtype), k_embed.to(k_dtype)
+
+
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    Expands key/value heads for Grouped Query Attention.
+    
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep).
+    The hidden states go from (batch, num_key_value_heads, seqlen, head_dim) to
+    (batch, num_attention_heads, seqlen, head_dim).
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+def eager_attention_forward(
+    module: nn.Module,
+    query: torch.Tensor,
+    key: torch.Tensor,
+    value: torch.Tensor,
+    attention_mask: Optional[torch.Tensor],
+    scaling: float,
+    dropout: float = 0.0,
+    **kwargs,
+) -> Tuple[torch.Tensor, torch.Tensor]:
+    """Standard eager attention implementation."""
+    key_states = repeat_kv(key, module.num_key_value_groups)
+    value_states = repeat_kv(value, module.num_key_value_groups)
+
+    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
+    if attention_mask is not None:
+        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+        attn_weights = attn_weights + causal_mask
+
+    attn_weights = F.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
+    attn_weights = F.dropout(attn_weights, p=dropout, training=module.training)
+    attn_output = torch.matmul(attn_weights, value_states)
+    attn_output = attn_output.transpose(1, 2).contiguous()
+
+    return attn_output, attn_weights
+
+
+# =============================================================================
+# Model Components
+# =============================================================================
+
+class IQuestCoderRMSNorm(nn.Module):
+    """Root Mean Square Layer Normalization.
+    
+    RMSNorm is computationally simpler than LayerNorm while achieving similar
+    performance. It normalizes the input by its RMS value.
+    """
+
+    def __init__(self, hidden_size: int, eps: float = 1e-6):
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)
+
+    def extra_repr(self) -> str:
+        return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}"
+
+
+class IQuestCoderRotaryEmbedding(nn.Module):
+    """Rotary Position Embedding (RoPE).
+    
+    Implements rotary positional embeddings as described in the RoFormer paper.
+    Supports various RoPE scaling methods for extended context lengths.
+    """
+
+    def __init__(self, config: IQuestCoderConfig, device=None):
+        super().__init__()
+        # BC: "rope_type" was originally "type"
+        if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
+            self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
+        else:
+            self.rope_type = "default"
+        self.max_seq_len_cached = config.max_position_embeddings
+        self.original_max_seq_len = config.max_position_embeddings
+
+        self.config = config
+        self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
+
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+        self.original_inv_freq = self.inv_freq
+
+    @torch.no_grad()
+    @dynamic_rope_update
+    def forward(self, x: torch.Tensor, position_ids: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        # Lazy recompute: accelerate meta-device init leaves inv_freq as zeros
+        if self.inv_freq is not None and self.inv_freq.numel() > 0 and (self.inv_freq == 0).all():
+            inv_freq, self.attention_scaling = self.rope_init_fn(self.config, None)
+            self.inv_freq = inv_freq.to(device=x.device, dtype=self.inv_freq.dtype)
+            self.original_inv_freq = self.inv_freq
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
+        position_ids_expanded = position_ids[:, None, :].float()
+
+        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
+        with torch.autocast(device_type=device_type, enabled=False):
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.cat((freqs, freqs), dim=-1)
+            cos = emb.cos() * self.attention_scaling
+            sin = emb.sin() * self.attention_scaling
+
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
+
+
+class IQuestCoderMLP(nn.Module):
+    """Feed-forward network with SwiGLU activation.
+    
+    Uses the gated linear unit variant with SiLU activation for improved
+    performance compared to standard FFN.
+    """
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.intermediate_size = config.intermediate_size
+        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=config.mlp_bias)
+        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=config.mlp_bias)
+        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=config.mlp_bias)
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        # SwiGLU: down_proj(act_fn(gate_proj(x)) * up_proj(x))
+        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+
+
+class IQuestCoderAttention(nn.Module):
+    """Multi-headed attention with support for Grouped Query Attention (GQA).
+    
+    Features:
+    - Grouped Query Attention for memory efficiency
+    - Optional QKV clipping for training stability (from OLMo)
+    - Optional sliding window attention (from Qwen2)
+    - Rotary Position Embeddings
+    """
+
+    def __init__(self, config: IQuestCoderConfig, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.head_dim = getattr(config, "head_dim", config.hidden_size // config.num_attention_heads)
+        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
+        self.scaling = self.head_dim ** -0.5
+        self.attention_dropout = config.attention_dropout
+        self.is_causal = True
+
+        # Projection layers
+        self.q_proj = nn.Linear(
+            config.hidden_size, config.num_attention_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.k_proj = nn.Linear(
+            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.v_proj = nn.Linear(
+            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.o_proj = nn.Linear(
+            config.num_attention_heads * self.head_dim, config.hidden_size, bias=config.attention_bias
+        )
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: Tuple[torch.Tensor, torch.Tensor],
+        attention_mask: Optional[torch.Tensor],
+        past_key_value: Optional[Cache] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        input_shape = hidden_states.shape[:-1]
+        hidden_shape = (*input_shape, -1, self.head_dim)
+
+        # Compute Q, K, V projections
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        # [OLMo Feature] Optional QKV clipping for training stability
+        if self.config.clip_qkv is not None:
+            query_states = query_states.clamp(min=-self.config.clip_qkv, max=self.config.clip_qkv)
+            key_states = key_states.clamp(min=-self.config.clip_qkv, max=self.config.clip_qkv)
+            value_states = value_states.clamp(min=-self.config.clip_qkv, max=self.config.clip_qkv)
+
+        # Reshape to (batch, heads, seq_len, head_dim)
+        query_states = query_states.view(hidden_shape).transpose(1, 2)
+        key_states = key_states.view(hidden_shape).transpose(1, 2)
+        value_states = value_states.view(hidden_shape).transpose(1, 2)
+
+        # Apply rotary position embeddings
+        cos, sin = position_embeddings
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
+
+        # Update KV cache if provided
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # [Qwen2 Feature] Sliding window attention
+        sliding_window = None
+        if (
+            self.config.use_sliding_window
+            and getattr(self.config, "sliding_window", None) is not None
+            and self.layer_idx >= self.config.max_window_layers
+        ):
+            sliding_window = self.config.sliding_window
+
+        # Select attention implementation
+        attention_interface: Callable = eager_attention_forward
+        if self.config._attn_implementation != "eager":
+            if self.config._attn_implementation == "sdpa" and kwargs.get("output_attentions", False):
+                logger.warning_once(
+                    "`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. "
+                    'Falling back to eager attention. This warning can be removed using the argument '
+                    '`attn_implementation="eager"` when loading the model.'
+                )
+            else:
+                attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
+
+        # Compute attention
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.attention_dropout,
+            scaling=self.scaling,
+            sliding_window=sliding_window,
+            **kwargs,
+        )
+
+        # Reshape and project output
+        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        return attn_output, attn_weights
+
+
+class IQuestCoderDecoderLayer(GradientCheckpointingLayer):
+    """Transformer decoder layer with pre-normalization.
+    
+    Architecture: Pre-RMSNorm -> Attention -> Residual -> Pre-RMSNorm -> MLP -> Residual
+    """
+
+    def __init__(self, config: IQuestCoderConfig, layer_idx: int):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = IQuestCoderAttention(config=config, layer_idx=layer_idx)
+        self.mlp = IQuestCoderMLP(config)
+        self.input_layernorm = IQuestCoderRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = IQuestCoderRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+        # Warn if sliding window is enabled but not properly supported
+        if config.use_sliding_window and config._attn_implementation != "flash_attention_2":
+            logger.warning_once(
+                f"Sliding Window Attention is enabled but not implemented for `{config._attn_implementation}`; "
+                "unexpected results may be encountered."
+            )
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+        cache_position: Optional[torch.LongTensor] = None,
+        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        # Pre-norm + Self Attention
+        residual = hidden_states
+        hidden_states = self.input_layernorm(hidden_states)
+
+        hidden_states, self_attn_weights = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+            cache_position=cache_position,
+            position_embeddings=position_embeddings,
+            **kwargs,
+        )
+        hidden_states = residual + hidden_states
+
+        # Pre-norm + MLP
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        return outputs
+
+
+# =============================================================================
+# Base Model
+# =============================================================================
+
+@auto_docstring
+class IQuestCoderPreTrainedModel(PreTrainedModel):
+    """Base class for IQuestCoder models."""
+
+    config_class = IQuestCoderConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["IQuestCoderDecoderLayer"]
+    _skip_keys_device_placement = ["past_key_values"]
+    _supports_flash_attn_2 = True
+    _supports_sdpa = True
+    _supports_flex_attn = True
+    _supports_cache_class = True
+    _supports_quantized_cache = True
+    _supports_static_cache = True
+    _supports_attention_backend = True
+
+    def _init_weights(self, module: nn.Module):
+        std = self.config.initializer_range
+        if isinstance(module, nn.Linear):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+        elif isinstance(module, IQuestCoderRMSNorm):
+            module.weight.data.fill_(1.0)
+
+
+@auto_docstring
+class IQuestCoderModel(IQuestCoderPreTrainedModel):
+    """
+    IQuestCoder Model outputting raw hidden-states without any specific head on top.
+    
+    This model is compatible with LLaMA weights while incorporating features from OLMo and Qwen2.
+    """
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
+        self.layers = nn.ModuleList(
+            [IQuestCoderDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self.norm = IQuestCoderRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.rotary_emb = IQuestCoderRotaryEmbedding(config=config)
+        self.gradient_checkpointing = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Embedding):
+        self.embed_tokens = value
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **flash_attn_kwargs: Unpack[FlashAttentionKwargs],
+    ) -> BaseModelOutputWithPast:
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
+
+        if self.gradient_checkpointing and self.training and use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`."
+            )
+            use_cache = False
+
+        if not isinstance(past_key_values, (type(None), Cache)):
+            raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.")
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        if use_cache and past_key_values is None:
+            past_key_values = DynamicCache()
+
+        if cache_position is None:
+            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+            cache_position = torch.arange(
+                past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
+            )
+
+        if position_ids is None:
+            position_ids = cache_position.unsqueeze(0)
+
+        causal_mask = self._update_causal_mask(
+            attention_mask, inputs_embeds, cache_position, past_key_values, output_attentions
+        )
+
+        hidden_states = inputs_embeds
+
+        # Create position embeddings to be shared across the decoder layers
+        position_embeddings = self.rotary_emb(hidden_states, position_ids)
+
+        # Decoder layers
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attns = () if output_attentions else None
+
+        for decoder_layer in self.layers[: self.config.num_hidden_layers]:
+            if output_hidden_states:
+                all_hidden_states += (hidden_states,)
+
+            layer_outputs = decoder_layer(
+                hidden_states,
+                attention_mask=causal_mask,
+                position_ids=position_ids,
+                past_key_value=past_key_values,
+                output_attentions=output_attentions,
+                use_cache=use_cache,
+                cache_position=cache_position,
+                position_embeddings=position_embeddings,
+                **flash_attn_kwargs,
+            )
+
+            hidden_states = layer_outputs[0]
+
+            if output_attentions:
+                all_self_attns += (layer_outputs[1],)
+
+        hidden_states = self.norm(hidden_states)
+
+        # Add hidden states from the last decoder layer
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=past_key_values if use_cache else None,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attns,
+        )
+
+    def _update_causal_mask(
+        self,
+        attention_mask: Union[torch.Tensor, "BlockMask"],
+        input_tensor: torch.Tensor,
+        cache_position: torch.Tensor,
+        past_key_values: Cache,
+        output_attentions: bool = False,
+    ):
+        if self.config._attn_implementation == "flash_attention_2":
+            if attention_mask is not None and past_key_values is not None:
+                is_padding_right = attention_mask[:, -1].sum().item() != input_tensor.size()[0]
+                if is_padding_right:
+                    raise ValueError(
+                        "You are attempting to perform batched generation with padding_side='right'. "
+                        "This may lead to unexpected behaviour for Flash Attention version of IQuestCoder. "
+                        "Make sure to call `tokenizer.padding_side = 'left'` before tokenizing the input."
+                    )
+            if attention_mask is not None and 0.0 in attention_mask:
+                return attention_mask
+            return None
+
+        if self.config._attn_implementation == "flex_attention":
+            if isinstance(attention_mask, torch.Tensor):
+                attention_mask = make_flex_block_causal_mask(attention_mask)
+            return attention_mask
+
+        past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+        using_static_cache = isinstance(past_key_values, StaticCache)
+        using_sliding_window_cache = isinstance(past_key_values, SlidingWindowCache)
+
+        if (
+            self.config._attn_implementation == "sdpa"
+            and not (using_static_cache or using_sliding_window_cache)
+            and not output_attentions
+        ):
+            if AttentionMaskConverter._ignore_causal_mask_sdpa(
+                attention_mask,
+                inputs_embeds=input_tensor,
+                past_key_values_length=past_seen_tokens,
+                sliding_window=self.config.sliding_window if self.config.use_sliding_window else None,
+                is_training=self.training,
+            ):
+                return None
+
+        dtype = input_tensor.dtype
+        min_dtype = torch.finfo(dtype).min
+        sequence_length = input_tensor.shape[1]
+
+        if using_sliding_window_cache or using_static_cache:
+            target_length = past_key_values.get_max_cache_shape()
+        else:
+            target_length = (
+                attention_mask.shape[-1]
+                if isinstance(attention_mask, torch.Tensor)
+                else past_seen_tokens + sequence_length + 1
+            )
+
+        causal_mask = self._prepare_4d_causal_attention_mask_with_cache_position(
+            attention_mask,
+            sequence_length=sequence_length,
+            target_length=target_length,
+            dtype=dtype,
+            cache_position=cache_position,
+            batch_size=input_tensor.shape[0],
+            config=self.config,
+            past_key_values=past_key_values,
+        )
+
+        if (
+            self.config._attn_implementation == "sdpa"
+            and attention_mask is not None
+            and attention_mask.device.type in ["cuda", "xpu", "npu"]
+            and not output_attentions
+        ):
+            causal_mask = AttentionMaskConverter._unmask_unattended(causal_mask, min_dtype)
+
+        return causal_mask
+
+    @staticmethod
+    def _prepare_4d_causal_attention_mask_with_cache_position(
+        attention_mask: torch.Tensor,
+        sequence_length: int,
+        target_length: int,
+        dtype: torch.dtype,
+        cache_position: torch.Tensor,
+        batch_size: int,
+        config: IQuestCoderConfig,
+        past_key_values: Cache,
+    ):
+        """Creates a causal 4D mask from a 2D mask, or returns the 4D mask if already provided."""
+        if attention_mask is not None and attention_mask.dim() == 4:
+            causal_mask = attention_mask
+        else:
+            min_dtype = torch.finfo(dtype).min
+            causal_mask = torch.full(
+                (sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=cache_position.device
+            )
+            diagonal_attend_mask = torch.arange(target_length, device=cache_position.device) > cache_position.reshape(
+                -1, 1
+            )
+            
+            # [Qwen2 Feature] Handle sliding window mask
+            if getattr(config, "use_sliding_window", False) and config.sliding_window is not None:
+                if not isinstance(past_key_values, SlidingWindowCache) or sequence_length > target_length:
+                    sliding_attend_mask = torch.arange(target_length, device=cache_position.device) <= (
+                        cache_position.reshape(-1, 1) - config.sliding_window
+                    )
+                    diagonal_attend_mask.bitwise_or_(sliding_attend_mask)
+            
+            causal_mask *= diagonal_attend_mask
+            causal_mask = causal_mask[None, None, :, :].expand(batch_size, 1, -1, -1)
+            
+            if attention_mask is not None:
+                causal_mask = causal_mask.clone()
+                if attention_mask.shape[-1] > target_length:
+                    attention_mask = attention_mask[:, :target_length]
+                mask_length = attention_mask.shape[-1]
+                padding_mask = causal_mask[:, :, :, :mask_length] + attention_mask[:, None, None, :].to(
+                    causal_mask.device
+                )
+                padding_mask = padding_mask == 0
+                causal_mask[:, :, :, :mask_length] = causal_mask[:, :, :, :mask_length].masked_fill(
+                    padding_mask, min_dtype
+                )
+
+        return causal_mask
+
+
+# =============================================================================
+# Model Heads
+# =============================================================================
+
+@auto_docstring
+class IQuestCoderForCausalLM(IQuestCoderPreTrainedModel, GenerationMixin):
+    """IQuestCoder Model with a language modeling head on top for causal LM."""
+
+    _tied_weights_keys = ["lm_head.weight"]
+    _tp_plan = {"lm_head": "colwise_rep"}
+    _pp_plan = {"lm_head": (["hidden_states"], ["logits"])}
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__(config)
+        self.model = IQuestCoderModel(config)
+        self.vocab_size = config.vocab_size
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Embedding):
+        self.model.embed_tokens = value
+
+    def get_output_embeddings(self) -> nn.Linear:
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings: nn.Linear):
+        self.lm_head = new_embeddings
+
+    def set_decoder(self, decoder: IQuestCoderModel):
+        self.model = decoder
+
+    def get_decoder(self) -> IQuestCoderModel:
+        return self.model
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        logits_to_keep: Union[int, torch.Tensor] = 0,
+        **kwargs
+    ) -> CausalLMOutputWithPast:
+        r"""
+        Args:
+            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Example:
+            ```python
+            >>> from transformers import AutoTokenizer
+            >>> from modeling_iquestcoder import IQuestCoderForCausalLM
+
+            >>> model = IQuestCoderForCausalLM.from_pretrained("path/to/IQuestCoder")
+            >>> tokenizer = AutoTokenizer.from_pretrained("path/to/IQuestCoder")
+
+            >>> prompt = "Hey, are you conscious? Can you talk to me?"
+            >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+            >>> # Generate
+            >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+            >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+            "Hey, are you conscious? Can you talk to me?\\nI'm not conscious, but I can talk to you."
+            ```
+        """
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+
+        # Decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs: BaseModelOutputWithPast = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            cache_position=cache_position,
+            **kwargs,
+        )
+
+        hidden_states = outputs.last_hidden_state
+        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
+        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
+        logits = self.lm_head(hidden_states[:, slice_indices, :])
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs)
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+@auto_docstring(
+    custom_intro="""
+    The IQuestCoder Model transformer with a sequence classification head on top (linear layer).
+
+    [`IQuestCoderForSequenceClassification`] uses the last token in order to do the classification, as other causal
+    models (e.g. GPT-2) do.
+
+    Since it does classification on the last token, it requires to know the position of the last token. If a
+    `pad_token_id` is defined in the configuration, it finds the last token that is not a padding token in each row.
+    If no `pad_token_id` is defined, it simply takes the last value in each row of the batch.
+    """
+)
+class IQuestCoderForSequenceClassification(IQuestCoderPreTrainedModel):
+    """IQuestCoder Model with a sequence classification head."""
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = IQuestCoderModel(config)
+        self.score = nn.Linear(config.hidden_size, self.num_labels, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Embedding):
+        self.model.embed_tokens = value
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+    ) -> SequenceClassifierOutputWithPast:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
+            If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        transformer_outputs: BaseModelOutputWithPast = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+        )
+        hidden_states = transformer_outputs.last_hidden_state
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size = input_ids.shape[0]
+        else:
+            batch_size = inputs_embeds.shape[0]
+
+        if self.config.pad_token_id is None and batch_size != 1:
+            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
+        if self.config.pad_token_id is None:
+            last_non_pad_token = -1
+        elif input_ids is not None:
+            non_pad_mask = (input_ids != self.config.pad_token_id).to(logits.device, torch.int32)
+            token_indices = torch.arange(input_ids.shape[-1], device=logits.device, dtype=torch.int32)
+            last_non_pad_token = (token_indices * non_pad_mask).argmax(-1)
+        else:
+            last_non_pad_token = -1
+            logger.warning_once(
+                f"{self.__class__.__name__} will not detect padding tokens in `inputs_embeds`. Results may be "
+                "unexpected if using padding tokens in conjunction with `inputs_embeds.`"
+            )
+
+        pooled_logits = logits[torch.arange(batch_size, device=logits.device), last_non_pad_token]
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(logits=logits, labels=labels, pooled_logits=pooled_logits, config=self.config)
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+
+@auto_docstring
+class IQuestCoderForTokenClassification(IQuestCoderPreTrainedModel):
+    """IQuestCoder Model with a token classification head."""
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = IQuestCoderModel(config)
+        if getattr(config, "classifier_dropout", None) is not None:
+            classifier_dropout = config.classifier_dropout
+        elif getattr(config, "hidden_dropout", None) is not None:
+            classifier_dropout = config.hidden_dropout
+        else:
+            classifier_dropout = 0.1
+        self.dropout = nn.Dropout(classifier_dropout)
+        self.score = nn.Linear(config.hidden_size, config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Embedding):
+        self.model.embed_tokens = value
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+    ) -> TokenClassifierOutput:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
+            If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        outputs: BaseModelOutputWithPast = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+        )
+        sequence_output = outputs.last_hidden_state
+        sequence_output = self.dropout(sequence_output)
+        logits = self.score(sequence_output)
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(logits, labels, self.config)
+
+        return TokenClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+@auto_docstring
+class IQuestCoderForQuestionAnswering(IQuestCoderPreTrainedModel):
+    """IQuestCoder Model with a span classification head for extractive question-answering."""
+
+    base_model_prefix = "transformer"
+
+    def __init__(self, config: IQuestCoderConfig):
+        super().__init__(config)
+        self.transformer = IQuestCoderModel(config)
+        self.qa_outputs = nn.Linear(config.hidden_size, 2)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.transformer.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Embedding):
+        self.transformer.embed_tokens = value
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        start_positions: Optional[torch.LongTensor] = None,
+        end_positions: Optional[torch.LongTensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        **kwargs,
+    ) -> QuestionAnsweringModelOutput:
+        outputs: BaseModelOutputWithPast = self.transformer(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+        )
+
+        sequence_output = outputs.last_hidden_state
+
+        logits = self.qa_outputs(sequence_output)
+        start_logits, end_logits = logits.split(1, dim=-1)
+        start_logits = start_logits.squeeze(-1).contiguous()
+        end_logits = end_logits.squeeze(-1).contiguous()
+
+        loss = None
+        if start_positions is not None and end_positions is not None:
+            loss = self.loss_function(start_logits, end_logits, start_positions, end_positions, **kwargs)
+
+        return QuestionAnsweringModelOutput(
+            loss=loss,
+            start_logits=start_logits,
+            end_logits=end_logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+__all__ = [
+    "IQuestCoderPreTrainedModel",
+    "IQuestCoderModel",
+    "IQuestCoderForCausalLM",
+    "IQuestCoderForSequenceClassification",
+    "IQuestCoderForTokenClassification",
+    "IQuestCoderForQuestionAnswering",
+]
+