Upload folder using huggingface_hub

README.md

@@ -1,3 +1,79 @@
 ---
-license: gpl-3.0
+library_name: transformers
+license: other
+base_model: microsoft/Phi-3.5-mini-instruct
+tags:
+- llama-factory
+- freeze
+- generated_from_trainer
+model-index:
+- name: microsoft/Phi-3.5-mini-instruct
+  results: []
 ---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# my DC sever
+https://discord.gg/yaTfFF6Ut2
+
+# 我正在計畫微調64K指令模型，請幫助我進行計畫
+Support me here if you're interested:
+# Ko-fi: https://ko-fi.com/ogodwin10
+
+# Phi-3.5-mini-instruct-24-9-29
+
+This model is a fine-tuned version of [microsoft/Phi-3.5-mini-instruct](https://huggingface.co/microsoft/Phi-3.5-mini-instruct) on the lmsys_chat dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.3194
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0001
+- train_batch_size: 2
+- eval_batch_size: 2
+- seed: 42
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 16
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: reduce_lr_on_plateau
+- training_steps: 1000
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 1.428         | 0.0160 | 100  | 1.3194          |
+| 1.2653        | 0.0320 | 200  | 1.3194          |
+| 1.3084        | 0.0480 | 300  | 1.3194          |
+| 1.3234        | 0.0641 | 400  | 1.3194          |
+| 1.4091        | 0.0801 | 500  | 1.3194          |
+| 1.2878        | 0.0961 | 600  | 1.3194          |
+| 1.2933        | 0.1121 | 700  | 1.3194          |
+| 1.3246        | 0.1281 | 800  | 1.3194          |
+| 1.2911        | 0.1441 | 900  | 1.3194          |
+| 1.4227        | 0.1602 | 1000 | 1.3194          |
+
+
+### Framework versions
+
+- Transformers 4.45.0
+- Pytorch 2.4.0+cu124
+- Datasets 2.19.1
+- Tokenizers 0.20.0

added_tokens.json

@@ -0,0 +1,13 @@
+{
+  "<|assistant|>": 32001,
+  "<|endoftext|>": 32000,
+  "<|end|>": 32007,
+  "<|placeholder1|>": 32002,
+  "<|placeholder2|>": 32003,
+  "<|placeholder3|>": 32004,
+  "<|placeholder4|>": 32005,
+  "<|placeholder5|>": 32008,
+  "<|placeholder6|>": 32009,
+  "<|system|>": 32006,
+  "<|user|>": 32010
+}

all_results.json

@@ -0,0 +1,12 @@
+{
+    "epoch": 0.16016016016016016,
+    "eval_loss": 1.3194454908370972,
+    "eval_runtime": 12.6122,
+    "eval_samples_per_second": 7.929,
+    "eval_steps_per_second": 3.964,
+    "total_flos": 2.9103944129622835e+17,
+    "train_loss": 1.3281120185852051,
+    "train_runtime": 3519.7932,
+    "train_samples_per_second": 4.546,
+    "train_steps_per_second": 0.284
+}

config.json

@@ -0,0 +1,139 @@
+{
+  "_name_or_path": "E:\\LLaMA-Factory\\saves\\Phi-3.5-mini-instruct\\sft",
+  "architectures": [
+    "Phi3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_phi3.Phi3Config",
+    "AutoModel": "modeling_phi3.Phi3ForCausalLM",
+    "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
+  },
+  "bos_token_id": 1,
+  "embd_pdrop": 0.0,
+  "eos_token_id": 32000,
+  "hidden_act": "silu",
+  "hidden_size": 3072,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "max_position_embeddings": 131072,
+  "model_type": "phi3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "original_max_position_embeddings": 4096,
+  "pad_token_id": 32000,
+  "resid_pdrop": 0.0,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "long_factor": [
+      1.0800000429153442,
+      1.1100000143051147,
+      1.1399999856948853,
+      1.340000033378601,
+      1.5899999141693115,
+      1.600000023841858,
+      1.6200000047683716,
+      2.620000123977661,
+      3.2300000190734863,
+      3.2300000190734863,
+      4.789999961853027,
+      7.400000095367432,
+      7.700000286102295,
+      9.09000015258789,
+      12.199999809265137,
+      17.670000076293945,
+      24.46000099182129,
+      28.57000160217285,
+      30.420001983642578,
+      30.840002059936523,
+      32.590003967285156,
+      32.93000411987305,
+      42.320003509521484,
+      44.96000289916992,
+      50.340003967285156,
+      50.45000457763672,
+      57.55000305175781,
+      57.93000411987305,
+      58.21000289916992,
+      60.1400032043457,
+      62.61000442504883,
+      62.62000274658203,
+      62.71000289916992,
+      63.1400032043457,
+      63.1400032043457,
+      63.77000427246094,
+      63.93000411987305,
+      63.96000289916992,
+      63.970001220703125,
+      64.02999877929688,
+      64.06999969482422,
+      64.08000183105469,
+      64.12000274658203,
+      64.41000366210938,
+      64.4800033569336,
+      64.51000213623047,
+      64.52999877929688,
+      64.83999633789062
+    ],
+    "short_factor": [
+      1.0,
+      1.0199999809265137,
+      1.0299999713897705,
+      1.0299999713897705,
+      1.0499999523162842,
+      1.0499999523162842,
+      1.0499999523162842,
+      1.0499999523162842,
+      1.0499999523162842,
+      1.0699999332427979,
+      1.0999999046325684,
+      1.1099998950958252,
+      1.1599998474121094,
+      1.1599998474121094,
+      1.1699998378753662,
+      1.2899998426437378,
+      1.339999794960022,
+      1.679999828338623,
+      1.7899998426437378,
+      1.8199998140335083,
+      1.8499997854232788,
+      1.8799997568130493,
+      1.9099997282028198,
+      1.9399996995925903,
+      1.9899996519088745,
+      2.0199997425079346,
+      2.0199997425079346,
+      2.0199997425079346,
+      2.0199997425079346,
+      2.0199997425079346,
+      2.0199997425079346,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0299997329711914,
+      2.0799996852874756,
+      2.0899996757507324,
+      2.189999580383301,
+      2.2199995517730713,
+      2.5899994373321533,
+      2.729999542236328,
+      2.749999523162842,
+      2.8399994373321533
+    ],
+    "type": "longrope"
+  },
+  "rope_theta": 10000.0,
+  "sliding_window": 262144,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.45.0",
+  "use_cache": false,
+  "vocab_size": 32064
+}

configuration_phi3.py

@@ -0,0 +1,227 @@
+# coding=utf-8
+# Copyright 2024 Microsoft and the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+""" Phi-3 model configuration"""
+
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+PHI3_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "microsoft/Phi-3-mini-4k-instruct": "https://huggingface.co/microsoft/Phi-3-mini-4k-instruct/resolve/main/config.json",
+    "microsoft/Phi-3-mini-128k-instruct": "https://huggingface.co/microsoft/Phi-3-mini-128k-instruct/resolve/main/config.json",
+}
+
+
+class Phi3Config(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`Phi3Model`]. It is used to instantiate a Phi-3
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the
+    [microsoft/Phi-3-mini-4k-instruct](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct).
+
+    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 32064):
+            Vocabulary size of the Phi-3 model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`Phi3Model`].
+        hidden_size (`int`, *optional*, defaults to 3072):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 8192):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer decoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        num_key_value_heads (`int`, *optional*):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. 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) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
+        resid_pdrop (`float`, *optional*, defaults to 0.0):
+            Dropout probability for mlp outputs.
+        embd_pdrop (`int`, *optional*, defaults to 0.0):
+            The dropout ratio for the embeddings.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio after computing the attention scores.
+        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 4096):
+            The maximum sequence length that this model might ever be used with.
+        original_max_position_embeddings (`int`, *optional*, defaults to 4096):
+            The maximum sequence length that this model was trained with. This is used to determine the size of the
+            original RoPE embeddings when using long scaling.
+        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 value used for the RMSNorm.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`. Whether to tie weight embeddings or not.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`dict`, *optional*):
+            The scaling strategy for the RoPE embeddings. If `None`, no scaling is applied. If a dictionary, it must
+            contain the following keys: `type`, `short_factor` and `long_factor`. The `type` must be `longrope` and 
+            the `short_factor` and `long_factor` must be lists of numbers with the same length as the hidden size 
+            divided by the number of attention heads divided by 2.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            The id of the "beginning-of-sequence" token.
+        eos_token_id (`int`, *optional*, defaults to 32000):
+            The id of the "end-of-sequence" token.
+        pad_token_id (`int`, *optional*, defaults to 32000):
+            The id of the padding token.
+        sliding_window (`int`, *optional*):
+            Sliding window attention window size. If `None`, no sliding window is applied.
+
+    Example:
+
+    ```python
+    >>> from transformers import Phi3Model, Phi3Config
+
+    >>> # Initializing a Phi-3 style configuration
+    >>> configuration = Phi3Config.from_pretrained("microsoft/Phi-3-mini-4k-instruct")
+
+    >>> # Initializing a model from the configuration
+    >>> model = Phi3Model(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "phi3"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=32064,
+        hidden_size=3072,
+        intermediate_size=8192,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=None,
+        resid_pdrop=0.0,
+        embd_pdrop=0.0,
+        attention_dropout=0.0,
+        hidden_act="silu",
+        max_position_embeddings=4096,
+        original_max_position_embeddings=4096,
+        initializer_range=0.02,
+        rms_norm_eps=1e-5,
+        use_cache=True,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        rope_scaling=None,
+        bos_token_id=1,
+        eos_token_id=32000,
+        pad_token_id=32000,
+        sliding_window=None,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.resid_pdrop = resid_pdrop
+        self.embd_pdrop = embd_pdrop
+        self.attention_dropout = attention_dropout
+        self.hidden_act = hidden_act
+        self.max_position_embeddings = max_position_embeddings
+        self.original_max_position_embeddings = original_max_position_embeddings
+        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._rope_scaling_adjustment()
+        self._rope_scaling_validation()
+        self.sliding_window = sliding_window
+
+        super().__init__(
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            pad_token_id=pad_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    def _rope_scaling_adjustment(self):
+        """
+        Adjust the `type` of the `rope_scaling` configuration for backward compatibility.
+        """
+        if self.rope_scaling is None:
+            return
+
+        rope_scaling_type = self.rope_scaling.get("type", None)
+
+        # For backward compatibility if previous version used "su" or "yarn"
+        if rope_scaling_type is not None and rope_scaling_type in ["su", "yarn"]:
+            self.rope_scaling["type"] = "longrope"
+
+    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) != 3:
+            raise ValueError(
+                "`rope_scaling` must be a dictionary with three fields, `type`, `short_factor` and `long_factor`, "
+                f"got {self.rope_scaling}"
+            )
+        rope_scaling_type = self.rope_scaling.get("type", None)
+        rope_scaling_short_factor = self.rope_scaling.get("short_factor", None)
+        rope_scaling_long_factor = self.rope_scaling.get("long_factor", None)
+        if rope_scaling_type is None or rope_scaling_type not in ["longrope"]:
+            raise ValueError(f"`rope_scaling`'s type field must be one of ['longrope'], got {rope_scaling_type}")
+        if not (
+            isinstance(rope_scaling_short_factor, list)
+            and all(isinstance(x, (int, float)) for x in rope_scaling_short_factor)
+        ):
+            raise ValueError(
+                f"`rope_scaling`'s short_factor field must be a list of numbers, got {rope_scaling_short_factor}"
+            )
+        if not len(rope_scaling_short_factor) == self.hidden_size // self.num_attention_heads // 2:
+            raise ValueError(
+                f"`rope_scaling`'s short_factor field must have length {self.hidden_size // self.num_attention_heads // 2}, got {len(rope_scaling_short_factor)}"
+            )
+        if not (
+            isinstance(rope_scaling_long_factor, list)
+            and all(isinstance(x, (int, float)) for x in rope_scaling_long_factor)
+        ):
+            raise ValueError(
+                f"`rope_scaling`'s long_factor field must be a list of numbers, got {rope_scaling_long_factor}"
+            )
+        if not len(rope_scaling_long_factor) == self.hidden_size // self.num_attention_heads // 2:
+            raise ValueError(
+                f"`rope_scaling`'s long_factor field must have length {self.hidden_size // self.num_attention_heads // 2}, got {len(rope_scaling_long_factor)}"
+            )

eval_results.json

@@ -0,0 +1,7 @@
+{
+    "epoch": 0.16016016016016016,
+    "eval_loss": 1.3194454908370972,
+    "eval_runtime": 12.6122,
+    "eval_samples_per_second": 7.929,
+    "eval_steps_per_second": 3.964
+}

generation_config.json

@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": [
+    32007,
+    32001,
+    32000
+  ],
+  "pad_token_id": 32000,
+  "transformers_version": "4.45.0"
+}

model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f490e5d0031feb5cafd1ee8e107ded943b98883f9f763f7e19c100798ea7c35
+size 4972489200

model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d1a02bd87851a25044e4816520a33bc41aaad1d1ebe48b866403b8a315e3576
+size 3122701904

model.safetensors.index.json

@@ -0,0 +1,202 @@
+{
+  "metadata": {
+    "total_size": 8095168512
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.gate_up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.qkv_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.qkv_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

modeling_phi3.py

@@ -0,0 +1,1570 @@
+# coding=utf-8
+# Copyright 2024 Microsoft and the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+""" PyTorch Phi-3 model."""
+
+import inspect
+import math
+import warnings
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
+
+from transformers.activations import ACT2FN
+from transformers.cache_utils import Cache, DynamicCache
+from transformers.modeling_attn_mask_utils import _prepare_4d_causal_attention_mask
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+    TokenClassifierOutput,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import (
+    add_code_sample_docstrings,
+    add_start_docstrings,
+    add_start_docstrings_to_model_forward,
+    is_flash_attn_2_available,
+    is_flash_attn_greater_or_equal_2_10,
+    logging,
+    replace_return_docstrings,
+)
+from .configuration_phi3 import Phi3Config
+
+
+logger = logging.get_logger(__name__)
+
+# Transformers scans dependencies in the modeling file, causing issues on conditional loading. The regex only ignores try/catch blocks, but not if statements
+# if is_flash_attn_2_available():
+_flash_supports_window_size = False
+try:
+    from flash_attn import flash_attn_func, flash_attn_varlen_func
+    from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input  # noqa
+
+    _flash_supports_window_size = "window_size" in list(inspect.signature(flash_attn_func).parameters)
+except ImportError as error:
+    logger.warning(
+        f"`flash-attention` package not found, consider installing for better performance: {error}."
+    )
+    if not _flash_supports_window_size:
+        logger.warning(
+            "Current `flash-attention` does not support `window_size`. Either upgrade or use `attn_implementation='eager'`."
+        )
+
+_CHECKPOINT_FOR_DOC = "microsoft/Phi-3-mini-4k-instruct"
+_CONFIG_FOR_DOC = "Phi3Config"
+
+PHI3_PRETRAINED_MODEL_ARCHIVE_LIST = [
+    "microsoft/Phi-3-mini-4k-instruct",
+    "microsoft/Phi-3-mini-128k-instruct",
+    # See all Phi-3 models at https://huggingface.co/models?filter=Phi-3
+]
+
+
+# Copied from transformers.models.llama.modeling_llama.LlamaRMSNorm with Llama->Phi3
+class Phi3RMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        Phi3RMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        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)
+
+
+# Copied from transformers.models.llama.modeling_llama._get_unpad_data
+def _get_unpad_data(attention_mask):
+    seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
+    indices = torch.nonzero(attention_mask.flatten(), as_tuple=False).flatten()
+    max_seqlen_in_batch = seqlens_in_batch.max().item()
+    cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.int32), (1, 0))
+    return (
+        indices,
+        cu_seqlens,
+        max_seqlen_in_batch,
+    )
+
+
+# Copied from transformers.models.gemma.modeling_gemma.GemmaRotaryEmbedding with gemma->phi3, Gemma->Phi3
+class Phi3RotaryEmbedding(nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
+        super().__init__()
+
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        self.register_buffer("inv_freq", None, persistent=False)
+
+    @torch.no_grad()
+    def forward(self, x, position_ids, seq_len=None):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        if self.inv_freq is None:
+            self.inv_freq = 1.0 / (
+                self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64, device=x.device).float() / self.dim)
+            )
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+        position_ids_expanded = position_ids[:, None, :].float()
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
+        device_type = x.device.type
+        device_type = device_type if isinstance(device_type, str) and 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()
+            sin = emb.sin()
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
+
+
+class Phi3LongRoPEScaledRotaryEmbedding(Phi3RotaryEmbedding):
+    def __init__(self, dim, config, device=None):
+        super().__init__(dim, config.max_position_embeddings, config.rope_theta, device)
+
+        self.short_factor = config.rope_scaling["short_factor"]
+        self.long_factor = config.rope_scaling["long_factor"]
+        self.original_max_position_embeddings = config.original_max_position_embeddings
+
+    @torch.no_grad()
+    def forward(self, x, position_ids, seq_len=None):
+        seq_len = seq_len or torch.max(position_ids) + 1
+        if seq_len > self.original_max_position_embeddings:
+            ext_factors = torch.tensor(self.long_factor, dtype=torch.float32, device=x.device)
+        else:
+            ext_factors = torch.tensor(self.short_factor, dtype=torch.float32, device=x.device)
+
+        inv_freq_shape = torch.arange(0, self.dim, 2, dtype=torch.int64, device=x.device).float() / self.dim
+        self.inv_freq = 1.0 / (ext_factors * self.base**inv_freq_shape)
+
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+        position_ids_expanded = position_ids[:, None, :].float()
+
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
+        device_type = x.device.type
+        device_type = device_type if isinstance(device_type, str) and 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)
+
+            scale = self.max_position_embeddings / self.original_max_position_embeddings
+            if scale <= 1.0:
+                scaling_factor = 1.0
+            else:
+                scaling_factor = math.sqrt(1 + math.log(scale) / math.log(self.original_max_position_embeddings))
+
+            cos = emb.cos() * scaling_factor
+            sin = emb.sin() * scaling_factor
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
+
+
+# Copied from transformers.models.llama.modeling_llama.rotate_half
+def rotate_half(x):
+    """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)
+
+
+# Copied from transformers.models.llama.modeling_llama.apply_rotary_pos_emb
+def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`, *optional*):
+            Deprecated and unused.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    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, k_embed
+
+
+class Phi3MLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+
+        self.config = config
+        self.gate_up_proj = nn.Linear(config.hidden_size, 2 * config.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(config.intermediate_size, config.hidden_size, bias=False)
+
+        self.activation_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, hidden_states: torch.FloatTensor) -> torch.FloatTensor:
+        up_states = self.gate_up_proj(hidden_states)
+
+        gate, up_states = up_states.chunk(2, dim=-1)
+        up_states = up_states * self.activation_fn(gate)
+
+        return self.down_proj(up_states)
+
+
+# Copied from transformers.models.llama.modeling_llama.repeat_kv with llama->phi
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    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)
+
+
+class Phi3Attention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: Phi3Config, layer_idx: Optional[int] = None):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        if layer_idx is None:
+            logger.warning_once(
+                f"Instantiating {self.__class__.__name__} without passing a `layer_idx` is not recommended and will "
+                "lead to errors during the forward call if caching is used. Please make sure to provide a `layer_idx` "
+                "when creating this class."
+            )
+
+        self.attention_dropout = config.attention_dropout
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+        self.original_max_position_embeddings = config.original_max_position_embeddings
+        self.rope_theta = config.rope_theta
+        self.rope_scaling = config.rope_scaling
+        self.is_causal = True
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+
+        op_size = self.num_heads * self.head_dim + 2 * (self.num_key_value_heads * self.head_dim)
+        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=False)
+        self.qkv_proj = nn.Linear(self.hidden_size, op_size, bias=False)
+        self._init_rope()
+
+    def _init_rope(self):
+        if self.rope_scaling is None:
+            self.rotary_emb = Phi3RotaryEmbedding(
+                self.head_dim,
+                max_position_embeddings=self.max_position_embeddings,
+                base=self.rope_theta,
+            )
+        else:
+            scaling_type = self.config.rope_scaling["type"]
+            if scaling_type == "longrope":
+                self.rotary_emb = Phi3LongRoPEScaledRotaryEmbedding(self.head_dim, self.config)
+            else:
+                raise ValueError(f"Unknown RoPE scaling type {scaling_type}")
+
+    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: bool = False,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        logger.warning_once("You are not running the flash-attention implementation, expect numerical differences.")
+
+        bsz, q_len, _ = hidden_states.size()
+
+        qkv = self.qkv_proj(hidden_states)
+        query_pos = self.num_heads * self.head_dim
+        query_states = qkv[..., :query_pos]
+        key_states = qkv[..., query_pos : query_pos + self.num_key_value_heads * self.head_dim]
+        value_states = qkv[..., query_pos + self.num_key_value_heads * self.head_dim :]
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            if self.layer_idx is None:
+                raise ValueError(
+                    f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} "
+                    "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class "
+                    "with a layer index."
+                )
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, position_ids, seq_len=kv_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # repeat k/v heads if n_kv_heads < n_heads
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(value_states.dtype)
+        attn_weights = nn.functional.dropout(attn_weights, p=self.attention_dropout, training=self.training)
+
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class Phi3FlashAttention2(Phi3Attention):
+    """
+    Phi-3 flash attention module. This module inherits from `Phi3Attention` as the weights of the module stays
+    untouched. The only required change would be on the forward pass where it needs to correctly call the public API of
+    flash attention and deal with padding tokens in case the input contains any of them.
+    """
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
+        # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignement, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
+        # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        # Phi3FlashAttention2 attention does not support output_attentions
+
+        if not _flash_supports_window_size:
+            logger.warning_once(
+                "The current flash attention version does not support sliding window attention. Please use `attn_implementation='eager'` or upgrade flash-attn library."
+            )
+            raise ValueError("The current flash attention version does not support sliding window attention.")
+
+        output_attentions = False
+
+        if "padding_mask" in kwargs:
+            warnings.warn(
+                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
+            )
+
+            # overwrite attention_mask with padding_mask
+            attention_mask = kwargs.pop("padding_mask")
+
+        bsz, q_len, _ = hidden_states.size()
+
+        qkv = self.qkv_proj(hidden_states)
+        query_pos = self.num_heads * self.head_dim
+        query_states = qkv[..., :query_pos]
+        key_states = qkv[..., query_pos : query_pos + self.num_key_value_heads * self.head_dim]
+        value_states = qkv[..., query_pos + self.num_key_value_heads * self.head_dim :]
+
+        # Flash attention requires the input to have the shape
+        # batch_size x seq_length x head_dim x hidden_dim
+        # therefore we just need to keep the original shape
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            if self.layer_idx is None:
+                raise ValueError(
+                    f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} "
+                    "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class "
+                    "with a layer index."
+                )
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+
+        # Because the input can be padded, the absolute sequence length depends on the max position id.
+        rotary_seq_len = max(kv_seq_len, position_ids[:, -1].max().item() + 1)
+        cos, sin = self.rotary_emb(value_states, position_ids, seq_len=rotary_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        use_sliding_windows = (
+            _flash_supports_window_size
+            and getattr(self.config, "sliding_window", None) is not None
+            and kv_seq_len > self.config.sliding_window
+        )
+
+        if past_key_value is not None:
+            # Activate slicing cache only if the config has a value `sliding_windows` attribute
+            cache_has_contents = past_key_value.get_seq_length(self.layer_idx) > 0
+            if (
+                getattr(self.config, "sliding_window", None) is not None
+                and kv_seq_len > self.config.sliding_window
+                and cache_has_contents
+            ):
+                slicing_tokens = 1 - self.config.sliding_window
+
+                past_key = past_key_value[self.layer_idx][0]
+                past_value = past_key_value[self.layer_idx][1]
+
+                past_key = past_key[:, :, slicing_tokens:, :].contiguous()
+                past_value = past_value[:, :, slicing_tokens:, :].contiguous()
+
+                if past_key.shape[-2] != self.config.sliding_window - 1:
+                    raise ValueError(
+                        f"past key must have a shape of (`batch_size, num_heads, self.config.sliding_window-1, head_dim`), got"
+                        f" {past_key.shape}"
+                    )
+
+                if attention_mask is not None:
+                    attention_mask = attention_mask[:, slicing_tokens:]
+                    attention_mask = torch.cat([attention_mask, torch.ones_like(attention_mask[:, -1:])], dim=-1)
+
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # repeat k/v heads if n_kv_heads < n_heads
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_dropout = self.attention_dropout if self.training else 0.0
+
+        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
+        # therefore the input hidden states gets silently casted in float32. Hence, we need
+        # cast them back in the correct dtype just to be sure everything works as expected.
+        # This might slowdown training & inference so it is recommended to not cast the LayerNorms
+        # in fp32.
+
+        if query_states.dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            # Handle the case where the model is quantized
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = self.config._pre_quantization_dtype
+            else:
+                target_dtype = self.qkv_proj.weight.dtype
+
+            logger.warning_once(
+                f"The input hidden states seems to be silently casted in float32, this might be related to"
+                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
+                f" {target_dtype}."
+            )
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        # Reashape to the expected shape for Flash Attention
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        attn_output = self._flash_attention_forward(
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            q_len,
+            dropout=attn_dropout,
+            use_sliding_windows=use_sliding_windows,
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+    # Copied from transformers.models.mistral.modeling_mistral.MistralFlashAttention2._flash_attention_forward
+    def _flash_attention_forward(
+        self,
+        query_states,
+        key_states,
+        value_states,
+        attention_mask,
+        query_length,
+        dropout=0.0,
+        softmax_scale=None,
+        use_sliding_windows=False,
+    ):
+        """
+        Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token
+        first unpad the input, then computes the attention scores and pad the final attention scores.
+
+        Args:
+            query_states (`torch.Tensor`):
+                Input query states to be passed to Flash Attention API
+            key_states (`torch.Tensor`):
+                Input key states to be passed to Flash Attention API
+            value_states (`torch.Tensor`):
+                Input value states to be passed to Flash Attention API
+            attention_mask (`torch.Tensor`):
+                The padding mask - corresponds to a tensor of size `(batch_size, seq_len)` where 0 stands for the
+                position of padding tokens and 1 for the position of non-padding tokens.
+            dropout (`float`):
+                Attention dropout
+            softmax_scale (`float`, *optional*):
+                The scaling of QK^T before applying softmax. Default to 1 / sqrt(head_dim)
+            use_sliding_windows (`bool`, *optional*):
+                Whether to activate sliding window attention.
+        """
+        if not self._flash_attn_uses_top_left_mask:
+            causal = self.is_causal
+        else:
+            # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in LlamaFlashAttention2 __init__.
+            causal = self.is_causal and query_length != 1
+
+        # Contains at least one padding token in the sequence
+        if attention_mask is not None:
+            batch_size = query_states.shape[0]
+            query_states, key_states, value_states, indices_q, cu_seq_lens, max_seq_lens = self._upad_input(
+                query_states, key_states, value_states, attention_mask, query_length
+            )
+
+            cu_seqlens_q, cu_seqlens_k = cu_seq_lens
+            max_seqlen_in_batch_q, max_seqlen_in_batch_k = max_seq_lens
+
+            if not use_sliding_windows:
+                attn_output_unpad = flash_attn_varlen_func(
+                    query_states,
+                    key_states,
+                    value_states,
+                    cu_seqlens_q=cu_seqlens_q,
+                    cu_seqlens_k=cu_seqlens_k,
+                    max_seqlen_q=max_seqlen_in_batch_q,
+                    max_seqlen_k=max_seqlen_in_batch_k,
+                    dropout_p=dropout,
+                    softmax_scale=softmax_scale,
+                    causal=causal,
+                )
+            else:
+                attn_output_unpad = flash_attn_varlen_func(
+                    query_states,
+                    key_states,
+                    value_states,
+                    cu_seqlens_q=cu_seqlens_q,
+                    cu_seqlens_k=cu_seqlens_k,
+                    max_seqlen_q=max_seqlen_in_batch_q,
+                    max_seqlen_k=max_seqlen_in_batch_k,
+                    dropout_p=dropout,
+                    softmax_scale=softmax_scale,
+                    causal=causal,
+                    window_size=(self.config.sliding_window, self.config.sliding_window),
+                )
+
+            attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length)
+        else:
+            if not use_sliding_windows:
+                attn_output = flash_attn_func(
+                    query_states,
+                    key_states,
+                    value_states,
+                    dropout,
+                    softmax_scale=softmax_scale,
+                    causal=causal,
+                )
+            else:
+                attn_output = flash_attn_func(
+                    query_states,
+                    key_states,
+                    value_states,
+                    dropout,
+                    softmax_scale=softmax_scale,
+                    causal=causal,
+                    window_size=(self.config.sliding_window, self.config.sliding_window),
+                )
+
+        return attn_output
+
+    # Copied from transformers.models.mistral.modeling_mistral.MistralFlashAttention2._upad_input
+    def _upad_input(self, query_layer, key_layer, value_layer, attention_mask, query_length):
+        batch_size, kv_seq_len, num_heads, head_dim = key_layer.shape
+
+        # On the first iteration we need to properly re-create the padding mask
+        # by slicing it on the proper place
+        if kv_seq_len != attention_mask.shape[-1]:
+            attention_mask_num_tokens = attention_mask.shape[-1]
+            attention_mask = attention_mask[:, attention_mask_num_tokens - kv_seq_len :]
+
+        indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask)
+
+        key_layer = index_first_axis(key_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k)
+        value_layer = index_first_axis(value_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k)
+
+        if query_length == kv_seq_len:
+            query_layer = index_first_axis(
+                query_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k
+            )
+            cu_seqlens_q = cu_seqlens_k
+            max_seqlen_in_batch_q = max_seqlen_in_batch_k
+            indices_q = indices_k
+        elif query_length == 1:
+            max_seqlen_in_batch_q = 1
+            cu_seqlens_q = torch.arange(
+                batch_size + 1, dtype=torch.int32, device=query_layer.device
+            )  # There is a memcpy here, that is very bad.
+            indices_q = cu_seqlens_q[:-1]
+            query_layer = query_layer.squeeze(1)
+        else:
+            # The -q_len: slice assumes left padding.
+            attention_mask = attention_mask[:, -query_length:]
+            query_layer, indices_q, cu_seqlens_q, max_seqlen_in_batch_q = unpad_input(query_layer, attention_mask)
+
+        return (
+            query_layer,
+            key_layer,
+            value_layer,
+            indices_q,
+            (cu_seqlens_q, cu_seqlens_k),
+            (max_seqlen_in_batch_q, max_seqlen_in_batch_k),
+        )
+
+
+# copied from transformers.models.llama.modeling_llama.LlamaSdpaAttention with Llama->Phi3
+# TODO @Arthur no longer copied from LLama after static cache
+class Phi3SdpaAttention(Phi3Attention):
+    """
+    Phi3 attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from
+    `Phi3Attention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to
+    SDPA API.
+    """
+
+    # Adapted from Phi3Attention.forward
+    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: bool = False,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if output_attentions:
+            # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
+            logger.warning_once(
+                "Phi3Model is using Phi3SdpaAttention, but `torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to the manual attention implementation, "
+                'but specifying the manual implementation will be required from Transformers version v5.0.0 onwards. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
+            )
+            return super().forward(
+                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,
+            )
+
+        bsz, q_len, _ = hidden_states.size()
+
+        qkv = self.qkv_proj(hidden_states)
+        query_pos = self.num_heads * self.head_dim
+        query_states = qkv[..., :query_pos]
+        key_states = qkv[..., query_pos : query_pos + self.num_key_value_heads * self.head_dim]
+        value_states = qkv[..., query_pos + self.num_key_value_heads * self.head_dim :]
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, position_ids, seq_len=kv_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+
+        # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
+        # Reference: https://github.com/pytorch/pytorch/issues/112577.
+        if query_states.device.type == "cuda" and attention_mask is not None:
+            query_states = query_states.contiguous()
+            key_states = key_states.contiguous()
+            value_states = value_states.contiguous()
+
+        attn_output = torch.nn.functional.scaled_dot_product_attention(
+            query_states,
+            key_states,
+            value_states,
+            attn_mask=attention_mask,
+            dropout_p=self.attention_dropout if self.training else 0.0,
+            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+            is_causal=self.is_causal and attention_mask is None and q_len > 1,
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.view(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        return attn_output, None, past_key_value
+
+
+PHI3_ATTENTION_CLASSES = {
+    "eager": Phi3Attention,
+    "flash_attention_2": Phi3FlashAttention2,
+    "sdpa": Phi3SdpaAttention,
+}
+
+
+class Phi3DecoderLayer(nn.Module):
+    def __init__(self, config: Phi3Config, layer_idx: int):
+        super().__init__()
+
+        self.config = config
+        self.self_attn = PHI3_ATTENTION_CLASSES[config._attn_implementation](config, layer_idx=layer_idx)
+
+        self.mlp = Phi3MLP(config)
+        self.input_layernorm = Phi3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+        self.resid_attn_dropout = nn.Dropout(config.resid_pdrop)
+        self.resid_mlp_dropout = nn.Dropout(config.resid_pdrop)
+        self.post_attention_layernorm = Phi3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+        **kwargs,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        if "padding_mask" in kwargs:
+            warnings.warn(
+                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
+            )
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`):
+                input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+            position_ids (`torch.LongTensor` of shape `({0})`, *optional*):
+                Indices of positions of each input sequence tokens in the position embeddings. Selected in the range
+                `[0, config.n_positions - 1]`. [What are position IDs?](../glossary#position-ids)
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            use_cache (`bool`, *optional*):
+                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
+                (see `past_key_values`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+        """
+
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+
+        # Self Attention
+        attn_outputs, self_attn_weights, present_key_value = 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,
+        )
+
+        hidden_states = residual + self.resid_attn_dropout(attn_outputs)
+
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + self.resid_mlp_dropout(hidden_states)
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+PHI3_START_DOCSTRING = r"""
+    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
+    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
+    etc.)
+
+    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
+    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
+    and behavior.
+
+    Parameters:
+        config ([`Phi3Config`]):
+            Model configuration class with all the parameters of the model. Initializing with a config file does not
+            load the weights associated with the model, only the configuration. Check out the
+            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
+"""
+
+
+@add_start_docstrings(
+    "The bare Phi-3 model outputting raw hidden-states without any specific head on top.",
+    PHI3_START_DOCSTRING,
+)
+class Phi3PreTrainedModel(PreTrainedModel):
+    config_class = Phi3Config
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["Phi3DecoderLayer"]
+    _skip_keys_device_placement = "past_key_values"
+    _supports_flash_attn_2 = True
+    _supports_sdpa = False
+    _supports_cache_class = True
+
+    _version = "0.0.5"
+
+    def _init_weights(self, 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_()
+
+
+PHI3_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
+            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
+            it.
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are input IDs?](../glossary#input-ids)
+        attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            [What are attention masks?](../glossary#attention-mask)
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            If `past_key_values` is used, optionally only the last `input_ids` have to be input (see
+            `past_key_values`).
+
+            If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`]
+            and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
+            information on the default strategy.
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+        position_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
+            config.n_positions - 1]`.
+
+            [What are position IDs?](../glossary#position-ids)
+        past_key_values (`Cache` or `tuple(tuple(torch.FloatTensor))`, *optional*):
+            Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
+            blocks) that can be used to speed up sequential decoding. This typically consists in the `past_key_values`
+            returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`.
+
+            Two formats are allowed:
+            - a [`~cache_utils.Cache`] instance;
+            - Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
+            shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`). This is also known as the legacy
+            cache format.
+
+            The model will output the same cache format that is fed as input. If no `past_key_values` are passed, the
+            legacy cache format will be returned.
+
+            If `past_key_values` are used, the user can optionally input only the last `input_ids` (those that don't
+            have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `input_ids`
+            of shape `(batch_size, sequence_length)`.
+        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
+            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This
+            is useful if you want more control over how to convert `input_ids` indices into associated vectors than the
+            model's internal embedding lookup matrix.
+        use_cache (`bool`, *optional*):
+            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
+            `past_key_values`).
+        output_attentions (`bool`, *optional*):
+            Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
+            tensors for more detail.
+        output_hidden_states (`bool`, *optional*):
+            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
+            more detail.
+        return_dict (`bool`, *optional*):
+            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+"""
+
+
+@add_start_docstrings(
+    "The bare Phi-3 model outputting raw hidden-states without any specific head on top.",
+    PHI3_START_DOCSTRING,
+)
+class Phi3Model(Phi3PreTrainedModel):
+    """
+    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`Phi3DecoderLayer`]
+
+    Args:
+        config: Phi3Config
+    """
+
+    def __init__(self, config: Phi3Config):
+        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.embed_dropout = nn.Dropout(config.embd_pdrop)
+        self.layers = nn.ModuleList(
+            [Phi3DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self._attn_implementation = config._attn_implementation
+        self.norm = Phi3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+        self.gradient_checkpointing = False
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(PHI3_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, 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
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # retrieve input_ids and inputs_embeds
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
+        elif input_ids is not None:
+            batch_size, seq_length = input_ids.shape[:2]
+        elif inputs_embeds is not None:
+            batch_size, seq_length = inputs_embeds.shape[:2]
+        else:
+            raise ValueError("You have to specify either input_ids or inputs_embeds")
+
+        past_key_values_length = 0
+
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning_once(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        if use_cache:
+            use_legacy_cache = not isinstance(past_key_values, Cache)
+            if use_legacy_cache:
+                past_key_values = DynamicCache.from_legacy_cache(past_key_values)
+            past_key_values_length = past_key_values.get_usable_length(seq_length)
+
+        if position_ids is None:
+            device = input_ids.device if input_ids is not None else inputs_embeds.device
+            position_ids = torch.arange(
+                past_key_values_length, seq_length + past_key_values_length, dtype=torch.long, device=device
+            )
+            position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
+        else:
+            position_ids = position_ids.view(-1, seq_length).long()
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        if attention_mask is not None and self._attn_implementation == "flash_attention_2" and use_cache:
+            is_padding_right = attention_mask[:, -1].sum().item() != batch_size
+            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 Phi3. Make sure to "
+                    " call `tokenizer.padding_side  = 'left'` before tokenizing the input. "
+                )
+
+        if self._attn_implementation == "flash_attention_2":
+            # 2d mask is passed through the layers
+            attention_mask = attention_mask if (attention_mask is not None and 0 in attention_mask) else None
+        else:
+            # 4d mask is passed through the layers
+            attention_mask = _prepare_4d_causal_attention_mask(
+                attention_mask,
+                (batch_size, seq_length),
+                inputs_embeds,
+                past_key_values_length,
+                sliding_window=self.config.sliding_window,
+            )
+
+        hidden_states = inputs_embeds
+
+        # decoder layers
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attns = () if output_attentions else None
+        next_decoder_cache = None
+
+        for decoder_layer in self.layers:
+            if output_hidden_states:
+                all_hidden_states += (hidden_states,)
+
+            if self.gradient_checkpointing and self.training:
+                layer_outputs = self._gradient_checkpointing_func(
+                    decoder_layer.__call__,
+                    hidden_states,
+                    attention_mask,
+                    position_ids,
+                    past_key_values,
+                    output_attentions,
+                    use_cache,
+                )
+            else:
+                layer_outputs = decoder_layer(
+                    hidden_states,
+                    attention_mask=attention_mask,
+                    position_ids=position_ids,
+                    past_key_value=past_key_values,
+                    output_attentions=output_attentions,
+                    use_cache=use_cache,
+                )
+
+            hidden_states = layer_outputs[0]
+
+            if use_cache:
+                next_decoder_cache = layer_outputs[2 if output_attentions else 1]
+
+            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,)
+
+        next_cache = None
+        if use_cache:
+            next_cache = next_decoder_cache.to_legacy_cache() if use_legacy_cache else next_decoder_cache
+        if not return_dict:
+            return tuple(v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns] if v is not None)
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=next_cache,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attns,
+        )
+
+
+class Phi3ForCausalLM(Phi3PreTrainedModel):
+    _tied_weights_keys = ["lm_head.weight"]
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.__init__ with Llama->Phi3
+    def __init__(self, config):
+        super().__init__(config)
+        self.model = Phi3Model(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()
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.get_input_embeddings
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.set_input_embeddings
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.get_output_embeddings
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.set_output_embeddings
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head = new_embeddings
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.set_decoder
+    def set_decoder(self, decoder):
+        self.model = decoder
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.get_decoder
+    def get_decoder(self):
+        return self.model
+
+    # Ignore copy
+    @add_start_docstrings_to_model_forward(PHI3_INPUTS_DOCSTRING)
+    @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = 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,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, 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]`.
+
+        Returns:
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, Phi3ForCausalLM
+
+        >>> model = Phi3ForCausalLM.from_pretrained("microsoft/phi-3-mini-4k-instruct")
+        >>> tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-3-mini-4k-instruct")
+
+        >>> prompt = "This is an example script ."
+        >>> 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]
+        'This is an example script .\n Certainly! Below is a sample script that demonstrates a simple task, such as calculating the sum'
+        ```"""
+
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = 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,
+            return_dict=return_dict,
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states)
+        logits = logits.float()
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    # Copied from transformers.models.persimmon.modeling_persimmon.PersimmonForCausalLM.prepare_inputs_for_generation
+    def prepare_inputs_for_generation(
+        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
+    ):
+        # When the first time input length reached long and short factor switching point, enforce re-compute cache
+        # It will cause downside of slower at this single token position, however, better than current failure.
+        if past_key_values and self.config.rope_scaling and input_ids.shape[1] >= self.config.original_max_position_embeddings + 1:
+            past_length = past_key_values.seen_tokens if isinstance(past_key_values, Cache) else past_key_values[0][0].shape[2]
+            if past_length <= self.config.original_max_position_embeddings:
+                past_key_values = None
+
+        if past_key_values is not None:
+            if isinstance(past_key_values, Cache):
+                cache_length = past_key_values.get_seq_length()
+                past_length = past_key_values.seen_tokens
+                max_cache_length = past_key_values.get_max_length()
+            else:
+                cache_length = past_length = past_key_values[0][0].shape[2]
+                max_cache_length = None
+
+            # Keep only the unprocessed tokens:
+            # 1 - If the length of the attention_mask exceeds the length of input_ids, then we are in a setting where
+            # some of the inputs are exclusively passed as part of the cache (e.g. when passing input_embeds as
+            # input)
+            if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]:
+                input_ids = input_ids[:, -(attention_mask.shape[1] - past_length) :]
+            # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard
+            # input_ids based on the past_length.
+            elif past_length < input_ids.shape[1]:
+                input_ids = input_ids[:, past_length:]
+            # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens.
+
+            # If we are about to go beyond the maximum cache length, we need to crop the input attention mask.
+            if (
+                max_cache_length is not None
+                and attention_mask is not None
+                and cache_length + input_ids.shape[1] > max_cache_length
+            ):
+                attention_mask = attention_mask[:, -max_cache_length:]
+
+        position_ids = kwargs.get("position_ids", None)
+        if attention_mask is not None and position_ids is None:
+            # create position_ids on the fly for batch generation
+            position_ids = attention_mask.long().cumsum(-1) - 1
+            position_ids.masked_fill_(attention_mask == 0, 1)
+            if past_key_values:
+                position_ids = position_ids[:, -input_ids.shape[1] :]
+
+        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
+        if inputs_embeds is not None and past_key_values is None:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        model_inputs.update(
+            {
+                "position_ids": position_ids,
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "attention_mask": attention_mask,
+            }
+        )
+        return model_inputs
+
+    @staticmethod
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM._reorder_cache
+    def _reorder_cache(past_key_values, beam_idx):
+        reordered_past = ()
+        for layer_past in past_key_values:
+            reordered_past += (
+                tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past),
+            )
+        return reordered_past
+
+
+@add_start_docstrings(
+    """
+    The [`Phi3Model`] with a sequence classification head on top (linear layer).
+
+    [`Phi3ForSequenceClassification`] 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. Since it cannot guess the
+    padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take the last value in
+    each row of the batch).
+    """,
+    PHI3_START_DOCSTRING,
+)
+# Copied from transformers.models.llama.modeling_llama.LlamaForSequenceClassification with Llama->Phi3, LLAMA->PHI3, self.transformer->self.model, transformer_outputs->model_outputs
+class Phi3ForSequenceClassification(Phi3PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = Phi3Model(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):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(PHI3_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = 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,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, 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).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        model_outputs = 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,
+            return_dict=return_dict,
+        )
+        hidden_states = model_outputs[0]
+        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:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                # if no pad token found, use modulo instead of reverse indexing for ONNX compatibility
+                sequence_lengths = torch.eq(input_ids, self.config.pad_token_id).int().argmax(-1) - 1
+                sequence_lengths = sequence_lengths % input_ids.shape[-1]
+                sequence_lengths = sequence_lengths.to(logits.device)
+            else:
+                sequence_lengths = -1
+
+        pooled_logits = logits[torch.arange(batch_size, device=logits.device), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            labels = labels.to(logits.device)
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(pooled_logits, labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(pooled_logits, labels)
+        if not return_dict:
+            output = (pooled_logits,) + model_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=model_outputs.past_key_values,
+            hidden_states=model_outputs.hidden_states,
+            attentions=model_outputs.attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    [`Phi3Model`] with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for
+    Named-Entity-Recognition (NER) tasks.
+    """,
+    PHI3_START_DOCSTRING,
+)
+# Copied from transformers.models.mpt.modeling_mpt.MptForTokenClassification with Mpt->Phi3,MPT->PHI3,self.transformer->self.model,transformer_outputs->model_outputs
+class Phi3ForTokenClassification(Phi3PreTrainedModel):
+    def __init__(self, config: Phi3Config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+
+        self.model = Phi3Model(config)
+        if hasattr(config, "classifier_dropout") and config.classifier_dropout is not None:
+            classifier_dropout = config.classifier_dropout
+        elif hasattr(config, "hidden_dropout") and config.hidden_dropout is not None:
+            classifier_dropout = config.hidden_dropout
+        else:
+            classifier_dropout = 0.1
+        self.dropout = nn.Dropout(classifier_dropout)
+        self.classifier = nn.Linear(config.hidden_size, config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings_to_model_forward(PHI3_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=TokenClassifierOutput,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        **deprecated_arguments,
+    ) -> Union[Tuple[torch.Tensor], 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).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        model_outputs = self.model(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = model_outputs[0]
+        hidden_states = self.dropout(hidden_states)
+        logits = self.classifier(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # move labels to correct device to enable model parallelism
+            labels = labels.to(logits.device)
+            batch_size, seq_length = labels.shape
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(
+                logits.view(batch_size * seq_length, self.num_labels), labels.view(batch_size * seq_length)
+            )
+
+        if not return_dict:
+            output = (logits,) + model_outputs[2:]
+            return ((loss,) + output) if loss is not None else output
+
+        return TokenClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=model_outputs.hidden_states,
+            attentions=model_outputs.attentions,
+        )

runs/Sep29_18-00-51_WIN/events.out.tfevents.1727604167.WIN.2364.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b8ab25c0854580bb4841eb0631dea2e5ba968995a2523e6020f07e088eec0dd
+size 32170

runs/Sep29_18-00-51_WIN/events.out.tfevents.1727607771.WIN.2364.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab265a3d8a853c8fe8a2205c5ba6bcd4bde16fac2d6404d1360628450e7f2250
+size 359

runs/Sep29_19-03-16_WIN/events.out.tfevents.1727607905.WIN.19292.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17898d5e9687170c4322a7698d0cbaf21cd85ae2bbeedecbe769917fc6fe8504
+size 32187

runs/Sep29_19-03-16_WIN/events.out.tfevents.1727611526.WIN.19292.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fe8660aba54aa136f2d27fe0ecbd0707f04f643c89a16bd536ecd1ef4102817
+size 359

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -0,0 +1,132 @@
+{
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "add_prefix_space": null,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "32000": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32001": {
+      "content": "<|assistant|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32002": {
+      "content": "<|placeholder1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32003": {
+      "content": "<|placeholder2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32004": {
+      "content": "<|placeholder3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32005": {
+      "content": "<|placeholder4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32006": {
+      "content": "<|system|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32007": {
+      "content": "<|end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32008": {
+      "content": "<|placeholder5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32009": {
+      "content": "<|placeholder6|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32010": {
+      "content": "<|user|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "chat_template": "{{ '<s>' }}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ '<|system|>\n' + system_message + '<|end|>\n' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|user|>\n' + content + '<|end|>\n<|assistant|>\n' }}{% elif message['role'] == 'assistant' %}{{ content + '<|end|>' + '\n' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|end|>",
+  "legacy": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "split_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 0.16016016016016016,
+    "total_flos": 2.9103944129622835e+17,
+    "train_loss": 1.3281120185852051,
+    "train_runtime": 3519.7932,
+    "train_samples_per_second": 4.546,
+    "train_steps_per_second": 0.284
+}

trainer_log.jsonl

@@ -0,0 +1,111 @@
+{"current_steps": 10, "total_steps": 1000, "loss": 1.2106, "learning_rate": 0.0001, "epoch": 0.0016016016016016017, "percentage": 1.0, "elapsed_time": "0:00:25", "remaining_time": "0:42:48"}
+{"current_steps": 20, "total_steps": 1000, "loss": 1.4712, "learning_rate": 0.0001, "epoch": 0.0032032032032032033, "percentage": 2.0, "elapsed_time": "0:00:51", "remaining_time": "0:42:15"}
+{"current_steps": 30, "total_steps": 1000, "loss": 1.2974, "learning_rate": 0.0001, "epoch": 0.004804804804804805, "percentage": 3.0, "elapsed_time": "0:01:18", "remaining_time": "0:42:26"}
+{"current_steps": 40, "total_steps": 1000, "loss": 1.304, "learning_rate": 0.0001, "epoch": 0.006406406406406407, "percentage": 4.0, "elapsed_time": "0:01:47", "remaining_time": "0:42:55"}
+{"current_steps": 50, "total_steps": 1000, "loss": 1.2863, "learning_rate": 0.0001, "epoch": 0.008008008008008008, "percentage": 5.0, "elapsed_time": "0:02:11", "remaining_time": "0:41:47"}
+{"current_steps": 60, "total_steps": 1000, "loss": 1.5074, "learning_rate": 0.0001, "epoch": 0.00960960960960961, "percentage": 6.0, "elapsed_time": "0:02:35", "remaining_time": "0:40:33"}
+{"current_steps": 70, "total_steps": 1000, "loss": 1.3915, "learning_rate": 0.0001, "epoch": 0.01121121121121121, "percentage": 7.0, "elapsed_time": "0:03:00", "remaining_time": "0:39:59"}
+{"current_steps": 80, "total_steps": 1000, "loss": 1.3476, "learning_rate": 0.0001, "epoch": 0.012812812812812813, "percentage": 8.0, "elapsed_time": "0:03:26", "remaining_time": "0:39:37"}
+{"current_steps": 90, "total_steps": 1000, "loss": 1.3663, "learning_rate": 0.0001, "epoch": 0.014414414414414415, "percentage": 9.0, "elapsed_time": "0:03:49", "remaining_time": "0:38:39"}
+{"current_steps": 100, "total_steps": 1000, "loss": 1.428, "learning_rate": 0.0001, "epoch": 0.016016016016016016, "percentage": 10.0, "elapsed_time": "0:04:12", "remaining_time": "0:37:55"}
+{"current_steps": 100, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.016016016016016016, "percentage": 10.0, "elapsed_time": "0:04:25", "remaining_time": "0:39:51"}
+{"current_steps": 110, "total_steps": 1000, "loss": 1.2343, "learning_rate": 0.0001, "epoch": 0.01761761761761762, "percentage": 11.0, "elapsed_time": "0:06:06", "remaining_time": "0:49:23"}
+{"current_steps": 120, "total_steps": 1000, "loss": 1.343, "learning_rate": 0.0001, "epoch": 0.01921921921921922, "percentage": 12.0, "elapsed_time": "0:06:28", "remaining_time": "0:47:32"}
+{"current_steps": 130, "total_steps": 1000, "loss": 1.3258, "learning_rate": 0.0001, "epoch": 0.02082082082082082, "percentage": 13.0, "elapsed_time": "0:06:55", "remaining_time": "0:46:22"}
+{"current_steps": 140, "total_steps": 1000, "loss": 1.3007, "learning_rate": 0.0001, "epoch": 0.02242242242242242, "percentage": 14.0, "elapsed_time": "0:07:23", "remaining_time": "0:45:25"}
+{"current_steps": 150, "total_steps": 1000, "loss": 1.35, "learning_rate": 0.0001, "epoch": 0.024024024024024024, "percentage": 15.0, "elapsed_time": "0:07:52", "remaining_time": "0:44:40"}
+{"current_steps": 160, "total_steps": 1000, "loss": 1.2401, "learning_rate": 0.0001, "epoch": 0.025625625625625627, "percentage": 16.0, "elapsed_time": "0:08:23", "remaining_time": "0:44:00"}
+{"current_steps": 170, "total_steps": 1000, "loss": 1.2047, "learning_rate": 0.0001, "epoch": 0.027227227227227226, "percentage": 17.0, "elapsed_time": "0:08:51", "remaining_time": "0:43:15"}
+{"current_steps": 180, "total_steps": 1000, "loss": 1.3125, "learning_rate": 0.0001, "epoch": 0.02882882882882883, "percentage": 18.0, "elapsed_time": "0:09:18", "remaining_time": "0:42:26"}
+{"current_steps": 190, "total_steps": 1000, "loss": 1.283, "learning_rate": 0.0001, "epoch": 0.03043043043043043, "percentage": 19.0, "elapsed_time": "0:09:45", "remaining_time": "0:41:37"}
+{"current_steps": 200, "total_steps": 1000, "loss": 1.2653, "learning_rate": 0.0001, "epoch": 0.03203203203203203, "percentage": 20.0, "elapsed_time": "0:10:12", "remaining_time": "0:40:48"}
+{"current_steps": 200, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.03203203203203203, "percentage": 20.0, "elapsed_time": "0:10:26", "remaining_time": "0:41:44"}
+{"current_steps": 210, "total_steps": 1000, "loss": 1.3293, "learning_rate": 0.0001, "epoch": 0.033633633633633635, "percentage": 21.0, "elapsed_time": "0:12:14", "remaining_time": "0:46:02"}
+{"current_steps": 220, "total_steps": 1000, "loss": 1.5406, "learning_rate": 0.0001, "epoch": 0.03523523523523524, "percentage": 22.0, "elapsed_time": "0:12:42", "remaining_time": "0:45:02"}
+{"current_steps": 230, "total_steps": 1000, "loss": 1.2827, "learning_rate": 0.0001, "epoch": 0.036836836836836834, "percentage": 23.0, "elapsed_time": "0:13:09", "remaining_time": "0:44:02"}
+{"current_steps": 240, "total_steps": 1000, "loss": 1.2914, "learning_rate": 0.0001, "epoch": 0.03843843843843844, "percentage": 24.0, "elapsed_time": "0:13:37", "remaining_time": "0:43:08"}
+{"current_steps": 250, "total_steps": 1000, "loss": 1.3335, "learning_rate": 0.0001, "epoch": 0.04004004004004004, "percentage": 25.0, "elapsed_time": "0:14:02", "remaining_time": "0:42:08"}
+{"current_steps": 260, "total_steps": 1000, "loss": 1.4173, "learning_rate": 0.0001, "epoch": 0.04164164164164164, "percentage": 26.0, "elapsed_time": "0:14:28", "remaining_time": "0:41:10"}
+{"current_steps": 270, "total_steps": 1000, "loss": 1.2077, "learning_rate": 0.0001, "epoch": 0.043243243243243246, "percentage": 27.0, "elapsed_time": "0:14:58", "remaining_time": "0:40:28"}
+{"current_steps": 280, "total_steps": 1000, "loss": 1.2665, "learning_rate": 0.0001, "epoch": 0.04484484484484484, "percentage": 28.0, "elapsed_time": "0:15:24", "remaining_time": "0:39:36"}
+{"current_steps": 290, "total_steps": 1000, "loss": 1.3744, "learning_rate": 0.0001, "epoch": 0.046446446446446445, "percentage": 29.0, "elapsed_time": "0:15:50", "remaining_time": "0:38:47"}
+{"current_steps": 300, "total_steps": 1000, "loss": 1.3084, "learning_rate": 0.0001, "epoch": 0.04804804804804805, "percentage": 30.0, "elapsed_time": "0:16:18", "remaining_time": "0:38:02"}
+{"current_steps": 300, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.04804804804804805, "percentage": 30.0, "elapsed_time": "0:16:31", "remaining_time": "0:38:33"}
+{"current_steps": 310, "total_steps": 1000, "loss": 1.3119, "learning_rate": 0.0001, "epoch": 0.04964964964964965, "percentage": 31.0, "elapsed_time": "0:18:15", "remaining_time": "0:40:38"}
+{"current_steps": 320, "total_steps": 1000, "loss": 1.1759, "learning_rate": 0.0001, "epoch": 0.051251251251251254, "percentage": 32.0, "elapsed_time": "0:18:45", "remaining_time": "0:39:51"}
+{"current_steps": 330, "total_steps": 1000, "loss": 1.4001, "learning_rate": 0.0001, "epoch": 0.05285285285285285, "percentage": 33.0, "elapsed_time": "0:19:08", "remaining_time": "0:38:51"}
+{"current_steps": 340, "total_steps": 1000, "loss": 1.3775, "learning_rate": 0.0001, "epoch": 0.05445445445445445, "percentage": 34.0, "elapsed_time": "0:19:32", "remaining_time": "0:37:56"}
+{"current_steps": 350, "total_steps": 1000, "loss": 1.2751, "learning_rate": 0.0001, "epoch": 0.056056056056056056, "percentage": 35.0, "elapsed_time": "0:19:58", "remaining_time": "0:37:05"}
+{"current_steps": 360, "total_steps": 1000, "loss": 1.3655, "learning_rate": 0.0001, "epoch": 0.05765765765765766, "percentage": 36.0, "elapsed_time": "0:20:23", "remaining_time": "0:36:15"}
+{"current_steps": 370, "total_steps": 1000, "loss": 1.3717, "learning_rate": 0.0001, "epoch": 0.05925925925925926, "percentage": 37.0, "elapsed_time": "0:20:50", "remaining_time": "0:35:29"}
+{"current_steps": 380, "total_steps": 1000, "loss": 1.22, "learning_rate": 0.0001, "epoch": 0.06086086086086086, "percentage": 38.0, "elapsed_time": "0:21:14", "remaining_time": "0:34:39"}
+{"current_steps": 390, "total_steps": 1000, "loss": 1.3391, "learning_rate": 0.0001, "epoch": 0.06246246246246246, "percentage": 39.0, "elapsed_time": "0:21:42", "remaining_time": "0:33:57"}
+{"current_steps": 400, "total_steps": 1000, "loss": 1.3234, "learning_rate": 0.0001, "epoch": 0.06406406406406406, "percentage": 40.0, "elapsed_time": "0:22:07", "remaining_time": "0:33:11"}
+{"current_steps": 400, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.06406406406406406, "percentage": 40.0, "elapsed_time": "0:22:20", "remaining_time": "0:33:31"}
+{"current_steps": 410, "total_steps": 1000, "loss": 1.1899, "learning_rate": 0.0001, "epoch": 0.06566566566566566, "percentage": 41.0, "elapsed_time": "0:24:05", "remaining_time": "0:34:40"}
+{"current_steps": 420, "total_steps": 1000, "loss": 1.4017, "learning_rate": 0.0001, "epoch": 0.06726726726726727, "percentage": 42.0, "elapsed_time": "0:24:28", "remaining_time": "0:33:47"}
+{"current_steps": 430, "total_steps": 1000, "loss": 1.4969, "learning_rate": 0.0001, "epoch": 0.06886886886886887, "percentage": 43.0, "elapsed_time": "0:24:51", "remaining_time": "0:32:56"}
+{"current_steps": 440, "total_steps": 1000, "loss": 1.2447, "learning_rate": 0.0001, "epoch": 0.07047047047047048, "percentage": 44.0, "elapsed_time": "0:25:17", "remaining_time": "0:32:11"}
+{"current_steps": 450, "total_steps": 1000, "loss": 1.3669, "learning_rate": 0.0001, "epoch": 0.07207207207207207, "percentage": 45.0, "elapsed_time": "0:25:43", "remaining_time": "0:31:27"}
+{"current_steps": 460, "total_steps": 1000, "loss": 1.5245, "learning_rate": 0.0001, "epoch": 0.07367367367367367, "percentage": 46.0, "elapsed_time": "0:26:09", "remaining_time": "0:30:42"}
+{"current_steps": 470, "total_steps": 1000, "loss": 1.4036, "learning_rate": 0.0001, "epoch": 0.07527527527527528, "percentage": 47.0, "elapsed_time": "0:26:34", "remaining_time": "0:29:58"}
+{"current_steps": 480, "total_steps": 1000, "loss": 1.4029, "learning_rate": 0.0001, "epoch": 0.07687687687687687, "percentage": 48.0, "elapsed_time": "0:26:57", "remaining_time": "0:29:12"}
+{"current_steps": 490, "total_steps": 1000, "loss": 1.5582, "learning_rate": 0.0001, "epoch": 0.07847847847847848, "percentage": 49.0, "elapsed_time": "0:27:21", "remaining_time": "0:28:28"}
+{"current_steps": 500, "total_steps": 1000, "loss": 1.4091, "learning_rate": 0.0001, "epoch": 0.08008008008008008, "percentage": 50.0, "elapsed_time": "0:27:45", "remaining_time": "0:27:45"}
+{"current_steps": 500, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.08008008008008008, "percentage": 50.0, "elapsed_time": "0:27:58", "remaining_time": "0:27:58"}
+{"current_steps": 510, "total_steps": 1000, "loss": 1.2554, "learning_rate": 0.0001, "epoch": 0.08168168168168168, "percentage": 51.0, "elapsed_time": "0:29:39", "remaining_time": "0:28:30"}
+{"current_steps": 520, "total_steps": 1000, "loss": 1.3238, "learning_rate": 0.0001, "epoch": 0.08328328328328329, "percentage": 52.0, "elapsed_time": "0:30:04", "remaining_time": "0:27:45"}
+{"current_steps": 530, "total_steps": 1000, "loss": 1.2115, "learning_rate": 0.0001, "epoch": 0.08488488488488488, "percentage": 53.0, "elapsed_time": "0:30:30", "remaining_time": "0:27:03"}
+{"current_steps": 540, "total_steps": 1000, "loss": 1.3589, "learning_rate": 0.0001, "epoch": 0.08648648648648649, "percentage": 54.0, "elapsed_time": "0:30:54", "remaining_time": "0:26:20"}
+{"current_steps": 550, "total_steps": 1000, "loss": 1.3267, "learning_rate": 0.0001, "epoch": 0.08808808808808809, "percentage": 55.0, "elapsed_time": "0:31:23", "remaining_time": "0:25:41"}
+{"current_steps": 560, "total_steps": 1000, "loss": 1.2068, "learning_rate": 0.0001, "epoch": 0.08968968968968968, "percentage": 56.0, "elapsed_time": "0:31:48", "remaining_time": "0:24:59"}
+{"current_steps": 570, "total_steps": 1000, "loss": 1.4504, "learning_rate": 0.0001, "epoch": 0.0912912912912913, "percentage": 57.0, "elapsed_time": "0:32:16", "remaining_time": "0:24:20"}
+{"current_steps": 580, "total_steps": 1000, "loss": 1.3775, "learning_rate": 0.0001, "epoch": 0.09289289289289289, "percentage": 58.0, "elapsed_time": "0:32:44", "remaining_time": "0:23:42"}
+{"current_steps": 590, "total_steps": 1000, "loss": 1.3212, "learning_rate": 0.0001, "epoch": 0.0944944944944945, "percentage": 59.0, "elapsed_time": "0:33:09", "remaining_time": "0:23:02"}
+{"current_steps": 600, "total_steps": 1000, "loss": 1.2878, "learning_rate": 0.0001, "epoch": 0.0960960960960961, "percentage": 60.0, "elapsed_time": "0:33:34", "remaining_time": "0:22:23"}
+{"current_steps": 600, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.0960960960960961, "percentage": 60.0, "elapsed_time": "0:33:48", "remaining_time": "0:22:32"}
+{"current_steps": 610, "total_steps": 1000, "loss": 1.4902, "learning_rate": 0.0001, "epoch": 0.09769769769769769, "percentage": 61.0, "elapsed_time": "0:35:31", "remaining_time": "0:22:42"}
+{"current_steps": 620, "total_steps": 1000, "loss": 1.203, "learning_rate": 0.0001, "epoch": 0.0992992992992993, "percentage": 62.0, "elapsed_time": "0:35:57", "remaining_time": "0:22:02"}
+{"current_steps": 630, "total_steps": 1000, "loss": 1.2783, "learning_rate": 0.0001, "epoch": 0.1009009009009009, "percentage": 63.0, "elapsed_time": "0:36:27", "remaining_time": "0:21:24"}
+{"current_steps": 640, "total_steps": 1000, "loss": 1.2326, "learning_rate": 0.0001, "epoch": 0.10250250250250251, "percentage": 64.0, "elapsed_time": "0:36:49", "remaining_time": "0:20:42"}
+{"current_steps": 650, "total_steps": 1000, "loss": 1.3132, "learning_rate": 0.0001, "epoch": 0.1041041041041041, "percentage": 65.0, "elapsed_time": "0:37:15", "remaining_time": "0:20:03"}
+{"current_steps": 660, "total_steps": 1000, "loss": 1.2495, "learning_rate": 0.0001, "epoch": 0.1057057057057057, "percentage": 66.0, "elapsed_time": "0:37:42", "remaining_time": "0:19:25"}
+{"current_steps": 670, "total_steps": 1000, "loss": 1.329, "learning_rate": 0.0001, "epoch": 0.10730730730730731, "percentage": 67.0, "elapsed_time": "0:38:08", "remaining_time": "0:18:47"}
+{"current_steps": 680, "total_steps": 1000, "loss": 1.3178, "learning_rate": 0.0001, "epoch": 0.1089089089089089, "percentage": 68.0, "elapsed_time": "0:38:34", "remaining_time": "0:18:09"}
+{"current_steps": 690, "total_steps": 1000, "loss": 1.3798, "learning_rate": 0.0001, "epoch": 0.11051051051051052, "percentage": 69.0, "elapsed_time": "0:38:59", "remaining_time": "0:17:31"}
+{"current_steps": 700, "total_steps": 1000, "loss": 1.2933, "learning_rate": 0.0001, "epoch": 0.11211211211211211, "percentage": 70.0, "elapsed_time": "0:39:27", "remaining_time": "0:16:54"}
+{"current_steps": 700, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.11211211211211211, "percentage": 70.0, "elapsed_time": "0:39:41", "remaining_time": "0:17:00"}
+{"current_steps": 710, "total_steps": 1000, "loss": 1.2721, "learning_rate": 0.0001, "epoch": 0.11371371371371371, "percentage": 71.0, "elapsed_time": "0:41:23", "remaining_time": "0:16:54"}
+{"current_steps": 720, "total_steps": 1000, "loss": 1.3859, "learning_rate": 0.0001, "epoch": 0.11531531531531532, "percentage": 72.0, "elapsed_time": "0:41:48", "remaining_time": "0:16:15"}
+{"current_steps": 730, "total_steps": 1000, "loss": 1.2536, "learning_rate": 0.0001, "epoch": 0.11691691691691691, "percentage": 73.0, "elapsed_time": "0:42:18", "remaining_time": "0:15:38"}
+{"current_steps": 740, "total_steps": 1000, "loss": 1.3625, "learning_rate": 0.0001, "epoch": 0.11851851851851852, "percentage": 74.0, "elapsed_time": "0:42:44", "remaining_time": "0:15:00"}
+{"current_steps": 750, "total_steps": 1000, "loss": 1.3386, "learning_rate": 0.0001, "epoch": 0.12012012012012012, "percentage": 75.0, "elapsed_time": "0:43:08", "remaining_time": "0:14:22"}
+{"current_steps": 760, "total_steps": 1000, "loss": 1.4023, "learning_rate": 0.0001, "epoch": 0.12172172172172172, "percentage": 76.0, "elapsed_time": "0:43:35", "remaining_time": "0:13:46"}
+{"current_steps": 770, "total_steps": 1000, "loss": 1.2977, "learning_rate": 0.0001, "epoch": 0.12332332332332333, "percentage": 77.0, "elapsed_time": "0:44:00", "remaining_time": "0:13:08"}
+{"current_steps": 780, "total_steps": 1000, "loss": 1.4091, "learning_rate": 0.0001, "epoch": 0.12492492492492492, "percentage": 78.0, "elapsed_time": "0:44:28", "remaining_time": "0:12:32"}
+{"current_steps": 790, "total_steps": 1000, "loss": 1.2513, "learning_rate": 0.0001, "epoch": 0.12652652652652652, "percentage": 79.0, "elapsed_time": "0:44:54", "remaining_time": "0:11:56"}
+{"current_steps": 800, "total_steps": 1000, "loss": 1.3246, "learning_rate": 0.0001, "epoch": 0.12812812812812813, "percentage": 80.0, "elapsed_time": "0:45:19", "remaining_time": "0:11:19"}
+{"current_steps": 800, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.12812812812812813, "percentage": 80.0, "elapsed_time": "0:45:32", "remaining_time": "0:11:23"}
+{"current_steps": 810, "total_steps": 1000, "loss": 1.3271, "learning_rate": 0.0001, "epoch": 0.12972972972972974, "percentage": 81.0, "elapsed_time": "0:47:18", "remaining_time": "0:11:05"}
+{"current_steps": 820, "total_steps": 1000, "loss": 1.3497, "learning_rate": 0.0001, "epoch": 0.13133133133133132, "percentage": 82.0, "elapsed_time": "0:47:49", "remaining_time": "0:10:29"}
+{"current_steps": 830, "total_steps": 1000, "loss": 1.2833, "learning_rate": 0.0001, "epoch": 0.13293293293293293, "percentage": 83.0, "elapsed_time": "0:48:17", "remaining_time": "0:09:53"}
+{"current_steps": 840, "total_steps": 1000, "loss": 1.3491, "learning_rate": 0.0001, "epoch": 0.13453453453453454, "percentage": 84.0, "elapsed_time": "0:48:42", "remaining_time": "0:09:16"}
+{"current_steps": 850, "total_steps": 1000, "loss": 1.2592, "learning_rate": 0.0001, "epoch": 0.13613613613613615, "percentage": 85.0, "elapsed_time": "0:49:05", "remaining_time": "0:08:39"}
+{"current_steps": 860, "total_steps": 1000, "loss": 1.3241, "learning_rate": 0.0001, "epoch": 0.13773773773773773, "percentage": 86.0, "elapsed_time": "0:49:31", "remaining_time": "0:08:03"}
+{"current_steps": 870, "total_steps": 1000, "loss": 1.3809, "learning_rate": 0.0001, "epoch": 0.13933933933933934, "percentage": 87.0, "elapsed_time": "0:49:59", "remaining_time": "0:07:28"}
+{"current_steps": 880, "total_steps": 1000, "loss": 1.294, "learning_rate": 0.0001, "epoch": 0.14094094094094095, "percentage": 88.0, "elapsed_time": "0:50:28", "remaining_time": "0:06:52"}
+{"current_steps": 890, "total_steps": 1000, "loss": 1.3025, "learning_rate": 0.0001, "epoch": 0.14254254254254253, "percentage": 89.0, "elapsed_time": "0:50:52", "remaining_time": "0:06:17"}
+{"current_steps": 900, "total_steps": 1000, "loss": 1.2911, "learning_rate": 0.0001, "epoch": 0.14414414414414414, "percentage": 90.0, "elapsed_time": "0:51:20", "remaining_time": "0:05:42"}
+{"current_steps": 900, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.14414414414414414, "percentage": 90.0, "elapsed_time": "0:51:33", "remaining_time": "0:05:43"}
+{"current_steps": 910, "total_steps": 1000, "loss": 1.3149, "learning_rate": 0.0001, "epoch": 0.14574574574574575, "percentage": 91.0, "elapsed_time": "0:53:14", "remaining_time": "0:05:15"}
+{"current_steps": 920, "total_steps": 1000, "loss": 1.2697, "learning_rate": 0.0001, "epoch": 0.14734734734734733, "percentage": 92.0, "elapsed_time": "0:53:41", "remaining_time": "0:04:40"}
+{"current_steps": 930, "total_steps": 1000, "loss": 1.3088, "learning_rate": 0.0001, "epoch": 0.14894894894894894, "percentage": 93.0, "elapsed_time": "0:54:07", "remaining_time": "0:04:04"}
+{"current_steps": 940, "total_steps": 1000, "loss": 1.3579, "learning_rate": 0.0001, "epoch": 0.15055055055055055, "percentage": 94.0, "elapsed_time": "0:54:33", "remaining_time": "0:03:28"}
+{"current_steps": 950, "total_steps": 1000, "loss": 1.4073, "learning_rate": 0.0001, "epoch": 0.15215215215215216, "percentage": 95.0, "elapsed_time": "0:55:00", "remaining_time": "0:02:53"}
+{"current_steps": 960, "total_steps": 1000, "loss": 1.2242, "learning_rate": 0.0001, "epoch": 0.15375375375375375, "percentage": 96.0, "elapsed_time": "0:55:24", "remaining_time": "0:02:18"}
+{"current_steps": 970, "total_steps": 1000, "loss": 1.3585, "learning_rate": 0.0001, "epoch": 0.15535535535535536, "percentage": 97.0, "elapsed_time": "0:55:50", "remaining_time": "0:01:43"}
+{"current_steps": 980, "total_steps": 1000, "loss": 1.2407, "learning_rate": 0.0001, "epoch": 0.15695695695695697, "percentage": 98.0, "elapsed_time": "0:56:15", "remaining_time": "0:01:08"}
+{"current_steps": 990, "total_steps": 1000, "loss": 1.2609, "learning_rate": 0.0001, "epoch": 0.15855855855855855, "percentage": 99.0, "elapsed_time": "0:56:42", "remaining_time": "0:00:34"}
+{"current_steps": 1000, "total_steps": 1000, "loss": 1.4227, "learning_rate": 0.0001, "epoch": 0.16016016016016016, "percentage": 100.0, "elapsed_time": "0:57:09", "remaining_time": "0:00:00"}
+{"current_steps": 1000, "total_steps": 1000, "eval_loss": 1.3194454908370972, "epoch": 0.16016016016016016, "percentage": 100.0, "elapsed_time": "0:57:22", "remaining_time": "0:00:00"}
+{"current_steps": 1000, "total_steps": 1000, "epoch": 0.16016016016016016, "percentage": 100.0, "elapsed_time": "0:58:37", "remaining_time": "0:00:00"}

trainer_state.json

@@ -0,0 +1,822 @@
+{
+  "best_metric": 1.3194454908370972,
+  "best_model_checkpoint": "saves/Llama-3.2-3B-Instruct/Phi-3.5-mini-instruct-24-9-29\\checkpoint-100",
+  "epoch": 0.16016016016016016,
+  "eval_steps": 100,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0016016016016016017,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2106,
+      "step": 10
+    },
+    {
+      "epoch": 0.0032032032032032033,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4712,
+      "step": 20
+    },
+    {
+      "epoch": 0.004804804804804805,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2974,
+      "step": 30
+    },
+    {
+      "epoch": 0.006406406406406407,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.304,
+      "step": 40
+    },
+    {
+      "epoch": 0.008008008008008008,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2863,
+      "step": 50
+    },
+    {
+      "epoch": 0.00960960960960961,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.5074,
+      "step": 60
+    },
+    {
+      "epoch": 0.01121121121121121,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3915,
+      "step": 70
+    },
+    {
+      "epoch": 0.012812812812812813,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3476,
+      "step": 80
+    },
+    {
+      "epoch": 0.014414414414414415,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3663,
+      "step": 90
+    },
+    {
+      "epoch": 0.016016016016016016,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.428,
+      "step": 100
+    },
+    {
+      "epoch": 0.016016016016016016,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.8069,
+      "eval_samples_per_second": 7.808,
+      "eval_steps_per_second": 3.904,
+      "step": 100
+    },
+    {
+      "epoch": 0.01761761761761762,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2343,
+      "step": 110
+    },
+    {
+      "epoch": 0.01921921921921922,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.343,
+      "step": 120
+    },
+    {
+      "epoch": 0.02082082082082082,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3258,
+      "step": 130
+    },
+    {
+      "epoch": 0.02242242242242242,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3007,
+      "step": 140
+    },
+    {
+      "epoch": 0.024024024024024024,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.35,
+      "step": 150
+    },
+    {
+      "epoch": 0.025625625625625627,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2401,
+      "step": 160
+    },
+    {
+      "epoch": 0.027227227227227226,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2047,
+      "step": 170
+    },
+    {
+      "epoch": 0.02882882882882883,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3125,
+      "step": 180
+    },
+    {
+      "epoch": 0.03043043043043043,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.283,
+      "step": 190
+    },
+    {
+      "epoch": 0.03203203203203203,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2653,
+      "step": 200
+    },
+    {
+      "epoch": 0.03203203203203203,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.8038,
+      "eval_samples_per_second": 7.81,
+      "eval_steps_per_second": 3.905,
+      "step": 200
+    },
+    {
+      "epoch": 0.033633633633633635,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3293,
+      "step": 210
+    },
+    {
+      "epoch": 0.03523523523523524,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.5406,
+      "step": 220
+    },
+    {
+      "epoch": 0.036836836836836834,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2827,
+      "step": 230
+    },
+    {
+      "epoch": 0.03843843843843844,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2914,
+      "step": 240
+    },
+    {
+      "epoch": 0.04004004004004004,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3335,
+      "step": 250
+    },
+    {
+      "epoch": 0.04164164164164164,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4173,
+      "step": 260
+    },
+    {
+      "epoch": 0.043243243243243246,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2077,
+      "step": 270
+    },
+    {
+      "epoch": 0.04484484484484484,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2665,
+      "step": 280
+    },
+    {
+      "epoch": 0.046446446446446445,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3744,
+      "step": 290
+    },
+    {
+      "epoch": 0.04804804804804805,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3084,
+      "step": 300
+    },
+    {
+      "epoch": 0.04804804804804805,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.9051,
+      "eval_samples_per_second": 7.749,
+      "eval_steps_per_second": 3.874,
+      "step": 300
+    },
+    {
+      "epoch": 0.04964964964964965,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3119,
+      "step": 310
+    },
+    {
+      "epoch": 0.051251251251251254,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.1759,
+      "step": 320
+    },
+    {
+      "epoch": 0.05285285285285285,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4001,
+      "step": 330
+    },
+    {
+      "epoch": 0.05445445445445445,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3775,
+      "step": 340
+    },
+    {
+      "epoch": 0.056056056056056056,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2751,
+      "step": 350
+    },
+    {
+      "epoch": 0.05765765765765766,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3655,
+      "step": 360
+    },
+    {
+      "epoch": 0.05925925925925926,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3717,
+      "step": 370
+    },
+    {
+      "epoch": 0.06086086086086086,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.22,
+      "step": 380
+    },
+    {
+      "epoch": 0.06246246246246246,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3391,
+      "step": 390
+    },
+    {
+      "epoch": 0.06406406406406406,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3234,
+      "step": 400
+    },
+    {
+      "epoch": 0.06406406406406406,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.7832,
+      "eval_samples_per_second": 7.823,
+      "eval_steps_per_second": 3.911,
+      "step": 400
+    },
+    {
+      "epoch": 0.06566566566566566,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.1899,
+      "step": 410
+    },
+    {
+      "epoch": 0.06726726726726727,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4017,
+      "step": 420
+    },
+    {
+      "epoch": 0.06886886886886887,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4969,
+      "step": 430
+    },
+    {
+      "epoch": 0.07047047047047048,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2447,
+      "step": 440
+    },
+    {
+      "epoch": 0.07207207207207207,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3669,
+      "step": 450
+    },
+    {
+      "epoch": 0.07367367367367367,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.5245,
+      "step": 460
+    },
+    {
+      "epoch": 0.07527527527527528,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4036,
+      "step": 470
+    },
+    {
+      "epoch": 0.07687687687687687,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4029,
+      "step": 480
+    },
+    {
+      "epoch": 0.07847847847847848,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.5582,
+      "step": 490
+    },
+    {
+      "epoch": 0.08008008008008008,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4091,
+      "step": 500
+    },
+    {
+      "epoch": 0.08008008008008008,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.8939,
+      "eval_samples_per_second": 7.756,
+      "eval_steps_per_second": 3.878,
+      "step": 500
+    },
+    {
+      "epoch": 0.08168168168168168,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2554,
+      "step": 510
+    },
+    {
+      "epoch": 0.08328328328328329,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3238,
+      "step": 520
+    },
+    {
+      "epoch": 0.08488488488488488,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2115,
+      "step": 530
+    },
+    {
+      "epoch": 0.08648648648648649,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3589,
+      "step": 540
+    },
+    {
+      "epoch": 0.08808808808808809,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3267,
+      "step": 550
+    },
+    {
+      "epoch": 0.08968968968968968,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2068,
+      "step": 560
+    },
+    {
+      "epoch": 0.0912912912912913,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4504,
+      "step": 570
+    },
+    {
+      "epoch": 0.09289289289289289,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3775,
+      "step": 580
+    },
+    {
+      "epoch": 0.0944944944944945,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3212,
+      "step": 590
+    },
+    {
+      "epoch": 0.0960960960960961,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2878,
+      "step": 600
+    },
+    {
+      "epoch": 0.0960960960960961,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.7841,
+      "eval_samples_per_second": 7.822,
+      "eval_steps_per_second": 3.911,
+      "step": 600
+    },
+    {
+      "epoch": 0.09769769769769769,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4902,
+      "step": 610
+    },
+    {
+      "epoch": 0.0992992992992993,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.203,
+      "step": 620
+    },
+    {
+      "epoch": 0.1009009009009009,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2783,
+      "step": 630
+    },
+    {
+      "epoch": 0.10250250250250251,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2326,
+      "step": 640
+    },
+    {
+      "epoch": 0.1041041041041041,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3132,
+      "step": 650
+    },
+    {
+      "epoch": 0.1057057057057057,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2495,
+      "step": 660
+    },
+    {
+      "epoch": 0.10730730730730731,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.329,
+      "step": 670
+    },
+    {
+      "epoch": 0.1089089089089089,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3178,
+      "step": 680
+    },
+    {
+      "epoch": 0.11051051051051052,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3798,
+      "step": 690
+    },
+    {
+      "epoch": 0.11211211211211211,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2933,
+      "step": 700
+    },
+    {
+      "epoch": 0.11211211211211211,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.8596,
+      "eval_samples_per_second": 7.776,
+      "eval_steps_per_second": 3.888,
+      "step": 700
+    },
+    {
+      "epoch": 0.11371371371371371,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2721,
+      "step": 710
+    },
+    {
+      "epoch": 0.11531531531531532,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3859,
+      "step": 720
+    },
+    {
+      "epoch": 0.11691691691691691,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2536,
+      "step": 730
+    },
+    {
+      "epoch": 0.11851851851851852,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3625,
+      "step": 740
+    },
+    {
+      "epoch": 0.12012012012012012,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3386,
+      "step": 750
+    },
+    {
+      "epoch": 0.12172172172172172,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4023,
+      "step": 760
+    },
+    {
+      "epoch": 0.12332332332332333,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2977,
+      "step": 770
+    },
+    {
+      "epoch": 0.12492492492492492,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4091,
+      "step": 780
+    },
+    {
+      "epoch": 0.12652652652652652,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2513,
+      "step": 790
+    },
+    {
+      "epoch": 0.12812812812812813,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3246,
+      "step": 800
+    },
+    {
+      "epoch": 0.12812812812812813,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.7102,
+      "eval_samples_per_second": 7.868,
+      "eval_steps_per_second": 3.934,
+      "step": 800
+    },
+    {
+      "epoch": 0.12972972972972974,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3271,
+      "step": 810
+    },
+    {
+      "epoch": 0.13133133133133132,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3497,
+      "step": 820
+    },
+    {
+      "epoch": 0.13293293293293293,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2833,
+      "step": 830
+    },
+    {
+      "epoch": 0.13453453453453454,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3491,
+      "step": 840
+    },
+    {
+      "epoch": 0.13613613613613615,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2592,
+      "step": 850
+    },
+    {
+      "epoch": 0.13773773773773773,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3241,
+      "step": 860
+    },
+    {
+      "epoch": 0.13933933933933934,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3809,
+      "step": 870
+    },
+    {
+      "epoch": 0.14094094094094095,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.294,
+      "step": 880
+    },
+    {
+      "epoch": 0.14254254254254253,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3025,
+      "step": 890
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2911,
+      "step": 900
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.7786,
+      "eval_samples_per_second": 7.826,
+      "eval_steps_per_second": 3.913,
+      "step": 900
+    },
+    {
+      "epoch": 0.14574574574574575,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3149,
+      "step": 910
+    },
+    {
+      "epoch": 0.14734734734734733,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2697,
+      "step": 920
+    },
+    {
+      "epoch": 0.14894894894894894,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3088,
+      "step": 930
+    },
+    {
+      "epoch": 0.15055055055055055,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3579,
+      "step": 940
+    },
+    {
+      "epoch": 0.15215215215215216,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4073,
+      "step": 950
+    },
+    {
+      "epoch": 0.15375375375375375,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2242,
+      "step": 960
+    },
+    {
+      "epoch": 0.15535535535535536,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.3585,
+      "step": 970
+    },
+    {
+      "epoch": 0.15695695695695697,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2407,
+      "step": 980
+    },
+    {
+      "epoch": 0.15855855855855855,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.2609,
+      "step": 990
+    },
+    {
+      "epoch": 0.16016016016016016,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001,
+      "loss": 1.4227,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16016016016016016,
+      "eval_loss": 1.3194454908370972,
+      "eval_runtime": 12.8395,
+      "eval_samples_per_second": 7.788,
+      "eval_steps_per_second": 3.894,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16016016016016016,
+      "step": 1000,
+      "total_flos": 2.9103944129622835e+17,
+      "train_loss": 1.3281120185852051,
+      "train_runtime": 3519.7932,
+      "train_samples_per_second": 4.546,
+      "train_steps_per_second": 0.284
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.9103944129622835e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac058a69c22218bb0c2078f569e04d18b3cb58ea5e7c7e8cbb9ce207f7a0662c
+size 5432