update

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 +1,8 @@
+{
+    "epoch": 1.9978094194961664,
+    "total_flos": 131893560147968.0,
+    "train_loss": 0.08742370063708604,
+    "train_runtime": 41055.7102,
+    "train_samples_per_second": 1.778,
+    "train_steps_per_second": 0.03
+}

config.json

@@ -0,0 +1,139 @@
+{
+  "_name_or_path": "./saves/phi3/sft/checkpoint-922",
+  "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": "bfloat16",
+  "transformers_version": "4.44.0",
+  "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)}"
+            )

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": "4.44.0"
+}

model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2632d9b32ff555d5395b67cb2c4e099102a587286a542db4dbaafb2f88874f9
+size 4972489328

model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1cc24d5e4f73e7894f5434ed9cbf0bcdc437fae1130065cae60d000c0849ce3
+size 2669692552

model.safetensors.index.json

@@ -0,0 +1,202 @@
+{
+  "metadata": {
+    "total_size": 7642159104
+  },
+  "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,
+        )

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": 1.9978094194961664,
+    "total_flos": 131893560147968.0,
+    "train_loss": 0.08742370063708604,
+    "train_runtime": 41055.7102,
+    "train_samples_per_second": 1.778,
+    "train_steps_per_second": 0.03
+}

trainer_log.jsonl

@@ -0,0 +1,244 @@
+{"current_steps": 5, "total_steps": 1216, "loss": 0.6863, "accuracy": 0.4000000059604645, "learning_rate": 4.0983606557377046e-08, "epoch": 0.008214676889375685, "percentage": 0.41, "elapsed_time": "0:02:53", "remaining_time": "11:41:09"}
+{"current_steps": 10, "total_steps": 1216, "loss": 0.6278, "accuracy": 0.786666750907898, "learning_rate": 8.196721311475409e-08, "epoch": 0.01642935377875137, "percentage": 0.82, "elapsed_time": "0:05:41", "remaining_time": "11:26:03"}
+{"current_steps": 15, "total_steps": 1216, "loss": 0.4689, "accuracy": 0.9466667175292969, "learning_rate": 1.2295081967213113e-07, "epoch": 0.024644030668127054, "percentage": 1.23, "elapsed_time": "0:08:29", "remaining_time": "11:19:41"}
+{"current_steps": 20, "total_steps": 1216, "loss": 0.3581, "accuracy": 0.9600000381469727, "learning_rate": 1.6393442622950818e-07, "epoch": 0.03285870755750274, "percentage": 1.64, "elapsed_time": "0:11:17", "remaining_time": "11:15:13"}
+{"current_steps": 25, "total_steps": 1216, "loss": 0.3047, "accuracy": 0.9200000166893005, "learning_rate": 2.0491803278688524e-07, "epoch": 0.04107338444687842, "percentage": 2.06, "elapsed_time": "0:14:05", "remaining_time": "11:11:24"}
+{"current_steps": 30, "total_steps": 1216, "loss": 0.2507, "accuracy": 0.9333333969116211, "learning_rate": 2.4590163934426226e-07, "epoch": 0.04928806133625411, "percentage": 2.47, "elapsed_time": "0:16:53", "remaining_time": "11:07:44"}
+{"current_steps": 35, "total_steps": 1216, "loss": 0.2291, "accuracy": 0.9466666579246521, "learning_rate": 2.868852459016393e-07, "epoch": 0.05750273822562979, "percentage": 2.88, "elapsed_time": "0:19:41", "remaining_time": "11:04:23"}
+{"current_steps": 40, "total_steps": 1216, "loss": 0.2061, "accuracy": 0.9333333373069763, "learning_rate": 3.2786885245901637e-07, "epoch": 0.06571741511500548, "percentage": 3.29, "elapsed_time": "0:22:29", "remaining_time": "11:01:27"}
+{"current_steps": 45, "total_steps": 1216, "loss": 0.1803, "accuracy": 0.9466667175292969, "learning_rate": 3.6885245901639347e-07, "epoch": 0.07393209200438117, "percentage": 3.7, "elapsed_time": "0:25:17", "remaining_time": "10:58:19"}
+{"current_steps": 50, "total_steps": 1216, "loss": 0.1609, "accuracy": 0.9600000381469727, "learning_rate": 4.0983606557377047e-07, "epoch": 0.08214676889375684, "percentage": 4.11, "elapsed_time": "0:28:06", "remaining_time": "10:55:20"}
+{"current_steps": 55, "total_steps": 1216, "loss": 0.1759, "accuracy": 0.9333333373069763, "learning_rate": 4.508196721311475e-07, "epoch": 0.09036144578313253, "percentage": 4.52, "elapsed_time": "0:30:53", "remaining_time": "10:52:13"}
+{"current_steps": 60, "total_steps": 1216, "loss": 0.166, "accuracy": 0.9200000166893005, "learning_rate": 4.918032786885245e-07, "epoch": 0.09857612267250822, "percentage": 4.93, "elapsed_time": "0:33:41", "remaining_time": "10:49:03"}
+{"current_steps": 65, "total_steps": 1216, "loss": 0.1564, "accuracy": 0.9866666793823242, "learning_rate": 4.999852034151641e-07, "epoch": 0.10679079956188389, "percentage": 5.35, "elapsed_time": "0:36:29", "remaining_time": "10:46:04"}
+{"current_steps": 70, "total_steps": 1216, "loss": 0.1336, "accuracy": 0.9466667175292969, "learning_rate": 4.999250952911133e-07, "epoch": 0.11500547645125958, "percentage": 5.76, "elapsed_time": "0:39:17", "remaining_time": "10:43:07"}
+{"current_steps": 75, "total_steps": 1216, "loss": 0.1269, "accuracy": 0.9733333587646484, "learning_rate": 4.998187619501184e-07, "epoch": 0.12322015334063527, "percentage": 6.17, "elapsed_time": "0:42:04", "remaining_time": "10:40:09"}
+{"current_steps": 80, "total_steps": 1216, "loss": 0.1336, "accuracy": 0.9200000762939453, "learning_rate": 4.996662230591989e-07, "epoch": 0.13143483023001096, "percentage": 6.58, "elapsed_time": "0:44:52", "remaining_time": "10:37:16"}
+{"current_steps": 85, "total_steps": 1216, "loss": 0.1122, "accuracy": 0.9200000762939453, "learning_rate": 4.994675068313813e-07, "epoch": 0.13964950711938665, "percentage": 6.99, "elapsed_time": "0:47:40", "remaining_time": "10:34:19"}
+{"current_steps": 90, "total_steps": 1216, "loss": 0.1182, "accuracy": 0.9466666579246521, "learning_rate": 4.992226500204806e-07, "epoch": 0.14786418400876233, "percentage": 7.4, "elapsed_time": "0:50:28", "remaining_time": "10:31:32"}
+{"current_steps": 95, "total_steps": 1216, "loss": 0.1667, "accuracy": 0.9733333587646484, "learning_rate": 4.989316979143029e-07, "epoch": 0.156078860898138, "percentage": 7.81, "elapsed_time": "0:53:16", "remaining_time": "10:28:38"}
+{"current_steps": 100, "total_steps": 1216, "loss": 0.1454, "accuracy": 0.9333333373069763, "learning_rate": 4.985947043262686e-07, "epoch": 0.16429353778751368, "percentage": 8.22, "elapsed_time": "0:56:04", "remaining_time": "10:25:48"}
+{"current_steps": 105, "total_steps": 1216, "loss": 0.1293, "accuracy": 0.9733333587646484, "learning_rate": 4.982117315854593e-07, "epoch": 0.17250821467688937, "percentage": 8.63, "elapsed_time": "0:58:52", "remaining_time": "10:22:54"}
+{"current_steps": 110, "total_steps": 1216, "loss": 0.1359, "accuracy": 0.9466666579246521, "learning_rate": 4.977828505250903e-07, "epoch": 0.18072289156626506, "percentage": 9.05, "elapsed_time": "1:01:40", "remaining_time": "10:20:02"}
+{"current_steps": 115, "total_steps": 1216, "loss": 0.1152, "accuracy": 0.9599999785423279, "learning_rate": 4.973081404694087e-07, "epoch": 0.18893756845564075, "percentage": 9.46, "elapsed_time": "1:04:27", "remaining_time": "10:17:08"}
+{"current_steps": 120, "total_steps": 1216, "loss": 0.1214, "accuracy": 0.9466667175292969, "learning_rate": 4.967876892190227e-07, "epoch": 0.19715224534501644, "percentage": 9.87, "elapsed_time": "1:07:15", "remaining_time": "10:14:19"}
+{"current_steps": 125, "total_steps": 1216, "loss": 0.1007, "accuracy": 0.9733333587646484, "learning_rate": 4.962215930346614e-07, "epoch": 0.20536692223439212, "percentage": 10.28, "elapsed_time": "1:10:04", "remaining_time": "10:11:35"}
+{"current_steps": 130, "total_steps": 1216, "loss": 0.0878, "accuracy": 1.0, "learning_rate": 4.956099566193716e-07, "epoch": 0.21358159912376778, "percentage": 10.69, "elapsed_time": "1:12:52", "remaining_time": "10:08:47"}
+{"current_steps": 135, "total_steps": 1216, "loss": 0.1162, "accuracy": 0.9600000381469727, "learning_rate": 4.949528930991521e-07, "epoch": 0.22179627601314347, "percentage": 11.1, "elapsed_time": "1:15:40", "remaining_time": "10:05:55"}
+{"current_steps": 140, "total_steps": 1216, "loss": 0.1138, "accuracy": 0.9600000381469727, "learning_rate": 4.9425052400203e-07, "epoch": 0.23001095290251916, "percentage": 11.51, "elapsed_time": "1:18:28", "remaining_time": "10:03:09"}
+{"current_steps": 145, "total_steps": 1216, "loss": 0.0997, "accuracy": 0.9599999785423279, "learning_rate": 4.935029792355834e-07, "epoch": 0.23822562979189485, "percentage": 11.92, "elapsed_time": "1:21:16", "remaining_time": "10:00:18"}
+{"current_steps": 150, "total_steps": 1216, "loss": 0.1249, "accuracy": 0.9600000381469727, "learning_rate": 4.927103970629147e-07, "epoch": 0.24644030668127054, "percentage": 12.34, "elapsed_time": "1:24:04", "remaining_time": "9:57:28"}
+{"current_steps": 155, "total_steps": 1216, "loss": 0.1166, "accuracy": 0.9600000381469727, "learning_rate": 4.918729240770775e-07, "epoch": 0.2546549835706462, "percentage": 12.75, "elapsed_time": "1:26:52", "remaining_time": "9:54:40"}
+{"current_steps": 160, "total_steps": 1216, "loss": 0.1238, "accuracy": 0.9466667175292969, "learning_rate": 4.909907151739633e-07, "epoch": 0.2628696604600219, "percentage": 13.16, "elapsed_time": "1:29:40", "remaining_time": "9:51:49"}
+{"current_steps": 165, "total_steps": 1216, "loss": 0.1276, "accuracy": 0.9733333587646484, "learning_rate": 4.900639335236526e-07, "epoch": 0.2710843373493976, "percentage": 13.57, "elapsed_time": "1:32:27", "remaining_time": "9:48:57"}
+{"current_steps": 170, "total_steps": 1216, "loss": 0.1045, "accuracy": 0.9600000381469727, "learning_rate": 4.890927505402359e-07, "epoch": 0.2792990142387733, "percentage": 13.98, "elapsed_time": "1:35:15", "remaining_time": "9:46:07"}
+{"current_steps": 175, "total_steps": 1216, "loss": 0.0838, "accuracy": 0.9600000381469727, "learning_rate": 4.880773458501089e-07, "epoch": 0.28751369112814895, "percentage": 14.39, "elapsed_time": "1:38:03", "remaining_time": "9:43:20"}
+{"current_steps": 180, "total_steps": 1216, "loss": 0.1141, "accuracy": 0.9333333373069763, "learning_rate": 4.870179072587498e-07, "epoch": 0.29572836801752467, "percentage": 14.8, "elapsed_time": "1:40:51", "remaining_time": "9:40:29"}
+{"current_steps": 185, "total_steps": 1216, "loss": 0.0915, "accuracy": 0.9866666793823242, "learning_rate": 4.859146307159841e-07, "epoch": 0.30394304490690033, "percentage": 15.21, "elapsed_time": "1:43:39", "remaining_time": "9:37:40"}
+{"current_steps": 190, "total_steps": 1216, "loss": 0.1261, "accuracy": 0.9599999785423279, "learning_rate": 4.847677202797414e-07, "epoch": 0.312157721796276, "percentage": 15.62, "elapsed_time": "1:46:27", "remaining_time": "9:34:50"}
+{"current_steps": 195, "total_steps": 1216, "loss": 0.0996, "accuracy": 0.9600000381469727, "learning_rate": 4.835773880783144e-07, "epoch": 0.3203723986856517, "percentage": 16.04, "elapsed_time": "1:49:15", "remaining_time": "9:32:02"}
+{"current_steps": 200, "total_steps": 1216, "loss": 0.0875, "accuracy": 0.9466667175292969, "learning_rate": 4.823438542711238e-07, "epoch": 0.32858707557502737, "percentage": 16.45, "elapsed_time": "1:52:03", "remaining_time": "9:29:13"}
+{"current_steps": 205, "total_steps": 1216, "loss": 0.1094, "accuracy": 0.9333333969116211, "learning_rate": 4.81067347007999e-07, "epoch": 0.3368017524644031, "percentage": 16.86, "elapsed_time": "1:54:50", "remaining_time": "9:26:23"}
+{"current_steps": 210, "total_steps": 1216, "loss": 0.0868, "accuracy": 0.9866666793823242, "learning_rate": 4.797481023869801e-07, "epoch": 0.34501642935377874, "percentage": 17.27, "elapsed_time": "1:57:38", "remaining_time": "9:23:32"}
+{"current_steps": 215, "total_steps": 1216, "loss": 0.084, "accuracy": 0.9866666793823242, "learning_rate": 4.783863644106502e-07, "epoch": 0.35323110624315446, "percentage": 17.68, "elapsed_time": "2:00:26", "remaining_time": "9:20:44"}
+{"current_steps": 220, "total_steps": 1216, "loss": 0.0717, "accuracy": 1.0, "learning_rate": 4.769823849410053e-07, "epoch": 0.3614457831325301, "percentage": 18.09, "elapsed_time": "2:03:14", "remaining_time": "9:17:57"}
+{"current_steps": 225, "total_steps": 1216, "loss": 0.1065, "accuracy": 0.9866666793823242, "learning_rate": 4.7553642365287127e-07, "epoch": 0.3696604600219058, "percentage": 18.5, "elapsed_time": "2:06:02", "remaining_time": "9:15:07"}
+{"current_steps": 230, "total_steps": 1216, "loss": 0.0947, "accuracy": 0.9733333587646484, "learning_rate": 4.7404874798587493e-07, "epoch": 0.3778751369112815, "percentage": 18.91, "elapsed_time": "2:08:50", "remaining_time": "9:12:19"}
+{"current_steps": 235, "total_steps": 1216, "loss": 0.1189, "accuracy": 0.9599999785423279, "learning_rate": 4.7251963309497965e-07, "epoch": 0.38608981380065716, "percentage": 19.33, "elapsed_time": "2:11:38", "remaining_time": "9:09:30"}
+{"current_steps": 240, "total_steps": 1216, "loss": 0.0903, "accuracy": 0.9600000381469727, "learning_rate": 4.709493617995938e-07, "epoch": 0.39430449069003287, "percentage": 19.74, "elapsed_time": "2:14:26", "remaining_time": "9:06:42"}
+{"current_steps": 245, "total_steps": 1216, "loss": 0.1183, "accuracy": 0.9466667175292969, "learning_rate": 4.6933822453126114e-07, "epoch": 0.40251916757940853, "percentage": 20.15, "elapsed_time": "2:17:14", "remaining_time": "9:03:53"}
+{"current_steps": 250, "total_steps": 1216, "loss": 0.0704, "accuracy": 0.9600000381469727, "learning_rate": 4.676865192799443e-07, "epoch": 0.41073384446878425, "percentage": 20.56, "elapsed_time": "2:20:01", "remaining_time": "9:01:03"}
+{"current_steps": 255, "total_steps": 1216, "loss": 0.1042, "accuracy": 0.9733333587646484, "learning_rate": 4.65994551538909e-07, "epoch": 0.4189485213581599, "percentage": 20.97, "elapsed_time": "2:22:49", "remaining_time": "8:58:14"}
+{"current_steps": 260, "total_steps": 1216, "loss": 0.1123, "accuracy": 0.9466666579246521, "learning_rate": 4.642626342482215e-07, "epoch": 0.42716319824753557, "percentage": 21.38, "elapsed_time": "2:25:37", "remaining_time": "8:55:25"}
+{"current_steps": 265, "total_steps": 1216, "loss": 0.0943, "accuracy": 0.9600000381469727, "learning_rate": 4.624910877368684e-07, "epoch": 0.4353778751369113, "percentage": 21.79, "elapsed_time": "2:28:25", "remaining_time": "8:52:38"}
+{"current_steps": 270, "total_steps": 1216, "loss": 0.0909, "accuracy": 0.9733333587646484, "learning_rate": 4.606802396635098e-07, "epoch": 0.44359255202628695, "percentage": 22.2, "elapsed_time": "2:31:13", "remaining_time": "8:49:50"}
+{"current_steps": 275, "total_steps": 1216, "loss": 0.0999, "accuracy": 0.9466666579246521, "learning_rate": 4.588304249558763e-07, "epoch": 0.45180722891566266, "percentage": 22.62, "elapsed_time": "2:34:01", "remaining_time": "8:47:02"}
+{"current_steps": 280, "total_steps": 1216, "loss": 0.1019, "accuracy": 0.9466667175292969, "learning_rate": 4.569419857488228e-07, "epoch": 0.4600219058050383, "percentage": 23.03, "elapsed_time": "2:36:49", "remaining_time": "8:44:13"}
+{"current_steps": 285, "total_steps": 1216, "loss": 0.0864, "accuracy": 0.9466667175292969, "learning_rate": 4.550152713210478e-07, "epoch": 0.46823658269441404, "percentage": 23.44, "elapsed_time": "2:39:37", "remaining_time": "8:41:25"}
+{"current_steps": 290, "total_steps": 1216, "loss": 0.0889, "accuracy": 0.9733333587646484, "learning_rate": 4.530506380304925e-07, "epoch": 0.4764512595837897, "percentage": 23.85, "elapsed_time": "2:42:24", "remaining_time": "8:38:36"}
+{"current_steps": 295, "total_steps": 1216, "loss": 0.091, "accuracy": 1.0, "learning_rate": 4.510484492484301e-07, "epoch": 0.4846659364731654, "percentage": 24.26, "elapsed_time": "2:45:13", "remaining_time": "8:35:49"}
+{"current_steps": 300, "total_steps": 1216, "loss": 0.1166, "accuracy": 0.9600000381469727, "learning_rate": 4.4900907529225797e-07, "epoch": 0.4928806133625411, "percentage": 24.67, "elapsed_time": "2:48:01", "remaining_time": "8:33:01"}
+{"current_steps": 305, "total_steps": 1216, "loss": 0.0689, "accuracy": 1.0, "learning_rate": 4.46932893357005e-07, "epoch": 0.5010952902519168, "percentage": 25.08, "elapsed_time": "2:50:49", "remaining_time": "8:30:13"}
+{"current_steps": 310, "total_steps": 1216, "loss": 0.1225, "accuracy": 0.9333333969116211, "learning_rate": 4.448202874455672e-07, "epoch": 0.5093099671412924, "percentage": 25.49, "elapsed_time": "2:53:37", "remaining_time": "8:27:24"}
+{"current_steps": 315, "total_steps": 1216, "loss": 0.0913, "accuracy": 0.9600000381469727, "learning_rate": 4.426716482976838e-07, "epoch": 0.5175246440306681, "percentage": 25.9, "elapsed_time": "2:56:25", "remaining_time": "8:24:36"}
+{"current_steps": 320, "total_steps": 1216, "loss": 0.1113, "accuracy": 0.9866666793823242, "learning_rate": 4.4048737331766774e-07, "epoch": 0.5257393209200438, "percentage": 26.32, "elapsed_time": "2:59:12", "remaining_time": "8:21:48"}
+{"current_steps": 325, "total_steps": 1216, "loss": 0.0909, "accuracy": 0.9333333373069763, "learning_rate": 4.3826786650090273e-07, "epoch": 0.5339539978094195, "percentage": 26.73, "elapsed_time": "3:02:00", "remaining_time": "8:18:59"}
+{"current_steps": 330, "total_steps": 1216, "loss": 0.1393, "accuracy": 0.9066666960716248, "learning_rate": 4.3601353835912235e-07, "epoch": 0.5421686746987951, "percentage": 27.14, "elapsed_time": "3:04:49", "remaining_time": "8:16:12"}
+{"current_steps": 335, "total_steps": 1216, "loss": 0.0884, "accuracy": 0.9866666793823242, "learning_rate": 4.337248058444831e-07, "epoch": 0.5503833515881709, "percentage": 27.55, "elapsed_time": "3:07:37", "remaining_time": "8:13:25"}
+{"current_steps": 340, "total_steps": 1216, "loss": 0.1089, "accuracy": 0.9066667556762695, "learning_rate": 4.3140209227244617e-07, "epoch": 0.5585980284775466, "percentage": 27.96, "elapsed_time": "3:10:25", "remaining_time": "8:10:36"}
+{"current_steps": 345, "total_steps": 1216, "loss": 0.093, "accuracy": 0.9733333587646484, "learning_rate": 4.2904582724348316e-07, "epoch": 0.5668127053669222, "percentage": 28.37, "elapsed_time": "3:13:13", "remaining_time": "8:07:48"}
+{"current_steps": 350, "total_steps": 1216, "loss": 0.0859, "accuracy": 0.9733333587646484, "learning_rate": 4.266564465636182e-07, "epoch": 0.5750273822562979, "percentage": 28.78, "elapsed_time": "3:16:01", "remaining_time": "8:05:01"}
+{"current_steps": 355, "total_steps": 1216, "loss": 0.0601, "accuracy": 1.0, "learning_rate": 4.242343921638234e-07, "epoch": 0.5832420591456736, "percentage": 29.19, "elapsed_time": "3:18:49", "remaining_time": "8:02:13"}
+{"current_steps": 360, "total_steps": 1216, "loss": 0.0879, "accuracy": 0.9466667175292969, "learning_rate": 4.2178011201828044e-07, "epoch": 0.5914567360350493, "percentage": 29.61, "elapsed_time": "3:21:40", "remaining_time": "7:59:32"}
+{"current_steps": 365, "total_steps": 1216, "loss": 0.104, "accuracy": 0.9600000381469727, "learning_rate": 4.1929406006152546e-07, "epoch": 0.5996714129244249, "percentage": 30.02, "elapsed_time": "3:24:29", "remaining_time": "7:56:45"}
+{"current_steps": 370, "total_steps": 1216, "loss": 0.0842, "accuracy": 1.0, "learning_rate": 4.167766961044906e-07, "epoch": 0.6078860898138007, "percentage": 30.43, "elapsed_time": "3:27:17", "remaining_time": "7:53:59"}
+{"current_steps": 375, "total_steps": 1216, "loss": 0.086, "accuracy": 0.9733333587646484, "learning_rate": 4.1422848574945923e-07, "epoch": 0.6161007667031764, "percentage": 30.84, "elapsed_time": "3:30:07", "remaining_time": "7:51:14"}
+{"current_steps": 380, "total_steps": 1216, "loss": 0.1158, "accuracy": 0.9600000381469727, "learning_rate": 4.1164990030394985e-07, "epoch": 0.624315443592552, "percentage": 31.25, "elapsed_time": "3:32:57", "remaining_time": "7:48:31"}
+{"current_steps": 385, "total_steps": 1216, "loss": 0.1037, "accuracy": 0.9600000381469727, "learning_rate": 4.09041416693545e-07, "epoch": 0.6325301204819277, "percentage": 31.66, "elapsed_time": "3:35:47", "remaining_time": "7:45:45"}
+{"current_steps": 390, "total_steps": 1216, "loss": 0.098, "accuracy": 0.9600000381469727, "learning_rate": 4.064035173736804e-07, "epoch": 0.6407447973713034, "percentage": 32.07, "elapsed_time": "3:38:36", "remaining_time": "7:42:59"}
+{"current_steps": 395, "total_steps": 1216, "loss": 0.076, "accuracy": 0.9866666793823242, "learning_rate": 4.0373669024041225e-07, "epoch": 0.6489594742606791, "percentage": 32.48, "elapsed_time": "3:41:24", "remaining_time": "7:40:12"}
+{"current_steps": 400, "total_steps": 1216, "loss": 0.0779, "accuracy": 0.9733333587646484, "learning_rate": 4.010414285401776e-07, "epoch": 0.6571741511500547, "percentage": 32.89, "elapsed_time": "3:44:15", "remaining_time": "7:37:29"}
+{"current_steps": 405, "total_steps": 1216, "loss": 0.1026, "accuracy": 0.9466667175292969, "learning_rate": 3.9831823077856565e-07, "epoch": 0.6653888280394304, "percentage": 33.31, "elapsed_time": "3:47:04", "remaining_time": "7:34:42"}
+{"current_steps": 410, "total_steps": 1216, "loss": 0.0833, "accuracy": 0.9733333587646484, "learning_rate": 3.95567600628115e-07, "epoch": 0.6736035049288062, "percentage": 33.72, "elapsed_time": "3:49:52", "remaining_time": "7:31:55"}
+{"current_steps": 415, "total_steps": 1216, "loss": 0.0921, "accuracy": 1.0, "learning_rate": 3.9279004683515783e-07, "epoch": 0.6818181818181818, "percentage": 34.13, "elapsed_time": "3:52:43", "remaining_time": "7:29:10"}
+{"current_steps": 420, "total_steps": 1216, "loss": 0.0666, "accuracy": 0.9600000381469727, "learning_rate": 3.8998608312572234e-07, "epoch": 0.6900328587075575, "percentage": 34.54, "elapsed_time": "3:55:31", "remaining_time": "7:26:23"}
+{"current_steps": 425, "total_steps": 1216, "loss": 0.0824, "accuracy": 0.9733333587646484, "learning_rate": 3.8715622811051753e-07, "epoch": 0.6982475355969332, "percentage": 34.95, "elapsed_time": "3:58:20", "remaining_time": "7:23:35"}
+{"current_steps": 430, "total_steps": 1216, "loss": 0.0968, "accuracy": 0.9466666579246521, "learning_rate": 3.843010051890114e-07, "epoch": 0.7064622124863089, "percentage": 35.36, "elapsed_time": "4:01:11", "remaining_time": "7:20:53"}
+{"current_steps": 435, "total_steps": 1216, "loss": 0.0907, "accuracy": 0.9866666793823242, "learning_rate": 3.8142094245262615e-07, "epoch": 0.7146768893756845, "percentage": 35.77, "elapsed_time": "4:04:00", "remaining_time": "7:18:05"}
+{"current_steps": 440, "total_steps": 1216, "loss": 0.0896, "accuracy": 0.9600000381469727, "learning_rate": 3.785165725870637e-07, "epoch": 0.7228915662650602, "percentage": 36.18, "elapsed_time": "4:06:50", "remaining_time": "7:15:20"}
+{"current_steps": 445, "total_steps": 1216, "loss": 0.089, "accuracy": 0.9733333587646484, "learning_rate": 3.7558843277378203e-07, "epoch": 0.731106243154436, "percentage": 36.6, "elapsed_time": "4:09:39", "remaining_time": "7:12:33"}
+{"current_steps": 450, "total_steps": 1216, "loss": 0.072, "accuracy": 0.9866666793823242, "learning_rate": 3.726370645906407e-07, "epoch": 0.7393209200438116, "percentage": 37.01, "elapsed_time": "4:12:28", "remaining_time": "7:09:45"}
+{"current_steps": 455, "total_steps": 1216, "loss": 0.0961, "accuracy": 0.9600000381469727, "learning_rate": 3.6966301391173204e-07, "epoch": 0.7475355969331873, "percentage": 37.42, "elapsed_time": "4:15:16", "remaining_time": "7:06:57"}
+{"current_steps": 460, "total_steps": 1216, "loss": 0.0663, "accuracy": 0.9866666793823242, "learning_rate": 3.6666683080641843e-07, "epoch": 0.755750273822563, "percentage": 37.83, "elapsed_time": "4:18:06", "remaining_time": "7:04:10"}
+{"current_steps": 465, "total_steps": 1216, "loss": 0.0698, "accuracy": 0.9733333587646484, "learning_rate": 3.636490694375937e-07, "epoch": 0.7639649507119387, "percentage": 38.24, "elapsed_time": "4:20:57", "remaining_time": "7:01:27"}
+{"current_steps": 470, "total_steps": 1216, "loss": 0.0643, "accuracy": 0.9733333587646484, "learning_rate": 3.6061028795918734e-07, "epoch": 0.7721796276013143, "percentage": 38.65, "elapsed_time": "4:23:45", "remaining_time": "6:58:39"}
+{"current_steps": 475, "total_steps": 1216, "loss": 0.0753, "accuracy": 0.9866666793823242, "learning_rate": 3.5755104841292974e-07, "epoch": 0.78039430449069, "percentage": 39.06, "elapsed_time": "4:26:34", "remaining_time": "6:55:51"}
+{"current_steps": 480, "total_steps": 1216, "loss": 0.0675, "accuracy": 0.9733333587646484, "learning_rate": 3.544719166243998e-07, "epoch": 0.7886089813800657, "percentage": 39.47, "elapsed_time": "4:29:24", "remaining_time": "6:53:05"}
+{"current_steps": 485, "total_steps": 1216, "loss": 0.0585, "accuracy": 0.9866666793823242, "learning_rate": 3.513734620983716e-07, "epoch": 0.7968236582694413, "percentage": 39.88, "elapsed_time": "4:32:13", "remaining_time": "6:50:18"}
+{"current_steps": 490, "total_steps": 1216, "loss": 0.0883, "accuracy": 0.9200000166893005, "learning_rate": 3.482562579134809e-07, "epoch": 0.8050383351588171, "percentage": 40.3, "elapsed_time": "4:35:03", "remaining_time": "6:47:31"}
+{"current_steps": 495, "total_steps": 1216, "loss": 0.0795, "accuracy": 0.9466666579246521, "learning_rate": 3.4512088061623073e-07, "epoch": 0.8132530120481928, "percentage": 40.71, "elapsed_time": "4:37:52", "remaining_time": "6:44:44"}
+{"current_steps": 500, "total_steps": 1216, "loss": 0.0646, "accuracy": 0.9599999785423279, "learning_rate": 3.419679101143555e-07, "epoch": 0.8214676889375685, "percentage": 41.12, "elapsed_time": "4:40:42", "remaining_time": "6:41:58"}
+{"current_steps": 505, "total_steps": 1216, "loss": 0.0854, "accuracy": 0.9733333587646484, "learning_rate": 3.387979295695632e-07, "epoch": 0.8296823658269441, "percentage": 41.53, "elapsed_time": "4:43:30", "remaining_time": "6:39:09"}
+{"current_steps": 510, "total_steps": 1216, "loss": 0.084, "accuracy": 0.9600000381469727, "learning_rate": 3.356115252896764e-07, "epoch": 0.8378970427163198, "percentage": 41.94, "elapsed_time": "4:46:18", "remaining_time": "6:36:20"}
+{"current_steps": 515, "total_steps": 1216, "loss": 0.076, "accuracy": 0.9466667175292969, "learning_rate": 3.3240928662019043e-07, "epoch": 0.8461117196056955, "percentage": 42.35, "elapsed_time": "4:49:06", "remaining_time": "6:33:31"}
+{"current_steps": 520, "total_steps": 1216, "loss": 0.1012, "accuracy": 0.9733333587646484, "learning_rate": 3.291918058352706e-07, "epoch": 0.8543263964950711, "percentage": 42.76, "elapsed_time": "4:51:54", "remaining_time": "6:30:42"}
+{"current_steps": 525, "total_steps": 1216, "loss": 0.0634, "accuracy": 1.0, "learning_rate": 3.259596780282074e-07, "epoch": 0.8625410733844469, "percentage": 43.17, "elapsed_time": "4:54:42", "remaining_time": "6:27:53"}
+{"current_steps": 530, "total_steps": 1216, "loss": 0.0671, "accuracy": 0.9866666793823242, "learning_rate": 3.2271350100134975e-07, "epoch": 0.8707557502738226, "percentage": 43.59, "elapsed_time": "4:57:30", "remaining_time": "6:25:04"}
+{"current_steps": 535, "total_steps": 1216, "loss": 0.0816, "accuracy": 1.0, "learning_rate": 3.1945387515553843e-07, "epoch": 0.8789704271631983, "percentage": 44.0, "elapsed_time": "5:00:18", "remaining_time": "6:22:15"}
+{"current_steps": 540, "total_steps": 1216, "loss": 0.0785, "accuracy": 0.9733333587646484, "learning_rate": 3.1618140337905764e-07, "epoch": 0.8871851040525739, "percentage": 44.41, "elapsed_time": "5:03:06", "remaining_time": "6:19:26"}
+{"current_steps": 545, "total_steps": 1216, "loss": 0.0577, "accuracy": 0.9600000381469727, "learning_rate": 3.128966909361271e-07, "epoch": 0.8953997809419496, "percentage": 44.82, "elapsed_time": "5:05:54", "remaining_time": "6:16:38"}
+{"current_steps": 550, "total_steps": 1216, "loss": 0.0805, "accuracy": 0.9600000381469727, "learning_rate": 3.096003453549549e-07, "epoch": 0.9036144578313253, "percentage": 45.23, "elapsed_time": "5:08:42", "remaining_time": "6:13:48"}
+{"current_steps": 555, "total_steps": 1216, "loss": 0.0696, "accuracy": 0.9866666793823242, "learning_rate": 3.06292976315371e-07, "epoch": 0.911829134720701, "percentage": 45.64, "elapsed_time": "5:11:30", "remaining_time": "6:11:00"}
+{"current_steps": 560, "total_steps": 1216, "loss": 0.0879, "accuracy": 0.9600000381469727, "learning_rate": 3.0297519553606324e-07, "epoch": 0.9200438116100766, "percentage": 46.05, "elapsed_time": "5:14:18", "remaining_time": "6:08:11"}
+{"current_steps": 565, "total_steps": 1216, "loss": 0.0885, "accuracy": 0.9600000381469727, "learning_rate": 2.996476166614363e-07, "epoch": 0.9282584884994524, "percentage": 46.46, "elapsed_time": "5:17:05", "remaining_time": "6:05:21"}
+{"current_steps": 570, "total_steps": 1216, "loss": 0.0877, "accuracy": 0.9733333587646484, "learning_rate": 2.963108551481142e-07, "epoch": 0.9364731653888281, "percentage": 46.88, "elapsed_time": "5:19:53", "remaining_time": "6:02:33"}
+{"current_steps": 575, "total_steps": 1216, "loss": 0.0648, "accuracy": 0.9733333587646484, "learning_rate": 2.929655281511075e-07, "epoch": 0.9446878422782037, "percentage": 47.29, "elapsed_time": "5:22:41", "remaining_time": "5:59:44"}
+{"current_steps": 580, "total_steps": 1216, "loss": 0.0808, "accuracy": 0.9733333587646484, "learning_rate": 2.896122544096667e-07, "epoch": 0.9529025191675794, "percentage": 47.7, "elapsed_time": "5:25:29", "remaining_time": "5:56:54"}
+{"current_steps": 585, "total_steps": 1216, "loss": 0.0951, "accuracy": 0.9466667175292969, "learning_rate": 2.8625165413284307e-07, "epoch": 0.9611171960569551, "percentage": 48.11, "elapsed_time": "5:28:17", "remaining_time": "5:54:06"}
+{"current_steps": 590, "total_steps": 1216, "loss": 0.0728, "accuracy": 0.9866666793823242, "learning_rate": 2.8288434888477626e-07, "epoch": 0.9693318729463308, "percentage": 48.52, "elapsed_time": "5:31:05", "remaining_time": "5:51:17"}
+{"current_steps": 595, "total_steps": 1216, "loss": 0.0873, "accuracy": 0.9599999785423279, "learning_rate": 2.795109614697326e-07, "epoch": 0.9775465498357064, "percentage": 48.93, "elapsed_time": "5:33:53", "remaining_time": "5:48:28"}
+{"current_steps": 600, "total_steps": 1216, "loss": 0.0896, "accuracy": 0.9866666793823242, "learning_rate": 2.761321158169134e-07, "epoch": 0.9857612267250822, "percentage": 49.34, "elapsed_time": "5:36:42", "remaining_time": "5:45:41"}
+{"current_steps": 605, "total_steps": 1216, "loss": 0.0826, "accuracy": 0.9599999785423279, "learning_rate": 2.727484368650553e-07, "epoch": 0.9939759036144579, "percentage": 49.75, "elapsed_time": "5:39:30", "remaining_time": "5:42:53"}
+{"current_steps": 610, "total_steps": 1216, "loss": 0.0613, "accuracy": 1.0, "learning_rate": 2.6936055044684425e-07, "epoch": 1.0021905805038336, "percentage": 50.16, "elapsed_time": "5:42:51", "remaining_time": "5:40:36"}
+{"current_steps": 615, "total_steps": 1216, "loss": 0.0479, "accuracy": 0.9600000381469727, "learning_rate": 2.659690831731631e-07, "epoch": 1.0104052573932092, "percentage": 50.58, "elapsed_time": "5:45:39", "remaining_time": "5:37:47"}
+{"current_steps": 620, "total_steps": 1216, "loss": 0.0375, "accuracy": 0.9866666793823242, "learning_rate": 2.6257466231719676e-07, "epoch": 1.0186199342825848, "percentage": 50.99, "elapsed_time": "5:48:29", "remaining_time": "5:35:00"}
+{"current_steps": 625, "total_steps": 1216, "loss": 0.0655, "accuracy": 0.9599999785423279, "learning_rate": 2.591779156984137e-07, "epoch": 1.0268346111719606, "percentage": 51.4, "elapsed_time": "5:51:19", "remaining_time": "5:32:12"}
+{"current_steps": 630, "total_steps": 1216, "loss": 0.0574, "accuracy": 0.9866666793823242, "learning_rate": 2.557794715664465e-07, "epoch": 1.0350492880613362, "percentage": 51.81, "elapsed_time": "5:54:07", "remaining_time": "5:29:23"}
+{"current_steps": 635, "total_steps": 1216, "loss": 0.0736, "accuracy": 0.9733333587646484, "learning_rate": 2.5237995848489417e-07, "epoch": 1.0432639649507118, "percentage": 52.22, "elapsed_time": "5:56:57", "remaining_time": "5:26:35"}
+{"current_steps": 640, "total_steps": 1216, "loss": 0.0747, "accuracy": 0.9733333587646484, "learning_rate": 2.48980005215064e-07, "epoch": 1.0514786418400877, "percentage": 52.63, "elapsed_time": "5:59:45", "remaining_time": "5:23:47"}
+{"current_steps": 645, "total_steps": 1216, "loss": 0.073, "accuracy": 0.9733333587646484, "learning_rate": 2.45580240599679e-07, "epoch": 1.0596933187294633, "percentage": 53.04, "elapsed_time": "6:02:34", "remaining_time": "5:20:58"}
+{"current_steps": 650, "total_steps": 1216, "loss": 0.0526, "accuracy": 0.9866666793823242, "learning_rate": 2.421812934465696e-07, "epoch": 1.067907995618839, "percentage": 53.45, "elapsed_time": "6:05:25", "remaining_time": "5:18:11"}
+{"current_steps": 655, "total_steps": 1216, "loss": 0.0702, "accuracy": 0.9600000381469727, "learning_rate": 2.3878379241237134e-07, "epoch": 1.0761226725082147, "percentage": 53.87, "elapsed_time": "6:08:14", "remaining_time": "5:15:23"}
+{"current_steps": 660, "total_steps": 1216, "loss": 0.0709, "accuracy": 0.9466667175292969, "learning_rate": 2.3538836588625077e-07, "epoch": 1.0843373493975903, "percentage": 54.28, "elapsed_time": "6:11:03", "remaining_time": "5:12:35"}
+{"current_steps": 665, "total_steps": 1216, "loss": 0.0699, "accuracy": 0.9866666793823242, "learning_rate": 2.3199564187368153e-07, "epoch": 1.0925520262869661, "percentage": 54.69, "elapsed_time": "6:13:52", "remaining_time": "5:09:47"}
+{"current_steps": 670, "total_steps": 1216, "loss": 0.0756, "accuracy": 0.9599999785423279, "learning_rate": 2.2860624788029013e-07, "epoch": 1.1007667031763417, "percentage": 55.1, "elapsed_time": "6:16:40", "remaining_time": "5:06:58"}
+{"current_steps": 675, "total_steps": 1216, "loss": 0.0701, "accuracy": 0.9466666579246521, "learning_rate": 2.2522081079579497e-07, "epoch": 1.1089813800657173, "percentage": 55.51, "elapsed_time": "6:19:30", "remaining_time": "5:04:09"}
+{"current_steps": 680, "total_steps": 1216, "loss": 0.0499, "accuracy": 1.0, "learning_rate": 2.2183995677805967e-07, "epoch": 1.1171960569550932, "percentage": 55.92, "elapsed_time": "6:22:18", "remaining_time": "5:01:21"}
+{"current_steps": 685, "total_steps": 1216, "loss": 0.0621, "accuracy": 0.9866666793823242, "learning_rate": 2.1846431113728062e-07, "epoch": 1.1254107338444688, "percentage": 56.33, "elapsed_time": "6:25:07", "remaining_time": "4:58:32"}
+{"current_steps": 690, "total_steps": 1216, "loss": 0.0415, "accuracy": 0.9866666793823242, "learning_rate": 2.1509449822033205e-07, "epoch": 1.1336254107338444, "percentage": 56.74, "elapsed_time": "6:27:56", "remaining_time": "4:55:44"}
+{"current_steps": 695, "total_steps": 1216, "loss": 0.0795, "accuracy": 0.9599999785423279, "learning_rate": 2.1173114129528957e-07, "epoch": 1.1418400876232202, "percentage": 57.15, "elapsed_time": "6:30:45", "remaining_time": "4:52:55"}
+{"current_steps": 700, "total_steps": 1216, "loss": 0.0802, "accuracy": 0.9733333587646484, "learning_rate": 2.0837486243615226e-07, "epoch": 1.1500547645125958, "percentage": 57.57, "elapsed_time": "6:33:34", "remaining_time": "4:50:07"}
+{"current_steps": 705, "total_steps": 1216, "loss": 0.073, "accuracy": 0.9866666793823242, "learning_rate": 2.0502628240778653e-07, "epoch": 1.1582694414019716, "percentage": 57.98, "elapsed_time": "6:36:23", "remaining_time": "4:47:18"}
+{"current_steps": 710, "total_steps": 1216, "loss": 0.043, "accuracy": 1.0, "learning_rate": 2.0168602055111173e-07, "epoch": 1.1664841182913472, "percentage": 58.39, "elapsed_time": "6:39:12", "remaining_time": "4:44:30"}
+{"current_steps": 715, "total_steps": 1216, "loss": 0.0523, "accuracy": 0.9600000381469727, "learning_rate": 1.9835469466854887e-07, "epoch": 1.1746987951807228, "percentage": 58.8, "elapsed_time": "6:42:03", "remaining_time": "4:41:43"}
+{"current_steps": 720, "total_steps": 1216, "loss": 0.069, "accuracy": 0.9466667175292969, "learning_rate": 1.9503292090975454e-07, "epoch": 1.1829134720700987, "percentage": 59.21, "elapsed_time": "6:44:51", "remaining_time": "4:38:54"}
+{"current_steps": 725, "total_steps": 1216, "loss": 0.0429, "accuracy": 0.9866666793823242, "learning_rate": 1.917213136576602e-07, "epoch": 1.1911281489594743, "percentage": 59.62, "elapsed_time": "6:47:40", "remaining_time": "4:36:05"}
+{"current_steps": 730, "total_steps": 1216, "loss": 0.0688, "accuracy": 0.9733333587646484, "learning_rate": 1.8842048541483756e-07, "epoch": 1.1993428258488499, "percentage": 60.03, "elapsed_time": "6:50:29", "remaining_time": "4:33:17"}
+{"current_steps": 735, "total_steps": 1216, "loss": 0.069, "accuracy": 0.9733333587646484, "learning_rate": 1.8513104669021314e-07, "epoch": 1.2075575027382257, "percentage": 60.44, "elapsed_time": "6:53:17", "remaining_time": "4:30:27"}
+{"current_steps": 740, "total_steps": 1216, "loss": 0.0678, "accuracy": 0.9600000381469727, "learning_rate": 1.8185360588615057e-07, "epoch": 1.2157721796276013, "percentage": 60.86, "elapsed_time": "6:56:05", "remaining_time": "4:27:38"}
+{"current_steps": 745, "total_steps": 1216, "loss": 0.0644, "accuracy": 0.9466667175292969, "learning_rate": 1.7858876918592232e-07, "epoch": 1.223986856516977, "percentage": 61.27, "elapsed_time": "6:58:52", "remaining_time": "4:24:49"}
+{"current_steps": 750, "total_steps": 1216, "loss": 0.0704, "accuracy": 0.9466667175292969, "learning_rate": 1.7533714044159299e-07, "epoch": 1.2322015334063527, "percentage": 61.68, "elapsed_time": "7:01:40", "remaining_time": "4:21:59"}
+{"current_steps": 755, "total_steps": 1216, "loss": 0.0531, "accuracy": 0.9733333587646484, "learning_rate": 1.7209932106233264e-07, "epoch": 1.2404162102957283, "percentage": 62.09, "elapsed_time": "7:04:27", "remaining_time": "4:19:10"}
+{"current_steps": 760, "total_steps": 1216, "loss": 0.0624, "accuracy": 0.9599999785423279, "learning_rate": 1.688759099031824e-07, "epoch": 1.248630887185104, "percentage": 62.5, "elapsed_time": "7:07:15", "remaining_time": "4:16:21"}
+{"current_steps": 765, "total_steps": 1216, "loss": 0.0407, "accuracy": 1.0, "learning_rate": 1.656675031542925e-07, "epoch": 1.2568455640744798, "percentage": 62.91, "elapsed_time": "7:10:03", "remaining_time": "4:13:32"}
+{"current_steps": 770, "total_steps": 1216, "loss": 0.0567, "accuracy": 0.9866666793823242, "learning_rate": 1.6247469423065343e-07, "epoch": 1.2650602409638554, "percentage": 63.32, "elapsed_time": "7:12:52", "remaining_time": "4:10:43"}
+{"current_steps": 775, "total_steps": 1216, "loss": 0.0526, "accuracy": 0.9866666793823242, "learning_rate": 1.5929807366233977e-07, "epoch": 1.273274917853231, "percentage": 63.73, "elapsed_time": "7:15:40", "remaining_time": "4:07:55"}
+{"current_steps": 780, "total_steps": 1216, "loss": 0.0508, "accuracy": 0.9600000381469727, "learning_rate": 1.5613822898528794e-07, "epoch": 1.2814895947426068, "percentage": 64.14, "elapsed_time": "7:18:28", "remaining_time": "4:05:06"}
+{"current_steps": 785, "total_steps": 1216, "loss": 0.0713, "accuracy": 0.9866666793823242, "learning_rate": 1.5299574463262794e-07, "epoch": 1.2897042716319824, "percentage": 64.56, "elapsed_time": "7:21:16", "remaining_time": "4:02:16"}
+{"current_steps": 790, "total_steps": 1216, "loss": 0.0521, "accuracy": 0.9600000381469727, "learning_rate": 1.4987120182658877e-07, "epoch": 1.297918948521358, "percentage": 64.97, "elapsed_time": "7:24:04", "remaining_time": "3:59:27"}
+{"current_steps": 795, "total_steps": 1216, "loss": 0.0717, "accuracy": 0.9733333587646484, "learning_rate": 1.4676517847099745e-07, "epoch": 1.3061336254107339, "percentage": 65.38, "elapsed_time": "7:26:52", "remaining_time": "3:56:38"}
+{"current_steps": 800, "total_steps": 1216, "loss": 0.0686, "accuracy": 1.0, "learning_rate": 1.4367824904439242e-07, "epoch": 1.3143483023001095, "percentage": 65.79, "elapsed_time": "7:29:40", "remaining_time": "3:53:49"}
+{"current_steps": 805, "total_steps": 1216, "loss": 0.0558, "accuracy": 0.9733333587646484, "learning_rate": 1.4061098449376985e-07, "epoch": 1.3225629791894853, "percentage": 66.2, "elapsed_time": "7:32:30", "remaining_time": "3:51:01"}
+{"current_steps": 810, "total_steps": 1216, "loss": 0.0641, "accuracy": 0.9600000381469727, "learning_rate": 1.375639521289836e-07, "epoch": 1.330777656078861, "percentage": 66.61, "elapsed_time": "7:35:20", "remaining_time": "3:48:13"}
+{"current_steps": 815, "total_steps": 1216, "loss": 0.0633, "accuracy": 1.0, "learning_rate": 1.3453771551781756e-07, "epoch": 1.3389923329682367, "percentage": 67.02, "elapsed_time": "7:38:10", "remaining_time": "3:45:25"}
+{"current_steps": 820, "total_steps": 1216, "loss": 0.0652, "accuracy": 0.9733333587646484, "learning_rate": 1.3153283438175034e-07, "epoch": 1.3472070098576123, "percentage": 67.43, "elapsed_time": "7:41:00", "remaining_time": "3:42:37"}
+{"current_steps": 825, "total_steps": 1216, "loss": 0.0445, "accuracy": 0.9733333587646484, "learning_rate": 1.2854986449243124e-07, "epoch": 1.355421686746988, "percentage": 67.85, "elapsed_time": "7:43:49", "remaining_time": "3:39:49"}
+{"current_steps": 830, "total_steps": 1216, "loss": 0.0624, "accuracy": 0.9866666793823242, "learning_rate": 1.2558935756888675e-07, "epoch": 1.3636363636363638, "percentage": 68.26, "elapsed_time": "7:46:38", "remaining_time": "3:37:01"}
+{"current_steps": 835, "total_steps": 1216, "loss": 0.0482, "accuracy": 0.9866666793823242, "learning_rate": 1.226518611754767e-07, "epoch": 1.3718510405257394, "percentage": 68.67, "elapsed_time": "7:49:27", "remaining_time": "3:34:12"}
+{"current_steps": 840, "total_steps": 1216, "loss": 0.0703, "accuracy": 0.9866666793823242, "learning_rate": 1.1973791862061871e-07, "epoch": 1.380065717415115, "percentage": 69.08, "elapsed_time": "7:52:16", "remaining_time": "3:31:23"}
+{"current_steps": 845, "total_steps": 1216, "loss": 0.0471, "accuracy": 1.0, "learning_rate": 1.1684806885630003e-07, "epoch": 1.3882803943044908, "percentage": 69.49, "elapsed_time": "7:55:06", "remaining_time": "3:28:35"}
+{"current_steps": 850, "total_steps": 1216, "loss": 0.0593, "accuracy": 1.0, "learning_rate": 1.1398284637839486e-07, "epoch": 1.3964950711938664, "percentage": 69.9, "elapsed_time": "7:57:55", "remaining_time": "3:25:47"}
+{"current_steps": 855, "total_steps": 1216, "loss": 0.0441, "accuracy": 1.0, "learning_rate": 1.1114278112780601e-07, "epoch": 1.404709748083242, "percentage": 70.31, "elapsed_time": "8:00:43", "remaining_time": "3:22:58"}
+{"current_steps": 860, "total_steps": 1216, "loss": 0.0629, "accuracy": 0.9733333587646484, "learning_rate": 1.08328398392449e-07, "epoch": 1.4129244249726178, "percentage": 70.72, "elapsed_time": "8:03:33", "remaining_time": "3:20:10"}
+{"current_steps": 865, "total_steps": 1216, "loss": 0.0464, "accuracy": 0.9600000381469727, "learning_rate": 1.0554021871009677e-07, "epoch": 1.4211391018619934, "percentage": 71.13, "elapsed_time": "8:06:24", "remaining_time": "3:17:22"}
+{"current_steps": 870, "total_steps": 1216, "loss": 0.0537, "accuracy": 1.0, "learning_rate": 1.0277875777210299e-07, "epoch": 1.429353778751369, "percentage": 71.55, "elapsed_time": "8:09:15", "remaining_time": "3:14:34"}
+{"current_steps": 875, "total_steps": 1216, "loss": 0.0514, "accuracy": 1.0, "learning_rate": 1.0004452632802158e-07, "epoch": 1.4375684556407449, "percentage": 71.96, "elapsed_time": "8:12:05", "remaining_time": "3:11:46"}
+{"current_steps": 880, "total_steps": 1216, "loss": 0.0474, "accuracy": 1.0, "learning_rate": 9.733803009114044e-08, "epoch": 1.4457831325301205, "percentage": 72.37, "elapsed_time": "8:14:54", "remaining_time": "3:08:58"}
+{"current_steps": 885, "total_steps": 1216, "loss": 0.0529, "accuracy": 0.9733333587646484, "learning_rate": 9.465976964494682e-08, "epoch": 1.453997809419496, "percentage": 72.78, "elapsed_time": "8:17:44", "remaining_time": "3:06:09"}
+{"current_steps": 890, "total_steps": 1216, "loss": 0.0605, "accuracy": 0.9733333587646484, "learning_rate": 9.201024035054053e-08, "epoch": 1.462212486308872, "percentage": 73.19, "elapsed_time": "8:20:32", "remaining_time": "3:03:20"}
+{"current_steps": 895, "total_steps": 1216, "loss": 0.0583, "accuracy": 1.0, "learning_rate": 8.938993225501495e-08, "epoch": 1.4704271631982475, "percentage": 73.6, "elapsed_time": "8:23:19", "remaining_time": "3:00:31"}
+{"current_steps": 900, "total_steps": 1216, "loss": 0.0603, "accuracy": 0.9733333587646484, "learning_rate": 8.679933000081879e-08, "epoch": 1.4786418400876231, "percentage": 74.01, "elapsed_time": "8:26:08", "remaining_time": "2:57:42"}
+{"current_steps": 905, "total_steps": 1216, "loss": 0.0587, "accuracy": 0.9866666793823242, "learning_rate": 8.423891273611855e-08, "epoch": 1.486856516976999, "percentage": 74.42, "elapsed_time": "8:28:56", "remaining_time": "2:54:53"}
+{"current_steps": 910, "total_steps": 1216, "loss": 0.0669, "accuracy": 1.0, "learning_rate": 8.170915402617739e-08, "epoch": 1.4950711938663745, "percentage": 74.84, "elapsed_time": "8:31:44", "remaining_time": "2:52:04"}
+{"current_steps": 915, "total_steps": 1216, "loss": 0.0412, "accuracy": 1.0, "learning_rate": 7.921052176576643e-08, "epoch": 1.5032858707557502, "percentage": 75.25, "elapsed_time": "8:34:33", "remaining_time": "2:49:16"}
+{"current_steps": 920, "total_steps": 1216, "loss": 0.0616, "accuracy": 0.9600000381469727, "learning_rate": 7.674347809262377e-08, "epoch": 1.511500547645126, "percentage": 75.66, "elapsed_time": "8:37:22", "remaining_time": "2:46:27"}
+{"current_steps": 925, "total_steps": 1216, "loss": 0.0599, "accuracy": 0.9866666793823242, "learning_rate": 7.430847930198009e-08, "epoch": 1.5197152245345018, "percentage": 76.07, "elapsed_time": "8:40:10", "remaining_time": "2:43:38"}
+{"current_steps": 930, "total_steps": 1216, "loss": 0.0357, "accuracy": 0.9733333587646484, "learning_rate": 7.190597576216384e-08, "epoch": 1.5279299014238772, "percentage": 76.48, "elapsed_time": "8:42:59", "remaining_time": "2:40:49"}
+{"current_steps": 935, "total_steps": 1216, "loss": 0.055, "accuracy": 0.9866666793823242, "learning_rate": 6.953641183130224e-08, "epoch": 1.536144578313253, "percentage": 76.89, "elapsed_time": "8:45:47", "remaining_time": "2:38:00"}
+{"current_steps": 940, "total_steps": 1216, "loss": 0.0495, "accuracy": 0.9733333587646484, "learning_rate": 6.720022577513507e-08, "epoch": 1.5443592552026288, "percentage": 77.3, "elapsed_time": "8:48:34", "remaining_time": "2:35:11"}
+{"current_steps": 945, "total_steps": 1216, "loss": 0.0797, "accuracy": 0.9600000381469727, "learning_rate": 6.489784968595444e-08, "epoch": 1.5525739320920042, "percentage": 77.71, "elapsed_time": "8:51:21", "remaining_time": "2:32:22"}
+{"current_steps": 950, "total_steps": 1216, "loss": 0.0502, "accuracy": 0.9600000381469727, "learning_rate": 6.262970940268652e-08, "epoch": 1.56078860898138, "percentage": 78.12, "elapsed_time": "8:54:09", "remaining_time": "2:29:33"}
+{"current_steps": 955, "total_steps": 1216, "loss": 0.0455, "accuracy": 0.9866666793823242, "learning_rate": 6.039622443213008e-08, "epoch": 1.5690032858707559, "percentage": 78.54, "elapsed_time": "8:56:56", "remaining_time": "2:26:44"}
+{"current_steps": 960, "total_steps": 1216, "loss": 0.0443, "accuracy": 1.0, "learning_rate": 5.8197807871366e-08, "epoch": 1.5772179627601315, "percentage": 78.95, "elapsed_time": "8:59:44", "remaining_time": "2:23:55"}
+{"current_steps": 965, "total_steps": 1216, "loss": 0.0789, "accuracy": 0.9333333969116211, "learning_rate": 5.6034866331352376e-08, "epoch": 1.585432639649507, "percentage": 79.36, "elapsed_time": "9:02:32", "remaining_time": "2:21:06"}
+{"current_steps": 970, "total_steps": 1216, "loss": 0.0487, "accuracy": 0.9866666793823242, "learning_rate": 5.390779986171934e-08, "epoch": 1.593647316538883, "percentage": 79.77, "elapsed_time": "9:05:19", "remaining_time": "2:18:18"}
+{"current_steps": 975, "total_steps": 1216, "loss": 0.0437, "accuracy": 0.9866666793823242, "learning_rate": 5.1817001876777314e-08, "epoch": 1.6018619934282585, "percentage": 80.18, "elapsed_time": "9:08:07", "remaining_time": "2:15:29"}
+{"current_steps": 980, "total_steps": 1216, "loss": 0.0679, "accuracy": 0.9466667175292969, "learning_rate": 4.9762859082752464e-08, "epoch": 1.6100766703176341, "percentage": 80.59, "elapsed_time": "9:10:55", "remaining_time": "2:12:40"}
+{"current_steps": 985, "total_steps": 1216, "loss": 0.0416, "accuracy": 0.9866666793823242, "learning_rate": 4.774575140626316e-08, "epoch": 1.61829134720701, "percentage": 81.0, "elapsed_time": "9:13:44", "remaining_time": "2:09:51"}
+{"current_steps": 990, "total_steps": 1216, "loss": 0.078, "accuracy": 0.9866666793823242, "learning_rate": 4.5766051924049975e-08, "epoch": 1.6265060240963856, "percentage": 81.41, "elapsed_time": "9:16:32", "remaining_time": "2:07:02"}
+{"current_steps": 995, "total_steps": 1216, "loss": 0.0653, "accuracy": 1.0, "learning_rate": 4.3824126793972934e-08, "epoch": 1.6347207009857612, "percentage": 81.83, "elapsed_time": "9:19:20", "remaining_time": "2:04:14"}
+{"current_steps": 1000, "total_steps": 1216, "loss": 0.0455, "accuracy": 0.9600000381469727, "learning_rate": 4.192033518728819e-08, "epoch": 1.642935377875137, "percentage": 82.24, "elapsed_time": "9:22:08", "remaining_time": "2:01:25"}
+{"current_steps": 1005, "total_steps": 1216, "loss": 0.0528, "accuracy": 1.0, "learning_rate": 4.0055029222217125e-08, "epoch": 1.6511500547645126, "percentage": 82.65, "elapsed_time": "9:24:56", "remaining_time": "1:58:36"}
+{"current_steps": 1010, "total_steps": 1216, "loss": 0.0704, "accuracy": 0.9466667175292969, "learning_rate": 3.8228553898819904e-08, "epoch": 1.6593647316538882, "percentage": 83.06, "elapsed_time": "9:27:44", "remaining_time": "1:55:47"}
+{"current_steps": 1015, "total_steps": 1216, "loss": 0.0365, "accuracy": 1.0, "learning_rate": 3.6441247035185416e-08, "epoch": 1.667579408543264, "percentage": 83.47, "elapsed_time": "9:30:32", "remaining_time": "1:52:59"}
+{"current_steps": 1020, "total_steps": 1216, "loss": 0.0555, "accuracy": 1.0, "learning_rate": 3.4693439204949855e-08, "epoch": 1.6757940854326396, "percentage": 83.88, "elapsed_time": "9:33:20", "remaining_time": "1:50:10"}
+{"current_steps": 1025, "total_steps": 1216, "loss": 0.0854, "accuracy": 0.9466666579246521, "learning_rate": 3.298545367615493e-08, "epoch": 1.6840087623220152, "percentage": 84.29, "elapsed_time": "9:36:08", "remaining_time": "1:47:21"}
+{"current_steps": 1030, "total_steps": 1216, "loss": 0.0719, "accuracy": 0.9733333587646484, "learning_rate": 3.13176063514575e-08, "epoch": 1.692223439211391, "percentage": 84.7, "elapsed_time": "9:38:55", "remaining_time": "1:44:32"}
+{"current_steps": 1035, "total_steps": 1216, "loss": 0.0603, "accuracy": 0.9733333587646484, "learning_rate": 2.96902057097011e-08, "epoch": 1.7004381161007667, "percentage": 85.12, "elapsed_time": "9:41:43", "remaining_time": "1:41:43"}
+{"current_steps": 1040, "total_steps": 1216, "loss": 0.0456, "accuracy": 0.9866666793823242, "learning_rate": 2.8103552748861475e-08, "epoch": 1.7086527929901423, "percentage": 85.53, "elapsed_time": "9:44:31", "remaining_time": "1:38:55"}
+{"current_steps": 1045, "total_steps": 1216, "loss": 0.0673, "accuracy": 1.0, "learning_rate": 2.65579409303745e-08, "epoch": 1.716867469879518, "percentage": 85.94, "elapsed_time": "9:47:19", "remaining_time": "1:36:06"}
+{"current_steps": 1050, "total_steps": 1216, "loss": 0.0714, "accuracy": 0.9333333373069763, "learning_rate": 2.505365612485874e-08, "epoch": 1.7250821467688937, "percentage": 86.35, "elapsed_time": "9:50:06", "remaining_time": "1:33:17"}
+{"current_steps": 1055, "total_steps": 1216, "loss": 0.0577, "accuracy": 0.9600000381469727, "learning_rate": 2.3590976559242275e-08, "epoch": 1.7332968236582693, "percentage": 86.76, "elapsed_time": "9:52:54", "remaining_time": "1:30:28"}
+{"current_steps": 1060, "total_steps": 1216, "loss": 0.0635, "accuracy": 0.9866666793823242, "learning_rate": 2.21701727653025e-08, "epoch": 1.7415115005476451, "percentage": 87.17, "elapsed_time": "9:55:42", "remaining_time": "1:27:40"}
+{"current_steps": 1065, "total_steps": 1216, "loss": 0.0617, "accuracy": 0.9866666793823242, "learning_rate": 2.0791507529629522e-08, "epoch": 1.749726177437021, "percentage": 87.58, "elapsed_time": "9:58:30", "remaining_time": "1:24:51"}
+{"current_steps": 1070, "total_steps": 1216, "loss": 0.0572, "accuracy": 0.9733333587646484, "learning_rate": 1.945523584502262e-08, "epoch": 1.7579408543263964, "percentage": 87.99, "elapsed_time": "10:01:18", "remaining_time": "1:22:02"}
+{"current_steps": 1075, "total_steps": 1216, "loss": 0.0406, "accuracy": 0.9733333587646484, "learning_rate": 1.8161604863327072e-08, "epoch": 1.7661555312157722, "percentage": 88.4, "elapsed_time": "10:04:06", "remaining_time": "1:19:14"}
+{"current_steps": 1080, "total_steps": 1216, "loss": 0.0575, "accuracy": 0.9866666793823242, "learning_rate": 1.691085384972235e-08, "epoch": 1.774370208105148, "percentage": 88.82, "elapsed_time": "10:06:54", "remaining_time": "1:16:25"}
+{"current_steps": 1085, "total_steps": 1216, "loss": 0.0299, "accuracy": 0.9866666793823242, "learning_rate": 1.570321413846845e-08, "epoch": 1.7825848849945234, "percentage": 89.23, "elapsed_time": "10:09:42", "remaining_time": "1:13:36"}
+{"current_steps": 1090, "total_steps": 1216, "loss": 0.0607, "accuracy": 0.9866666793823242, "learning_rate": 1.4538909090118846e-08, "epoch": 1.7907995618838992, "percentage": 89.64, "elapsed_time": "10:12:30", "remaining_time": "1:10:48"}
+{"current_steps": 1095, "total_steps": 1216, "loss": 0.0563, "accuracy": 0.9466667175292969, "learning_rate": 1.3418154050208936e-08, "epoch": 1.799014238773275, "percentage": 90.05, "elapsed_time": "10:15:18", "remaining_time": "1:07:59"}
+{"current_steps": 1100, "total_steps": 1216, "loss": 0.036, "accuracy": 1.0, "learning_rate": 1.2341156309426447e-08, "epoch": 1.8072289156626506, "percentage": 90.46, "elapsed_time": "10:18:05", "remaining_time": "1:05:10"}
+{"current_steps": 1105, "total_steps": 1216, "loss": 0.0328, "accuracy": 0.9866666793823242, "learning_rate": 1.130811506527149e-08, "epoch": 1.8154435925520263, "percentage": 90.87, "elapsed_time": "10:20:53", "remaining_time": "1:02:22"}
+{"current_steps": 1110, "total_steps": 1216, "loss": 0.0392, "accuracy": 0.9866666793823242, "learning_rate": 1.0319221385213934e-08, "epoch": 1.823658269441402, "percentage": 91.28, "elapsed_time": "10:23:40", "remaining_time": "0:59:33"}
+{"current_steps": 1115, "total_steps": 1216, "loss": 0.066, "accuracy": 0.9466666579246521, "learning_rate": 9.374658171354411e-09, "epoch": 1.8318729463307777, "percentage": 91.69, "elapsed_time": "10:26:28", "remaining_time": "0:56:44"}
+{"current_steps": 1120, "total_steps": 1216, "loss": 0.0479, "accuracy": 1.0, "learning_rate": 8.474600126594983e-09, "epoch": 1.8400876232201533, "percentage": 92.11, "elapsed_time": "10:29:16", "remaining_time": "0:53:56"}
+{"current_steps": 1125, "total_steps": 1216, "loss": 0.0435, "accuracy": 0.9866666793823242, "learning_rate": 7.619213722327184e-09, "epoch": 1.8483023001095291, "percentage": 92.52, "elapsed_time": "10:32:04", "remaining_time": "0:51:07"}
+{"current_steps": 1130, "total_steps": 1216, "loss": 0.0669, "accuracy": 0.9866666793823242, "learning_rate": 6.808657167641896e-09, "epoch": 1.8565169769989047, "percentage": 92.93, "elapsed_time": "10:34:52", "remaining_time": "0:48:19"}
+{"current_steps": 1135, "total_steps": 1216, "loss": 0.0523, "accuracy": 0.9866666793823242, "learning_rate": 6.043080380067539e-09, "epoch": 1.8647316538882803, "percentage": 93.34, "elapsed_time": "10:37:40", "remaining_time": "0:45:30"}
+{"current_steps": 1140, "total_steps": 1216, "loss": 0.0601, "accuracy": 1.0, "learning_rate": 5.322624957841998e-09, "epoch": 1.8729463307776562, "percentage": 93.75, "elapsed_time": "10:40:28", "remaining_time": "0:42:41"}
+{"current_steps": 1145, "total_steps": 1216, "loss": 0.0757, "accuracy": 0.9733333587646484, "learning_rate": 4.647424153723101e-09, "epoch": 1.8811610076670318, "percentage": 94.16, "elapsed_time": "10:43:16", "remaining_time": "0:39:53"}
+{"current_steps": 1150, "total_steps": 1216, "loss": 0.049, "accuracy": 0.9866666793823242, "learning_rate": 4.0176028503425826e-09, "epoch": 1.8893756845564074, "percentage": 94.57, "elapsed_time": "10:46:03", "remaining_time": "0:37:04"}
+{"current_steps": 1155, "total_steps": 1216, "loss": 0.0481, "accuracy": 0.9733333587646484, "learning_rate": 3.433277537108481e-09, "epoch": 1.8975903614457832, "percentage": 94.98, "elapsed_time": "10:48:51", "remaining_time": "0:34:16"}
+{"current_steps": 1160, "total_steps": 1216, "loss": 0.0572, "accuracy": 0.9733333587646484, "learning_rate": 2.8945562886593944e-09, "epoch": 1.9058050383351588, "percentage": 95.39, "elapsed_time": "10:51:38", "remaining_time": "0:31:27"}
+{"current_steps": 1165, "total_steps": 1216, "loss": 0.0698, "accuracy": 0.9600000381469727, "learning_rate": 2.4015387448756976e-09, "epoch": 1.9140197152245344, "percentage": 95.81, "elapsed_time": "10:54:26", "remaining_time": "0:28:38"}
+{"current_steps": 1170, "total_steps": 1216, "loss": 0.0483, "accuracy": 0.9866666793823242, "learning_rate": 1.954316092450281e-09, "epoch": 1.9222343921139102, "percentage": 96.22, "elapsed_time": "10:57:14", "remaining_time": "0:25:50"}
+{"current_steps": 1175, "total_steps": 1216, "loss": 0.0453, "accuracy": 1.0, "learning_rate": 1.5529710480231272e-09, "epoch": 1.9304490690032858, "percentage": 96.63, "elapsed_time": "11:00:02", "remaining_time": "0:23:01"}
+{"current_steps": 1180, "total_steps": 1216, "loss": 0.0718, "accuracy": 0.9733333587646484, "learning_rate": 1.1975778428823524e-09, "epoch": 1.9386637458926614, "percentage": 97.04, "elapsed_time": "11:02:49", "remaining_time": "0:20:13"}
+{"current_steps": 1185, "total_steps": 1216, "loss": 0.0371, "accuracy": 0.9600000381469727, "learning_rate": 8.882022092346064e-10, "epoch": 1.9468784227820373, "percentage": 97.45, "elapsed_time": "11:05:37", "remaining_time": "0:17:24"}
+{"current_steps": 1190, "total_steps": 1216, "loss": 0.042, "accuracy": 0.9866666793823242, "learning_rate": 6.249013680474368e-10, "epoch": 1.9550930996714129, "percentage": 97.86, "elapsed_time": "11:08:24", "remaining_time": "0:14:36"}
+{"current_steps": 1195, "total_steps": 1216, "loss": 0.0791, "accuracy": 0.9733333587646484, "learning_rate": 4.0772401846608794e-10, "epoch": 1.9633077765607885, "percentage": 98.27, "elapsed_time": "11:11:12", "remaining_time": "0:11:47"}
+{"current_steps": 1200, "total_steps": 1216, "loss": 0.057, "accuracy": 1.0, "learning_rate": 2.367103288061223e-10, "epoch": 1.9715224534501643, "percentage": 98.68, "elapsed_time": "11:14:00", "remaining_time": "0:08:59"}
+{"current_steps": 1205, "total_steps": 1216, "loss": 0.0365, "accuracy": 0.9866666793823242, "learning_rate": 1.1189192912416933e-10, "epoch": 1.9797371303395401, "percentage": 99.1, "elapsed_time": "11:16:48", "remaining_time": "0:06:10"}
+{"current_steps": 1210, "total_steps": 1216, "loss": 0.0535, "accuracy": 0.9733333587646484, "learning_rate": 3.329190536757731e-11, "epoch": 1.9879518072289155, "percentage": 99.51, "elapsed_time": "11:19:36", "remaining_time": "0:03:22"}
+{"current_steps": 1215, "total_steps": 1216, "loss": 0.0471, "accuracy": 0.9600000381469727, "learning_rate": 9.247951046897906e-13, "epoch": 1.9961664841182913, "percentage": 99.92, "elapsed_time": "11:22:23", "remaining_time": "0:00:33"}
+{"current_steps": 1216, "total_steps": 1216, "epoch": 1.9978094194961664, "percentage": 100.0, "elapsed_time": "11:24:13", "remaining_time": "0:00:00"}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d7ec14d067d57bc9a8fd40eab76a3931a232e4eca121571c52a96ff477151bf
+size 7032