Upload folder using huggingface_hub

README.md

@@ -0,0 +1,65 @@
+---
+base_model: Qwen/Qwen-1_8B
+library_name: peft
+license: other
+tags:
+- llama-factory
+- lora
+- generated_from_trainer
+model-index:
+- name: train_2024-09-02-15-46-54
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# train_2024-09-02-15-46-54
+
+This model is a fine-tuned version of [Qwen/Qwen-1_8B](https://huggingface.co/Qwen/Qwen-1_8B) on the glaive_toolcall_en dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.3859
+- Num Input Tokens Seen: 1596080
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-05
+- train_batch_size: 2
+- eval_batch_size: 2
+- seed: 42
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 16
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- num_epochs: 3.0
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss | Input Tokens Seen |
+|:-------------:|:------:|:----:|:---------------:|:-----------------:|
+| 0.4726        | 1.7778 | 100  | 0.3941          | 949424            |
+
+
+### Framework versions
+
+- PEFT 0.12.0
+- Transformers 4.44.2
+- Pytorch 2.3.1+cu121
+- Datasets 2.21.0
+- Tokenizers 0.19.1

adapter_config.json

@@ -0,0 +1,31 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen-1_8B",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn",
+    "w2",
+    "c_proj",
+    "w1"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3944faf4720d867e867863b0e244d4f412b6f008b9359fcc98360ebb37bf510b
+size 26867400

all_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 2.986666666666667,
+    "eval_loss": 0.38587039709091187,
+    "eval_runtime": 13.1955,
+    "eval_samples_per_second": 7.578,
+    "eval_steps_per_second": 3.789,
+    "num_input_tokens_seen": 1596080,
+    "total_flos": 1.467473931042816e+16,
+    "train_loss": 0.5455739086582547,
+    "train_runtime": 1063.0707,
+    "train_samples_per_second": 2.54,
+    "train_steps_per_second": 0.158
+}

checkpoint-100/README.md

@@ -0,0 +1,202 @@
+---
+base_model: Qwen/Qwen-1_8B
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.12.0

checkpoint-100/adapter_config.json

@@ -0,0 +1,31 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen-1_8B",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn",
+    "w2",
+    "c_proj",
+    "w1"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-100/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64238ab16528a40f2c5fbcde5419a3e8d83e298bd7b8e677cbcaa1ea66ceb7c4
+size 26867400

checkpoint-100/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a469617e9cb3688483ed2ba66fcdc7dcaa3e584dc768982c6f6b7442fcf46982
+size 53875386

checkpoint-100/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42477e3febec09c1d5f853ced62e06fb0b175d9d96e488c83c57aeeb062d8c97
+size 14244

checkpoint-100/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:367336456ed85b45075ea2f90d51ce174bed627c54e3e0ef29c073a10e04faa6
+size 1064

checkpoint-100/special_tokens_map.json

@@ -0,0 +1,10 @@
+{
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

checkpoint-100/tokenization_qwen.py

@@ -0,0 +1,276 @@
+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Tokenization classes for QWen."""
+
+import base64
+import logging
+import os
+import unicodedata
+from typing import Collection, Dict, List, Set, Tuple, Union
+
+import tiktoken
+from transformers import PreTrainedTokenizer, AddedToken
+
+logger = logging.getLogger(__name__)
+
+
+VOCAB_FILES_NAMES = {"vocab_file": "qwen.tiktoken"}
+
+PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
+ENDOFTEXT = "<|endoftext|>"
+IMSTART = "<|im_start|>"
+IMEND = "<|im_end|>"
+# as the default behavior is changed to allow special tokens in
+# regular texts, the surface forms of special tokens need to be
+# as different as possible to minimize the impact
+EXTRAS = tuple((f"<|extra_{i}|>" for i in range(205)))
+# changed to use actual index to avoid misconfiguration with vocabulary expansion
+SPECIAL_START_ID = 151643
+SPECIAL_TOKENS = tuple(
+    enumerate(
+        (
+            (
+                ENDOFTEXT,
+                IMSTART,
+                IMEND,
+            )
+            + EXTRAS
+        ),
+        start=SPECIAL_START_ID,
+    )
+)
+SPECIAL_TOKENS_SET = set(t for i, t in SPECIAL_TOKENS)
+
+
+def _load_tiktoken_bpe(tiktoken_bpe_file: str) -> Dict[bytes, int]:
+    with open(tiktoken_bpe_file, "rb") as f:
+        contents = f.read()
+    return {
+        base64.b64decode(token): int(rank)
+        for token, rank in (line.split() for line in contents.splitlines() if line)
+    }
+
+
+class QWenTokenizer(PreTrainedTokenizer):
+    """QWen tokenizer."""
+
+    vocab_files_names = VOCAB_FILES_NAMES
+
+    def __init__(
+        self,
+        vocab_file,
+        errors="replace",
+        extra_vocab_file=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        # how to handle errors in decoding UTF-8 byte sequences
+        # use ignore if you are in streaming inference
+        self.errors = errors  
+
+        self.mergeable_ranks = _load_tiktoken_bpe(vocab_file)  # type: Dict[bytes, int]
+        self.special_tokens = {
+            token: index
+            for index, token in SPECIAL_TOKENS
+        }
+
+        # try load extra vocab from file
+        if extra_vocab_file is not None:
+            used_ids = set(self.mergeable_ranks.values()) | set(self.special_tokens.values())
+            extra_mergeable_ranks = _load_tiktoken_bpe(extra_vocab_file)
+            for token, index in extra_mergeable_ranks.items():
+                if token in self.mergeable_ranks:
+                    logger.info(f"extra token {token} exists, skipping")
+                    continue
+                if index in used_ids:
+                    logger.info(f'the index {index} for extra token {token} exists, skipping')
+                    continue
+                self.mergeable_ranks[token] = index
+            # the index may be sparse after this, but don't worry tiktoken.Encoding will handle this
+
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        assert (
+            len(self.mergeable_ranks) + len(self.special_tokens) == enc.n_vocab
+        ), f"{len(self.mergeable_ranks) + len(self.special_tokens)} != {enc.n_vocab} in encoding"
+
+        self.decoder = {
+            v: k for k, v in self.mergeable_ranks.items()
+        }  # type: dict[int, bytes|str]
+        self.decoder.update({v: k for k, v in self.special_tokens.items()})
+
+        self.tokenizer = enc  # type: tiktoken.Encoding
+
+        self.eod_id = self.tokenizer.eot_token
+        self.im_start_id = self.special_tokens[IMSTART]
+        self.im_end_id = self.special_tokens[IMEND]
+
+    def __getstate__(self):
+        # for pickle lovers
+        state = self.__dict__.copy()
+        del state["tokenizer"]
+        return state
+
+    def __setstate__(self, state):
+        # tokenizer is not python native; don't pass it; rebuild it
+        self.__dict__.update(state)
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        self.tokenizer = enc
+
+    def __len__(self) -> int:
+        return self.tokenizer.n_vocab
+
+    def get_vocab(self) -> Dict[bytes, int]:
+        return self.mergeable_ranks
+
+    def convert_tokens_to_ids(
+        self, tokens: Union[bytes, str, List[Union[bytes, str]]]
+    ) -> List[int]:
+        ids = []
+        if isinstance(tokens, (str, bytes)):
+            if tokens in self.special_tokens:
+                return self.special_tokens[tokens]
+            else:
+                return self.mergeable_ranks.get(tokens)
+        for token in tokens:
+            if token in self.special_tokens:
+                ids.append(self.special_tokens[token])
+            else:
+                ids.append(self.mergeable_ranks.get(token))
+        return ids
+
+    def _add_tokens(
+        self,
+        new_tokens: Union[List[str], List[AddedToken]],
+        special_tokens: bool = False,
+    ) -> int:
+        if not special_tokens and new_tokens:
+            raise ValueError("Adding regular tokens is not supported")
+        for token in new_tokens:
+            surface_form = token.content if isinstance(token, AddedToken) else token
+            if surface_form not in SPECIAL_TOKENS_SET:
+                raise ValueError("Adding unknown special tokens is not supported")
+        return 0
+
+    def save_vocabulary(self, save_directory: str, **kwargs) -> Tuple[str]:
+        """
+        Save only the vocabulary of the tokenizer (vocabulary).
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        file_path = os.path.join(save_directory, "qwen.tiktoken")
+        with open(file_path, "w", encoding="utf8") as w:
+            for k, v in self.mergeable_ranks.items():
+                line = base64.b64encode(k).decode("utf8") + " " + str(v) + "\n"
+                w.write(line)
+        return (file_path,)
+
+    def tokenize(
+        self,
+        text: str,
+        allowed_special: Union[Set, str] = "all",
+        disallowed_special: Union[Collection, str] = (),
+        **kwargs,
+    ) -> List[Union[bytes, str]]:
+        """
+        Converts a string in a sequence of tokens.
+
+        Args:
+            text (`str`):
+                The sequence to be encoded.
+            allowed_special (`Literal["all"]` or `set`):
+                The surface forms of the tokens to be encoded as special tokens in regular texts.
+                Default to "all".
+            disallowed_special (`Literal["all"]` or `Collection`):
+                The surface forms of the tokens that should not be in regular texts and trigger errors.
+                Default to an empty tuple.
+
+            kwargs (additional keyword arguments, *optional*):
+                Will be passed to the underlying model specific encode method.
+
+        Returns:
+            `List[bytes|str]`: The list of tokens.
+        """
+        tokens = []
+        text = unicodedata.normalize("NFC", text)
+
+        # this implementation takes a detour: text -> token id -> token surface forms
+        for t in self.tokenizer.encode(
+            text, allowed_special=allowed_special, disallowed_special=disallowed_special
+        ):
+            tokens.append(self.decoder[t])
+        return tokens
+
+    def convert_tokens_to_string(self, tokens: List[Union[bytes, str]]) -> str:
+        """
+        Converts a sequence of tokens in a single string.
+        """
+        text = ""
+        temp = b""
+        for t in tokens:
+            if isinstance(t, str):
+                if temp:
+                    text += temp.decode("utf-8", errors=self.errors)
+                    temp = b""
+                text += t
+            elif isinstance(t, bytes):
+                temp += t
+            else:
+                raise TypeError("token should only be of type types or str")
+        if temp:
+            text += temp.decode("utf-8", errors=self.errors)
+        return text
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_vocab
+
+    def _convert_id_to_token(self, index: int) -> Union[bytes, str]:
+        """Converts an id to a token, special tokens included"""
+        if index in self.decoder:
+            return self.decoder[index]
+        raise ValueError("unknown ids")
+
+    def _convert_token_to_id(self, token: Union[bytes, str]) -> int:
+        """Converts a token to an id using the vocab, special tokens included"""
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        if token in self.mergeable_ranks:
+            return self.mergeable_ranks[token]
+        raise ValueError("unknown token")
+
+    def _tokenize(self, text: str, **kwargs):
+        """
+        Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
+        vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
+
+        Do NOT take care of added tokens.
+        """
+        raise NotImplementedError
+
+    def _decode(
+        self,
+        token_ids: Union[int, List[int]],
+        skip_special_tokens: bool = False,
+        errors: str = None,
+        **kwargs,
+    ) -> str:
+        if isinstance(token_ids, int):
+            token_ids = [token_ids]
+        if skip_special_tokens:
+            token_ids = [i for i in token_ids if i < self.eod_id]
+        return self.tokenizer.decode(token_ids, errors=errors or self.errors)

checkpoint-100/tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "added_tokens_decoder": {},
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "chat_template": "{% 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_message + '\n' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '\nAssistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + '\n' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

checkpoint-100/trainer_state.json

@@ -0,0 +1,202 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.7777777777777777,
+  "eval_steps": 100,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.7499012351036072,
+      "learning_rate": 4.989080197352834e-05,
+      "loss": 0.9542,
+      "num_input_tokens_seen": 47168,
+      "step": 5
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.6703861355781555,
+      "learning_rate": 4.956416183083221e-05,
+      "loss": 0.7834,
+      "num_input_tokens_seen": 100640,
+      "step": 10
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.6384798884391785,
+      "learning_rate": 4.9022933048627496e-05,
+      "loss": 0.7296,
+      "num_input_tokens_seen": 150048,
+      "step": 15
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.36834755539894104,
+      "learning_rate": 4.827184371610511e-05,
+      "loss": 0.6653,
+      "num_input_tokens_seen": 196928,
+      "step": 20
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.2962658405303955,
+      "learning_rate": 4.731745523109029e-05,
+      "loss": 0.683,
+      "num_input_tokens_seen": 244592,
+      "step": 25
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.33365580439567566,
+      "learning_rate": 4.6168104980707107e-05,
+      "loss": 0.5651,
+      "num_input_tokens_seen": 293488,
+      "step": 30
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.4523426294326782,
+      "learning_rate": 4.4833833507280884e-05,
+      "loss": 0.5411,
+      "num_input_tokens_seen": 338160,
+      "step": 35
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.41214215755462646,
+      "learning_rate": 4.332629679574566e-05,
+      "loss": 0.572,
+      "num_input_tokens_seen": 382768,
+      "step": 40
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3807964026927948,
+      "learning_rate": 4.16586644488001e-05,
+      "loss": 0.6106,
+      "num_input_tokens_seen": 430928,
+      "step": 45
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.45064786076545715,
+      "learning_rate": 3.9845504639337535e-05,
+      "loss": 0.5552,
+      "num_input_tokens_seen": 477520,
+      "step": 50
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.4501620829105377,
+      "learning_rate": 3.790265684518767e-05,
+      "loss": 0.5659,
+      "num_input_tokens_seen": 526640,
+      "step": 55
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.24308599531650543,
+      "learning_rate": 3.5847093477938956e-05,
+      "loss": 0.58,
+      "num_input_tokens_seen": 577184,
+      "step": 60
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 0.33691608905792236,
+      "learning_rate": 3.369677161463068e-05,
+      "loss": 0.5429,
+      "num_input_tokens_seen": 625344,
+      "step": 65
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 0.2771952748298645,
+      "learning_rate": 3.147047612756302e-05,
+      "loss": 0.4384,
+      "num_input_tokens_seen": 670224,
+      "step": 70
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4307265281677246,
+      "learning_rate": 2.918765558261841e-05,
+      "loss": 0.5749,
+      "num_input_tokens_seen": 714880,
+      "step": 75
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 0.4149227440357208,
+      "learning_rate": 2.686825233966061e-05,
+      "loss": 0.4222,
+      "num_input_tokens_seen": 752944,
+      "step": 80
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 0.44140687584877014,
+      "learning_rate": 2.4532528339227452e-05,
+      "loss": 0.556,
+      "num_input_tokens_seen": 802064,
+      "step": 85
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.376556396484375,
+      "learning_rate": 2.2200888097417307e-05,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 852112,
+      "step": 90
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 0.3596157431602478,
+      "learning_rate": 1.9893700455257996e-05,
+      "loss": 0.5085,
+      "num_input_tokens_seen": 905184,
+      "step": 95
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.4776301085948944,
+      "learning_rate": 1.7631120639727393e-05,
+      "loss": 0.4726,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "eval_loss": 0.39410004019737244,
+      "eval_runtime": 11.4337,
+      "eval_samples_per_second": 8.746,
+      "eval_steps_per_second": 4.373,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 168,
+  "num_input_tokens_seen": 949424,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8729230173339648.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-100/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3714cd435343e43fff99200ac2402d2ab09a63941e76fb7bd73195c6e4f4472
+size 5368

checkpoint-168/README.md

@@ -0,0 +1,202 @@
+---
+base_model: Qwen/Qwen-1_8B
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.12.0

checkpoint-168/adapter_config.json

@@ -0,0 +1,31 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen-1_8B",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn",
+    "w2",
+    "c_proj",
+    "w1"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-168/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3944faf4720d867e867863b0e244d4f412b6f008b9359fcc98360ebb37bf510b
+size 26867400

checkpoint-168/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ac14cfb7644f827b5f35128e3320eaa98f6f87f7ad323a5b94ba2a253d83bd5
+size 53875386

checkpoint-168/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40cb1869983733651486e797294b9d075e35ed911b745abc673b44d5ac187b23
+size 14244

checkpoint-168/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffd46888908b2132048cd2c5b54269699691843b9c7405080bb376310c5ce3c5
+size 1064

checkpoint-168/special_tokens_map.json

@@ -0,0 +1,10 @@
+{
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

checkpoint-168/tokenization_qwen.py

@@ -0,0 +1,276 @@
+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Tokenization classes for QWen."""
+
+import base64
+import logging
+import os
+import unicodedata
+from typing import Collection, Dict, List, Set, Tuple, Union
+
+import tiktoken
+from transformers import PreTrainedTokenizer, AddedToken
+
+logger = logging.getLogger(__name__)
+
+
+VOCAB_FILES_NAMES = {"vocab_file": "qwen.tiktoken"}
+
+PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
+ENDOFTEXT = "<|endoftext|>"
+IMSTART = "<|im_start|>"
+IMEND = "<|im_end|>"
+# as the default behavior is changed to allow special tokens in
+# regular texts, the surface forms of special tokens need to be
+# as different as possible to minimize the impact
+EXTRAS = tuple((f"<|extra_{i}|>" for i in range(205)))
+# changed to use actual index to avoid misconfiguration with vocabulary expansion
+SPECIAL_START_ID = 151643
+SPECIAL_TOKENS = tuple(
+    enumerate(
+        (
+            (
+                ENDOFTEXT,
+                IMSTART,
+                IMEND,
+            )
+            + EXTRAS
+        ),
+        start=SPECIAL_START_ID,
+    )
+)
+SPECIAL_TOKENS_SET = set(t for i, t in SPECIAL_TOKENS)
+
+
+def _load_tiktoken_bpe(tiktoken_bpe_file: str) -> Dict[bytes, int]:
+    with open(tiktoken_bpe_file, "rb") as f:
+        contents = f.read()
+    return {
+        base64.b64decode(token): int(rank)
+        for token, rank in (line.split() for line in contents.splitlines() if line)
+    }
+
+
+class QWenTokenizer(PreTrainedTokenizer):
+    """QWen tokenizer."""
+
+    vocab_files_names = VOCAB_FILES_NAMES
+
+    def __init__(
+        self,
+        vocab_file,
+        errors="replace",
+        extra_vocab_file=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        # how to handle errors in decoding UTF-8 byte sequences
+        # use ignore if you are in streaming inference
+        self.errors = errors  
+
+        self.mergeable_ranks = _load_tiktoken_bpe(vocab_file)  # type: Dict[bytes, int]
+        self.special_tokens = {
+            token: index
+            for index, token in SPECIAL_TOKENS
+        }
+
+        # try load extra vocab from file
+        if extra_vocab_file is not None:
+            used_ids = set(self.mergeable_ranks.values()) | set(self.special_tokens.values())
+            extra_mergeable_ranks = _load_tiktoken_bpe(extra_vocab_file)
+            for token, index in extra_mergeable_ranks.items():
+                if token in self.mergeable_ranks:
+                    logger.info(f"extra token {token} exists, skipping")
+                    continue
+                if index in used_ids:
+                    logger.info(f'the index {index} for extra token {token} exists, skipping')
+                    continue
+                self.mergeable_ranks[token] = index
+            # the index may be sparse after this, but don't worry tiktoken.Encoding will handle this
+
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        assert (
+            len(self.mergeable_ranks) + len(self.special_tokens) == enc.n_vocab
+        ), f"{len(self.mergeable_ranks) + len(self.special_tokens)} != {enc.n_vocab} in encoding"
+
+        self.decoder = {
+            v: k for k, v in self.mergeable_ranks.items()
+        }  # type: dict[int, bytes|str]
+        self.decoder.update({v: k for k, v in self.special_tokens.items()})
+
+        self.tokenizer = enc  # type: tiktoken.Encoding
+
+        self.eod_id = self.tokenizer.eot_token
+        self.im_start_id = self.special_tokens[IMSTART]
+        self.im_end_id = self.special_tokens[IMEND]
+
+    def __getstate__(self):
+        # for pickle lovers
+        state = self.__dict__.copy()
+        del state["tokenizer"]
+        return state
+
+    def __setstate__(self, state):
+        # tokenizer is not python native; don't pass it; rebuild it
+        self.__dict__.update(state)
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        self.tokenizer = enc
+
+    def __len__(self) -> int:
+        return self.tokenizer.n_vocab
+
+    def get_vocab(self) -> Dict[bytes, int]:
+        return self.mergeable_ranks
+
+    def convert_tokens_to_ids(
+        self, tokens: Union[bytes, str, List[Union[bytes, str]]]
+    ) -> List[int]:
+        ids = []
+        if isinstance(tokens, (str, bytes)):
+            if tokens in self.special_tokens:
+                return self.special_tokens[tokens]
+            else:
+                return self.mergeable_ranks.get(tokens)
+        for token in tokens:
+            if token in self.special_tokens:
+                ids.append(self.special_tokens[token])
+            else:
+                ids.append(self.mergeable_ranks.get(token))
+        return ids
+
+    def _add_tokens(
+        self,
+        new_tokens: Union[List[str], List[AddedToken]],
+        special_tokens: bool = False,
+    ) -> int:
+        if not special_tokens and new_tokens:
+            raise ValueError("Adding regular tokens is not supported")
+        for token in new_tokens:
+            surface_form = token.content if isinstance(token, AddedToken) else token
+            if surface_form not in SPECIAL_TOKENS_SET:
+                raise ValueError("Adding unknown special tokens is not supported")
+        return 0
+
+    def save_vocabulary(self, save_directory: str, **kwargs) -> Tuple[str]:
+        """
+        Save only the vocabulary of the tokenizer (vocabulary).
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        file_path = os.path.join(save_directory, "qwen.tiktoken")
+        with open(file_path, "w", encoding="utf8") as w:
+            for k, v in self.mergeable_ranks.items():
+                line = base64.b64encode(k).decode("utf8") + " " + str(v) + "\n"
+                w.write(line)
+        return (file_path,)
+
+    def tokenize(
+        self,
+        text: str,
+        allowed_special: Union[Set, str] = "all",
+        disallowed_special: Union[Collection, str] = (),
+        **kwargs,
+    ) -> List[Union[bytes, str]]:
+        """
+        Converts a string in a sequence of tokens.
+
+        Args:
+            text (`str`):
+                The sequence to be encoded.
+            allowed_special (`Literal["all"]` or `set`):
+                The surface forms of the tokens to be encoded as special tokens in regular texts.
+                Default to "all".
+            disallowed_special (`Literal["all"]` or `Collection`):
+                The surface forms of the tokens that should not be in regular texts and trigger errors.
+                Default to an empty tuple.
+
+            kwargs (additional keyword arguments, *optional*):
+                Will be passed to the underlying model specific encode method.
+
+        Returns:
+            `List[bytes|str]`: The list of tokens.
+        """
+        tokens = []
+        text = unicodedata.normalize("NFC", text)
+
+        # this implementation takes a detour: text -> token id -> token surface forms
+        for t in self.tokenizer.encode(
+            text, allowed_special=allowed_special, disallowed_special=disallowed_special
+        ):
+            tokens.append(self.decoder[t])
+        return tokens
+
+    def convert_tokens_to_string(self, tokens: List[Union[bytes, str]]) -> str:
+        """
+        Converts a sequence of tokens in a single string.
+        """
+        text = ""
+        temp = b""
+        for t in tokens:
+            if isinstance(t, str):
+                if temp:
+                    text += temp.decode("utf-8", errors=self.errors)
+                    temp = b""
+                text += t
+            elif isinstance(t, bytes):
+                temp += t
+            else:
+                raise TypeError("token should only be of type types or str")
+        if temp:
+            text += temp.decode("utf-8", errors=self.errors)
+        return text
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_vocab
+
+    def _convert_id_to_token(self, index: int) -> Union[bytes, str]:
+        """Converts an id to a token, special tokens included"""
+        if index in self.decoder:
+            return self.decoder[index]
+        raise ValueError("unknown ids")
+
+    def _convert_token_to_id(self, token: Union[bytes, str]) -> int:
+        """Converts a token to an id using the vocab, special tokens included"""
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        if token in self.mergeable_ranks:
+            return self.mergeable_ranks[token]
+        raise ValueError("unknown token")
+
+    def _tokenize(self, text: str, **kwargs):
+        """
+        Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
+        vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
+
+        Do NOT take care of added tokens.
+        """
+        raise NotImplementedError
+
+    def _decode(
+        self,
+        token_ids: Union[int, List[int]],
+        skip_special_tokens: bool = False,
+        errors: str = None,
+        **kwargs,
+    ) -> str:
+        if isinstance(token_ids, int):
+            token_ids = [token_ids]
+        if skip_special_tokens:
+            token_ids = [i for i in token_ids if i < self.eod_id]
+        return self.tokenizer.decode(token_ids, errors=errors or self.errors)

checkpoint-168/tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "added_tokens_decoder": {},
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "chat_template": "{% 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_message + '\n' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '\nAssistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + '\n' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

checkpoint-168/trainer_state.json

@@ -0,0 +1,306 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.986666666666667,
+  "eval_steps": 100,
+  "global_step": 168,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.7499012351036072,
+      "learning_rate": 4.989080197352834e-05,
+      "loss": 0.9542,
+      "num_input_tokens_seen": 47168,
+      "step": 5
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.6703861355781555,
+      "learning_rate": 4.956416183083221e-05,
+      "loss": 0.7834,
+      "num_input_tokens_seen": 100640,
+      "step": 10
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.6384798884391785,
+      "learning_rate": 4.9022933048627496e-05,
+      "loss": 0.7296,
+      "num_input_tokens_seen": 150048,
+      "step": 15
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.36834755539894104,
+      "learning_rate": 4.827184371610511e-05,
+      "loss": 0.6653,
+      "num_input_tokens_seen": 196928,
+      "step": 20
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.2962658405303955,
+      "learning_rate": 4.731745523109029e-05,
+      "loss": 0.683,
+      "num_input_tokens_seen": 244592,
+      "step": 25
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.33365580439567566,
+      "learning_rate": 4.6168104980707107e-05,
+      "loss": 0.5651,
+      "num_input_tokens_seen": 293488,
+      "step": 30
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.4523426294326782,
+      "learning_rate": 4.4833833507280884e-05,
+      "loss": 0.5411,
+      "num_input_tokens_seen": 338160,
+      "step": 35
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.41214215755462646,
+      "learning_rate": 4.332629679574566e-05,
+      "loss": 0.572,
+      "num_input_tokens_seen": 382768,
+      "step": 40
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3807964026927948,
+      "learning_rate": 4.16586644488001e-05,
+      "loss": 0.6106,
+      "num_input_tokens_seen": 430928,
+      "step": 45
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.45064786076545715,
+      "learning_rate": 3.9845504639337535e-05,
+      "loss": 0.5552,
+      "num_input_tokens_seen": 477520,
+      "step": 50
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.4501620829105377,
+      "learning_rate": 3.790265684518767e-05,
+      "loss": 0.5659,
+      "num_input_tokens_seen": 526640,
+      "step": 55
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.24308599531650543,
+      "learning_rate": 3.5847093477938956e-05,
+      "loss": 0.58,
+      "num_input_tokens_seen": 577184,
+      "step": 60
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 0.33691608905792236,
+      "learning_rate": 3.369677161463068e-05,
+      "loss": 0.5429,
+      "num_input_tokens_seen": 625344,
+      "step": 65
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 0.2771952748298645,
+      "learning_rate": 3.147047612756302e-05,
+      "loss": 0.4384,
+      "num_input_tokens_seen": 670224,
+      "step": 70
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4307265281677246,
+      "learning_rate": 2.918765558261841e-05,
+      "loss": 0.5749,
+      "num_input_tokens_seen": 714880,
+      "step": 75
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 0.4149227440357208,
+      "learning_rate": 2.686825233966061e-05,
+      "loss": 0.4222,
+      "num_input_tokens_seen": 752944,
+      "step": 80
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 0.44140687584877014,
+      "learning_rate": 2.4532528339227452e-05,
+      "loss": 0.556,
+      "num_input_tokens_seen": 802064,
+      "step": 85
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.376556396484375,
+      "learning_rate": 2.2200888097417307e-05,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 852112,
+      "step": 90
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 0.3596157431602478,
+      "learning_rate": 1.9893700455257996e-05,
+      "loss": 0.5085,
+      "num_input_tokens_seen": 905184,
+      "step": 95
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.4776301085948944,
+      "learning_rate": 1.7631120639727393e-05,
+      "loss": 0.4726,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "eval_loss": 0.39410004019737244,
+      "eval_runtime": 11.4337,
+      "eval_samples_per_second": 8.746,
+      "eval_steps_per_second": 4.373,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.26946938037872314,
+      "learning_rate": 1.5432914190872757e-05,
+      "loss": 0.4874,
+      "num_input_tokens_seen": 996112,
+      "step": 105
+    },
+    {
+      "epoch": 1.9555555555555557,
+      "grad_norm": 0.37138107419013977,
+      "learning_rate": 1.331828429317345e-05,
+      "loss": 0.5235,
+      "num_input_tokens_seen": 1047520,
+      "step": 110
+    },
+    {
+      "epoch": 2.0444444444444443,
+      "grad_norm": 0.33912232518196106,
+      "learning_rate": 1.130570401955322e-05,
+      "loss": 0.4984,
+      "num_input_tokens_seen": 1096832,
+      "step": 115
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.4000737965106964,
+      "learning_rate": 9.412754953531663e-06,
+      "loss": 0.4777,
+      "num_input_tokens_seen": 1151104,
+      "step": 120
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.5536447167396545,
+      "learning_rate": 7.65597359928646e-06,
+      "loss": 0.4918,
+      "num_input_tokens_seen": 1194592,
+      "step": 125
+    },
+    {
+      "epoch": 2.311111111111111,
+      "grad_norm": 0.42406854033470154,
+      "learning_rate": 6.050706921363672e-06,
+      "loss": 0.4573,
+      "num_input_tokens_seen": 1235104,
+      "step": 130
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.37090376019477844,
+      "learning_rate": 4.610978276018496e-06,
+      "loss": 0.4549,
+      "num_input_tokens_seen": 1280560,
+      "step": 135
+    },
+    {
+      "epoch": 2.488888888888889,
+      "grad_norm": 0.3070351779460907,
+      "learning_rate": 3.3493649053890326e-06,
+      "loss": 0.5422,
+      "num_input_tokens_seen": 1329792,
+      "step": 140
+    },
+    {
+      "epoch": 2.5777777777777775,
+      "grad_norm": 0.3824547529220581,
+      "learning_rate": 2.2768880646947268e-06,
+      "loss": 0.4968,
+      "num_input_tokens_seen": 1380976,
+      "step": 145
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.4585011899471283,
+      "learning_rate": 1.4029167422908107e-06,
+      "loss": 0.4361,
+      "num_input_tokens_seen": 1423712,
+      "step": 150
+    },
+    {
+      "epoch": 2.7555555555555555,
+      "grad_norm": 0.3712801933288574,
+      "learning_rate": 7.350858136652261e-07,
+      "loss": 0.4517,
+      "num_input_tokens_seen": 1470000,
+      "step": 155
+    },
+    {
+      "epoch": 2.8444444444444446,
+      "grad_norm": 0.4640035629272461,
+      "learning_rate": 2.7922934437178695e-07,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 1515168,
+      "step": 160
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.24545951187610626,
+      "learning_rate": 3.9329624554584884e-08,
+      "loss": 0.4861,
+      "num_input_tokens_seen": 1566432,
+      "step": 165
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 168,
+  "num_input_tokens_seen": 1596080,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.467473931042816e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-168/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3714cd435343e43fff99200ac2402d2ab09a63941e76fb7bd73195c6e4f4472
+size 5368

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.986666666666667,
+    "eval_loss": 0.38587039709091187,
+    "eval_runtime": 13.1955,
+    "eval_samples_per_second": 7.578,
+    "eval_steps_per_second": 3.789,
+    "num_input_tokens_seen": 1596080
+}

llamaboard_config.yaml

@@ -0,0 +1,67 @@
+top.booster: auto
+top.checkpoint_path: []
+top.finetuning_type: lora
+top.model_name: Qwen-1.8B
+top.quantization_bit: none
+top.quantization_method: bitsandbytes
+top.rope_scaling: none
+top.template: default
+top.visual_inputs: false
+train.additional_target: ''
+train.badam_mode: layer
+train.badam_switch_interval: 50
+train.badam_switch_mode: ascending
+train.badam_update_ratio: 0.05
+train.batch_size: 2
+train.compute_type: fp16
+train.create_new_adapter: false
+train.cutoff_len: 1024
+train.dataset:
+- glaive_toolcall_en
+train.dataset_dir: data
+train.ds_offload: false
+train.ds_stage: none
+train.freeze_extra_modules: ''
+train.freeze_trainable_layers: 2
+train.freeze_trainable_modules: all
+train.galore_rank: 16
+train.galore_scale: 0.25
+train.galore_target: all
+train.galore_update_interval: 200
+train.gradient_accumulation_steps: 8
+train.learning_rate: 5e-5
+train.logging_steps: 5
+train.lora_alpha: 16
+train.lora_dropout: 0
+train.lora_rank: 8
+train.lora_target: ''
+train.loraplus_lr_ratio: 0
+train.lr_scheduler_type: cosine
+train.mask_history: false
+train.max_grad_norm: '1.0'
+train.max_samples: '100000'
+train.neat_packing: false
+train.neftune_alpha: 0
+train.num_train_epochs: '3.0'
+train.optim: adamw_torch
+train.packing: false
+train.ppo_score_norm: false
+train.ppo_whiten_rewards: false
+train.pref_beta: 0.1
+train.pref_ftx: 0
+train.pref_loss: sigmoid
+train.report_to: false
+train.resize_vocab: false
+train.reward_model: null
+train.save_steps: 100
+train.shift_attn: false
+train.train_on_prompt: false
+train.training_stage: Supervised Fine-Tuning
+train.use_badam: false
+train.use_dora: false
+train.use_galore: false
+train.use_llama_pro: false
+train.use_pissa: false
+train.use_rslora: false
+train.val_size: 0.1
+train.warmup_steps: 0

running_log.txt

@@ -0,0 +1,353 @@
+[INFO|parser.py:355] 2024-09-02 15:49:30,713 >> Process rank: 0, device: cuda:0, n_gpu: 1, distributed training: False, compute dtype: torch.float16
+
+[INFO|tokenization_utils_base.py:2269] 2024-09-02 15:49:32,191 >> loading file qwen.tiktoken from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\qwen.tiktoken
+
+[INFO|tokenization_utils_base.py:2269] 2024-09-02 15:49:32,192 >> loading file added_tokens.json from cache at None
+
+[INFO|tokenization_utils_base.py:2269] 2024-09-02 15:49:32,192 >> loading file special_tokens_map.json from cache at None
+
+[INFO|tokenization_utils_base.py:2269] 2024-09-02 15:49:32,192 >> loading file tokenizer_config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\tokenizer_config.json
+
+[INFO|tokenization_utils_base.py:2269] 2024-09-02 15:49:32,192 >> loading file tokenizer.json from cache at None
+
+[INFO|template.py:270] 2024-09-02 15:49:32,688 >> Add eos token: <|endoftext|>
+
+[INFO|template.py:375] 2024-09-02 15:49:32,689 >> Add pad token: <|endoftext|>
+
+[INFO|loader.py:52] 2024-09-02 15:49:32,690 >> Loading dataset llamafactory/glaive_toolcall_en...
+
+[INFO|configuration_utils.py:733] 2024-09-02 15:49:40,666 >> loading configuration file config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\config.json
+
+[INFO|configuration_utils.py:733] 2024-09-02 15:49:41,313 >> loading configuration file config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\config.json
+
+[INFO|configuration_utils.py:800] 2024-09-02 15:49:41,314 >> Model config QWenConfig {
+  "_name_or_path": "Qwen/Qwen-1_8B",
+  "architectures": [
+    "QWenLMHeadModel"
+  ],
+  "attn_dropout_prob": 0.0,
+  "auto_map": {
+    "AutoConfig": "Qwen/Qwen-1_8B--configuration_qwen.QWenConfig",
+    "AutoModelForCausalLM": "Qwen/Qwen-1_8B--modeling_qwen.QWenLMHeadModel"
+  },
+  "bf16": false,
+  "emb_dropout_prob": 0.0,
+  "fp16": false,
+  "fp32": false,
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "kv_channels": 128,
+  "layer_norm_epsilon": 1e-06,
+  "max_position_embeddings": 8192,
+  "model_type": "qwen",
+  "no_bias": true,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "onnx_safe": null,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 1.0,
+  "scale_attn_weights": true,
+  "seq_length": 8192,
+  "softmax_in_fp32": false,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "QWenTokenizer",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "use_cache_kernel": false,
+  "use_cache_quantization": false,
+  "use_dynamic_ntk": true,
+  "use_flash_attn": "auto",
+  "use_logn_attn": true,
+  "vocab_size": 151936
+}
+
+
+[INFO|modeling_utils.py:3678] 2024-09-02 15:49:41,731 >> loading weights file model.safetensors from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\model.safetensors.index.json
+
+[INFO|modeling_utils.py:1606] 2024-09-02 16:16:57,834 >> Instantiating QWenLMHeadModel model under default dtype torch.float16.
+
+[INFO|configuration_utils.py:1038] 2024-09-02 16:16:57,838 >> Generate config GenerationConfig {}
+
+
+[INFO|modeling_utils.py:4507] 2024-09-02 16:17:02,439 >> All model checkpoint weights were used when initializing QWenLMHeadModel.
+
+
+[INFO|modeling_utils.py:4515] 2024-09-02 16:17:02,440 >> All the weights of QWenLMHeadModel were initialized from the model checkpoint at Qwen/Qwen-1_8B.
+If your task is similar to the task the model of the checkpoint was trained on, you can already use QWenLMHeadModel for predictions without further training.
+
+[INFO|modeling_utils.py:4003] 2024-09-02 16:17:12,457 >> Generation config file not found, using a generation config created from the model config.
+
+[WARNING|checkpointing.py:70] 2024-09-02 16:17:12,489 >> You are using the old GC format, some features (e.g. BAdam) will be invalid.
+
+[INFO|checkpointing.py:103] 2024-09-02 16:17:12,491 >> Gradient checkpointing enabled.
+
+[INFO|attention.py:86] 2024-09-02 16:17:12,493 >> Using vanilla attention implementation.
+
+[INFO|adapter.py:302] 2024-09-02 16:17:12,493 >> Upcasting trainable params to float32.
+
+[INFO|adapter.py:158] 2024-09-02 16:17:12,494 >> Fine-tuning method: LoRA
+
+[INFO|misc.py:56] 2024-09-02 16:17:12,497 >> Found linear modules: c_attn,w2,c_proj,w1
+
+[INFO|loader.py:196] 2024-09-02 16:17:13,257 >> trainable params: 6,709,248 || all params: 1,843,537,920 || trainable%: 0.3639
+
+[INFO|trainer.py:648] 2024-09-02 16:17:13,289 >> Using auto half precision backend
+
+[INFO|trainer.py:2134] 2024-09-02 16:17:13,776 >> ***** Running training *****
+
+[INFO|trainer.py:2135] 2024-09-02 16:17:13,777 >>   Num examples = 900
+
+[INFO|trainer.py:2136] 2024-09-02 16:17:13,778 >>   Num Epochs = 3
+
+[INFO|trainer.py:2137] 2024-09-02 16:17:13,779 >>   Instantaneous batch size per device = 2
+
+[INFO|trainer.py:2140] 2024-09-02 16:17:13,780 >>   Total train batch size (w. parallel, distributed & accumulation) = 16
+
+[INFO|trainer.py:2141] 2024-09-02 16:17:13,781 >>   Gradient Accumulation steps = 8
+
+[INFO|trainer.py:2142] 2024-09-02 16:17:13,782 >>   Total optimization steps = 168
+
+[INFO|trainer.py:2143] 2024-09-02 16:17:13,788 >>   Number of trainable parameters = 6,709,248
+
+[INFO|callbacks.py:319] 2024-09-02 16:17:47,749 >> {'loss': 0.9542, 'learning_rate': 4.9891e-05, 'epoch': 0.09, 'throughput': 1389.23}
+
+[INFO|callbacks.py:319] 2024-09-02 16:18:21,440 >> {'loss': 0.7834, 'learning_rate': 4.9564e-05, 'epoch': 0.18, 'throughput': 1487.79}
+
+[INFO|callbacks.py:319] 2024-09-02 16:18:54,465 >> {'loss': 0.7296, 'learning_rate': 4.9023e-05, 'epoch': 0.27, 'throughput': 1490.51}
+
+[INFO|callbacks.py:319] 2024-09-02 16:19:25,676 >> {'loss': 0.6653, 'learning_rate': 4.8272e-05, 'epoch': 0.36, 'throughput': 1493.24}
+
+[INFO|callbacks.py:319] 2024-09-02 16:19:55,399 >> {'loss': 0.6830, 'learning_rate': 4.7317e-05, 'epoch': 0.44, 'throughput': 1513.54}
+
+[INFO|callbacks.py:319] 2024-09-02 16:20:25,205 >> {'loss': 0.5651, 'learning_rate': 4.6168e-05, 'epoch': 0.53, 'throughput': 1533.3}
+
+[INFO|callbacks.py:319] 2024-09-02 16:20:56,227 >> {'loss': 0.5411, 'learning_rate': 4.4834e-05, 'epoch': 0.62, 'throughput': 1520.29}
+
+[INFO|callbacks.py:319] 2024-09-02 16:21:25,069 >> {'loss': 0.5720, 'learning_rate': 4.3326e-05, 'epoch': 0.71, 'throughput': 1523.32}
+
+[INFO|callbacks.py:319] 2024-09-02 16:21:55,779 >> {'loss': 0.6106, 'learning_rate': 4.1659e-05, 'epoch': 0.80, 'throughput': 1528.21}
+
+[INFO|callbacks.py:319] 2024-09-02 16:22:26,052 >> {'loss': 0.5552, 'learning_rate': 3.9846e-05, 'epoch': 0.89, 'throughput': 1529.26}
+
+[INFO|callbacks.py:319] 2024-09-02 16:22:58,326 >> {'loss': 0.5659, 'learning_rate': 3.7903e-05, 'epoch': 0.98, 'throughput': 1528.57}
+
+[INFO|callbacks.py:319] 2024-09-02 16:23:31,779 >> {'loss': 0.5800, 'learning_rate': 3.5847e-05, 'epoch': 1.07, 'throughput': 1527.01}
+
+[INFO|callbacks.py:319] 2024-09-02 16:24:02,611 >> {'loss': 0.5429, 'learning_rate': 3.3697e-05, 'epoch': 1.16, 'throughput': 1529.65}
+
+[INFO|callbacks.py:319] 2024-09-02 16:24:31,189 >> {'loss': 0.4384, 'learning_rate': 3.1470e-05, 'epoch': 1.24, 'throughput': 1532.32}
+
+[INFO|callbacks.py:319] 2024-09-02 16:25:02,456 >> {'loss': 0.5749, 'learning_rate': 2.9188e-05, 'epoch': 1.33, 'throughput': 1525.37}
+
+[INFO|callbacks.py:319] 2024-09-02 16:25:30,813 >> {'loss': 0.4222, 'learning_rate': 2.6868e-05, 'epoch': 1.42, 'throughput': 1514.93}
+
+[INFO|callbacks.py:319] 2024-09-02 16:26:01,816 >> {'loss': 0.5560, 'learning_rate': 2.4533e-05, 'epoch': 1.51, 'throughput': 1519.0}
+
+[INFO|callbacks.py:319] 2024-09-02 16:26:33,656 >> {'loss': 0.4963, 'learning_rate': 2.2201e-05, 'epoch': 1.60, 'throughput': 1522.01}
+
+[INFO|callbacks.py:319] 2024-09-02 16:27:08,164 >> {'loss': 0.5085, 'learning_rate': 1.9894e-05, 'epoch': 1.69, 'throughput': 1522.94}
+
+[INFO|callbacks.py:319] 2024-09-02 16:27:37,551 >> {'loss': 0.4726, 'learning_rate': 1.7631e-05, 'epoch': 1.78, 'throughput': 1522.11}
+
+[INFO|trainer.py:3819] 2024-09-02 16:27:37,560 >> 
+***** Running Evaluation *****
+
+[INFO|trainer.py:3821] 2024-09-02 16:27:37,560 >>   Num examples = 100
+
+[INFO|trainer.py:3824] 2024-09-02 16:27:37,561 >>   Batch size = 2
+
+[INFO|trainer.py:3503] 2024-09-02 16:27:48,996 >> Saving model checkpoint to saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-100
+
+[INFO|configuration_utils.py:733] 2024-09-02 16:27:51,471 >> loading configuration file config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\config.json
+
+[INFO|configuration_utils.py:800] 2024-09-02 16:27:51,473 >> Model config QWenConfig {
+  "architectures": [
+    "QWenLMHeadModel"
+  ],
+  "attn_dropout_prob": 0.0,
+  "auto_map": {
+    "AutoConfig": "Qwen/Qwen-1_8B--configuration_qwen.QWenConfig",
+    "AutoModelForCausalLM": "Qwen/Qwen-1_8B--modeling_qwen.QWenLMHeadModel"
+  },
+  "bf16": false,
+  "emb_dropout_prob": 0.0,
+  "fp16": false,
+  "fp32": false,
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "kv_channels": 128,
+  "layer_norm_epsilon": 1e-06,
+  "max_position_embeddings": 8192,
+  "model_type": "qwen",
+  "no_bias": true,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "onnx_safe": null,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 1.0,
+  "scale_attn_weights": true,
+  "seq_length": 8192,
+  "softmax_in_fp32": false,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "QWenTokenizer",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "use_cache_kernel": false,
+  "use_cache_quantization": false,
+  "use_dynamic_ntk": true,
+  "use_flash_attn": "auto",
+  "use_logn_attn": true,
+  "vocab_size": 151936
+}
+
+
+[INFO|tokenization_utils_base.py:2684] 2024-09-02 16:27:51,605 >> tokenizer config file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-100\tokenizer_config.json
+
+[INFO|tokenization_utils_base.py:2693] 2024-09-02 16:27:51,607 >> Special tokens file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-100\special_tokens_map.json
+
+[INFO|callbacks.py:319] 2024-09-02 16:28:22,125 >> {'loss': 0.4874, 'learning_rate': 1.5433e-05, 'epoch': 1.87, 'throughput': 1490.45}
+
+[INFO|callbacks.py:319] 2024-09-02 16:28:55,988 >> {'loss': 0.5235, 'learning_rate': 1.3318e-05, 'epoch': 1.96, 'throughput': 1491.79}
+
+[INFO|callbacks.py:319] 2024-09-02 16:29:28,437 >> {'loss': 0.4984, 'learning_rate': 1.1306e-05, 'epoch': 2.04, 'throughput': 1493.02}
+
+[INFO|callbacks.py:319] 2024-09-02 16:30:01,857 >> {'loss': 0.4777, 'learning_rate': 9.4128e-06, 'epoch': 2.13, 'throughput': 1498.71}
+
+[INFO|callbacks.py:319] 2024-09-02 16:30:30,832 >> {'loss': 0.4918, 'learning_rate': 7.6560e-06, 'epoch': 2.22, 'throughput': 1498.79}
+
+[INFO|callbacks.py:319] 2024-09-02 16:31:00,464 >> {'loss': 0.4573, 'learning_rate': 6.0507e-06, 'epoch': 2.31, 'throughput': 1494.08}
+
+[INFO|callbacks.py:319] 2024-09-02 16:31:30,565 >> {'loss': 0.4549, 'learning_rate': 4.6110e-06, 'epoch': 2.40, 'throughput': 1494.64}
+
+[INFO|callbacks.py:319] 2024-09-02 16:32:01,843 >> {'loss': 0.5422, 'learning_rate': 3.3494e-06, 'epoch': 2.49, 'throughput': 1497.43}
+
+[INFO|callbacks.py:319] 2024-09-02 16:32:34,313 >> {'loss': 0.4968, 'learning_rate': 2.2769e-06, 'epoch': 2.58, 'throughput': 1500.22}
+
+[INFO|callbacks.py:319] 2024-09-02 16:33:04,696 >> {'loss': 0.4361, 'learning_rate': 1.4029e-06, 'epoch': 2.67, 'throughput': 1497.23}
+
+[INFO|callbacks.py:319] 2024-09-02 16:33:34,901 >> {'loss': 0.4517, 'learning_rate': 7.3509e-07, 'epoch': 2.76, 'throughput': 1498.31}
+
+[INFO|callbacks.py:319] 2024-09-02 16:34:05,001 >> {'loss': 0.4099, 'learning_rate': 2.7923e-07, 'epoch': 2.84, 'throughput': 1498.38}
+
+[INFO|callbacks.py:319] 2024-09-02 16:34:34,884 >> {'loss': 0.4861, 'learning_rate': 3.9330e-08, 'epoch': 2.93, 'throughput': 1504.61}
+
+[INFO|trainer.py:3503] 2024-09-02 16:34:53,706 >> Saving model checkpoint to saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-168
+
+[INFO|configuration_utils.py:733] 2024-09-02 16:34:56,148 >> loading configuration file config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\config.json
+
+[INFO|configuration_utils.py:800] 2024-09-02 16:34:56,150 >> Model config QWenConfig {
+  "architectures": [
+    "QWenLMHeadModel"
+  ],
+  "attn_dropout_prob": 0.0,
+  "auto_map": {
+    "AutoConfig": "Qwen/Qwen-1_8B--configuration_qwen.QWenConfig",
+    "AutoModelForCausalLM": "Qwen/Qwen-1_8B--modeling_qwen.QWenLMHeadModel"
+  },
+  "bf16": false,
+  "emb_dropout_prob": 0.0,
+  "fp16": false,
+  "fp32": false,
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "kv_channels": 128,
+  "layer_norm_epsilon": 1e-06,
+  "max_position_embeddings": 8192,
+  "model_type": "qwen",
+  "no_bias": true,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "onnx_safe": null,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 1.0,
+  "scale_attn_weights": true,
+  "seq_length": 8192,
+  "softmax_in_fp32": false,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "QWenTokenizer",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "use_cache_kernel": false,
+  "use_cache_quantization": false,
+  "use_dynamic_ntk": true,
+  "use_flash_attn": "auto",
+  "use_logn_attn": true,
+  "vocab_size": 151936
+}
+
+
+[INFO|tokenization_utils_base.py:2684] 2024-09-02 16:34:56,239 >> tokenizer config file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-168\tokenizer_config.json
+
+[INFO|tokenization_utils_base.py:2693] 2024-09-02 16:34:56,240 >> Special tokens file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\checkpoint-168\special_tokens_map.json
+
+[INFO|trainer.py:2394] 2024-09-02 16:34:56,859 >> 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
+[INFO|trainer.py:3503] 2024-09-02 16:34:56,864 >> Saving model checkpoint to saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54
+
+[INFO|configuration_utils.py:733] 2024-09-02 16:34:58,089 >> loading configuration file config.json from cache at C:\Users\22320\.cache\huggingface\hub\models--Qwen--Qwen-1_8B\snapshots\fa6e214ccbbc6a55235c26ef406355b6bfdf5eed\config.json
+
+[INFO|configuration_utils.py:800] 2024-09-02 16:34:58,092 >> Model config QWenConfig {
+  "architectures": [
+    "QWenLMHeadModel"
+  ],
+  "attn_dropout_prob": 0.0,
+  "auto_map": {
+    "AutoConfig": "Qwen/Qwen-1_8B--configuration_qwen.QWenConfig",
+    "AutoModelForCausalLM": "Qwen/Qwen-1_8B--modeling_qwen.QWenLMHeadModel"
+  },
+  "bf16": false,
+  "emb_dropout_prob": 0.0,
+  "fp16": false,
+  "fp32": false,
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "kv_channels": 128,
+  "layer_norm_epsilon": 1e-06,
+  "max_position_embeddings": 8192,
+  "model_type": "qwen",
+  "no_bias": true,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "onnx_safe": null,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 1.0,
+  "scale_attn_weights": true,
+  "seq_length": 8192,
+  "softmax_in_fp32": false,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "QWenTokenizer",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "use_cache_kernel": false,
+  "use_cache_quantization": false,
+  "use_dynamic_ntk": true,
+  "use_flash_attn": "auto",
+  "use_logn_attn": true,
+  "vocab_size": 151936
+}
+
+
+[INFO|tokenization_utils_base.py:2684] 2024-09-02 16:34:58,235 >> tokenizer config file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\tokenizer_config.json
+
+[INFO|tokenization_utils_base.py:2693] 2024-09-02 16:34:58,238 >> Special tokens file saved in saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54\special_tokens_map.json
+
+[WARNING|ploting.py:89] 2024-09-02 16:34:59,062 >> No metric eval_accuracy to plot.
+
+[INFO|trainer.py:3819] 2024-09-02 16:34:59,082 >> 
+***** Running Evaluation *****
+
+[INFO|trainer.py:3821] 2024-09-02 16:34:59,084 >>   Num examples = 100
+
+[INFO|trainer.py:3824] 2024-09-02 16:34:59,086 >>   Batch size = 2
+
+[INFO|modelcard.py:449] 2024-09-02 16:35:12,284 >> Dropping the following result as it does not have all the necessary fields:
+{'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}}
+

special_tokens_map.json

@@ -0,0 +1,10 @@
+{
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

tokenization_qwen.py

@@ -0,0 +1,276 @@
+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Tokenization classes for QWen."""
+
+import base64
+import logging
+import os
+import unicodedata
+from typing import Collection, Dict, List, Set, Tuple, Union
+
+import tiktoken
+from transformers import PreTrainedTokenizer, AddedToken
+
+logger = logging.getLogger(__name__)
+
+
+VOCAB_FILES_NAMES = {"vocab_file": "qwen.tiktoken"}
+
+PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
+ENDOFTEXT = "<|endoftext|>"
+IMSTART = "<|im_start|>"
+IMEND = "<|im_end|>"
+# as the default behavior is changed to allow special tokens in
+# regular texts, the surface forms of special tokens need to be
+# as different as possible to minimize the impact
+EXTRAS = tuple((f"<|extra_{i}|>" for i in range(205)))
+# changed to use actual index to avoid misconfiguration with vocabulary expansion
+SPECIAL_START_ID = 151643
+SPECIAL_TOKENS = tuple(
+    enumerate(
+        (
+            (
+                ENDOFTEXT,
+                IMSTART,
+                IMEND,
+            )
+            + EXTRAS
+        ),
+        start=SPECIAL_START_ID,
+    )
+)
+SPECIAL_TOKENS_SET = set(t for i, t in SPECIAL_TOKENS)
+
+
+def _load_tiktoken_bpe(tiktoken_bpe_file: str) -> Dict[bytes, int]:
+    with open(tiktoken_bpe_file, "rb") as f:
+        contents = f.read()
+    return {
+        base64.b64decode(token): int(rank)
+        for token, rank in (line.split() for line in contents.splitlines() if line)
+    }
+
+
+class QWenTokenizer(PreTrainedTokenizer):
+    """QWen tokenizer."""
+
+    vocab_files_names = VOCAB_FILES_NAMES
+
+    def __init__(
+        self,
+        vocab_file,
+        errors="replace",
+        extra_vocab_file=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        # how to handle errors in decoding UTF-8 byte sequences
+        # use ignore if you are in streaming inference
+        self.errors = errors  
+
+        self.mergeable_ranks = _load_tiktoken_bpe(vocab_file)  # type: Dict[bytes, int]
+        self.special_tokens = {
+            token: index
+            for index, token in SPECIAL_TOKENS
+        }
+
+        # try load extra vocab from file
+        if extra_vocab_file is not None:
+            used_ids = set(self.mergeable_ranks.values()) | set(self.special_tokens.values())
+            extra_mergeable_ranks = _load_tiktoken_bpe(extra_vocab_file)
+            for token, index in extra_mergeable_ranks.items():
+                if token in self.mergeable_ranks:
+                    logger.info(f"extra token {token} exists, skipping")
+                    continue
+                if index in used_ids:
+                    logger.info(f'the index {index} for extra token {token} exists, skipping')
+                    continue
+                self.mergeable_ranks[token] = index
+            # the index may be sparse after this, but don't worry tiktoken.Encoding will handle this
+
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        assert (
+            len(self.mergeable_ranks) + len(self.special_tokens) == enc.n_vocab
+        ), f"{len(self.mergeable_ranks) + len(self.special_tokens)} != {enc.n_vocab} in encoding"
+
+        self.decoder = {
+            v: k for k, v in self.mergeable_ranks.items()
+        }  # type: dict[int, bytes|str]
+        self.decoder.update({v: k for k, v in self.special_tokens.items()})
+
+        self.tokenizer = enc  # type: tiktoken.Encoding
+
+        self.eod_id = self.tokenizer.eot_token
+        self.im_start_id = self.special_tokens[IMSTART]
+        self.im_end_id = self.special_tokens[IMEND]
+
+    def __getstate__(self):
+        # for pickle lovers
+        state = self.__dict__.copy()
+        del state["tokenizer"]
+        return state
+
+    def __setstate__(self, state):
+        # tokenizer is not python native; don't pass it; rebuild it
+        self.__dict__.update(state)
+        enc = tiktoken.Encoding(
+            "Qwen",
+            pat_str=PAT_STR,
+            mergeable_ranks=self.mergeable_ranks,
+            special_tokens=self.special_tokens,
+        )
+        self.tokenizer = enc
+
+    def __len__(self) -> int:
+        return self.tokenizer.n_vocab
+
+    def get_vocab(self) -> Dict[bytes, int]:
+        return self.mergeable_ranks
+
+    def convert_tokens_to_ids(
+        self, tokens: Union[bytes, str, List[Union[bytes, str]]]
+    ) -> List[int]:
+        ids = []
+        if isinstance(tokens, (str, bytes)):
+            if tokens in self.special_tokens:
+                return self.special_tokens[tokens]
+            else:
+                return self.mergeable_ranks.get(tokens)
+        for token in tokens:
+            if token in self.special_tokens:
+                ids.append(self.special_tokens[token])
+            else:
+                ids.append(self.mergeable_ranks.get(token))
+        return ids
+
+    def _add_tokens(
+        self,
+        new_tokens: Union[List[str], List[AddedToken]],
+        special_tokens: bool = False,
+    ) -> int:
+        if not special_tokens and new_tokens:
+            raise ValueError("Adding regular tokens is not supported")
+        for token in new_tokens:
+            surface_form = token.content if isinstance(token, AddedToken) else token
+            if surface_form not in SPECIAL_TOKENS_SET:
+                raise ValueError("Adding unknown special tokens is not supported")
+        return 0
+
+    def save_vocabulary(self, save_directory: str, **kwargs) -> Tuple[str]:
+        """
+        Save only the vocabulary of the tokenizer (vocabulary).
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        file_path = os.path.join(save_directory, "qwen.tiktoken")
+        with open(file_path, "w", encoding="utf8") as w:
+            for k, v in self.mergeable_ranks.items():
+                line = base64.b64encode(k).decode("utf8") + " " + str(v) + "\n"
+                w.write(line)
+        return (file_path,)
+
+    def tokenize(
+        self,
+        text: str,
+        allowed_special: Union[Set, str] = "all",
+        disallowed_special: Union[Collection, str] = (),
+        **kwargs,
+    ) -> List[Union[bytes, str]]:
+        """
+        Converts a string in a sequence of tokens.
+
+        Args:
+            text (`str`):
+                The sequence to be encoded.
+            allowed_special (`Literal["all"]` or `set`):
+                The surface forms of the tokens to be encoded as special tokens in regular texts.
+                Default to "all".
+            disallowed_special (`Literal["all"]` or `Collection`):
+                The surface forms of the tokens that should not be in regular texts and trigger errors.
+                Default to an empty tuple.
+
+            kwargs (additional keyword arguments, *optional*):
+                Will be passed to the underlying model specific encode method.
+
+        Returns:
+            `List[bytes|str]`: The list of tokens.
+        """
+        tokens = []
+        text = unicodedata.normalize("NFC", text)
+
+        # this implementation takes a detour: text -> token id -> token surface forms
+        for t in self.tokenizer.encode(
+            text, allowed_special=allowed_special, disallowed_special=disallowed_special
+        ):
+            tokens.append(self.decoder[t])
+        return tokens
+
+    def convert_tokens_to_string(self, tokens: List[Union[bytes, str]]) -> str:
+        """
+        Converts a sequence of tokens in a single string.
+        """
+        text = ""
+        temp = b""
+        for t in tokens:
+            if isinstance(t, str):
+                if temp:
+                    text += temp.decode("utf-8", errors=self.errors)
+                    temp = b""
+                text += t
+            elif isinstance(t, bytes):
+                temp += t
+            else:
+                raise TypeError("token should only be of type types or str")
+        if temp:
+            text += temp.decode("utf-8", errors=self.errors)
+        return text
+
+    @property
+    def vocab_size(self):
+        return self.tokenizer.n_vocab
+
+    def _convert_id_to_token(self, index: int) -> Union[bytes, str]:
+        """Converts an id to a token, special tokens included"""
+        if index in self.decoder:
+            return self.decoder[index]
+        raise ValueError("unknown ids")
+
+    def _convert_token_to_id(self, token: Union[bytes, str]) -> int:
+        """Converts a token to an id using the vocab, special tokens included"""
+        if token in self.special_tokens:
+            return self.special_tokens[token]
+        if token in self.mergeable_ranks:
+            return self.mergeable_ranks[token]
+        raise ValueError("unknown token")
+
+    def _tokenize(self, text: str, **kwargs):
+        """
+        Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
+        vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
+
+        Do NOT take care of added tokens.
+        """
+        raise NotImplementedError
+
+    def _decode(
+        self,
+        token_ids: Union[int, List[int]],
+        skip_special_tokens: bool = False,
+        errors: str = None,
+        **kwargs,
+    ) -> str:
+        if isinstance(token_ids, int):
+            token_ids = [token_ids]
+        if skip_special_tokens:
+            token_ids = [i for i in token_ids if i < self.eod_id]
+        return self.tokenizer.decode(token_ids, errors=errors or self.errors)

tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "added_tokens_decoder": {},
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "chat_template": "{% 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_message + '\n' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '\nAssistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + '\n' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 2.986666666666667,
+    "num_input_tokens_seen": 1596080,
+    "total_flos": 1.467473931042816e+16,
+    "train_loss": 0.5455739086582547,
+    "train_runtime": 1063.0707,
+    "train_samples_per_second": 2.54,
+    "train_steps_per_second": 0.158
+}

trainer_log.jsonl

@@ -0,0 +1,35 @@
+{"current_steps": 5, "total_steps": 168, "loss": 0.9542, "learning_rate": 4.989080197352834e-05, "epoch": 0.08888888888888889, "percentage": 2.98, "elapsed_time": "0:00:33", "remaining_time": "0:18:26", "throughput": 1389.23, "total_tokens": 47168}
+{"current_steps": 10, "total_steps": 168, "loss": 0.7834, "learning_rate": 4.956416183083221e-05, "epoch": 0.17777777777777778, "percentage": 5.95, "elapsed_time": "0:01:07", "remaining_time": "0:17:48", "throughput": 1487.79, "total_tokens": 100640}
+{"current_steps": 15, "total_steps": 168, "loss": 0.7296, "learning_rate": 4.9022933048627496e-05, "epoch": 0.26666666666666666, "percentage": 8.93, "elapsed_time": "0:01:40", "remaining_time": "0:17:06", "throughput": 1490.51, "total_tokens": 150048}
+{"current_steps": 20, "total_steps": 168, "loss": 0.6653, "learning_rate": 4.827184371610511e-05, "epoch": 0.35555555555555557, "percentage": 11.9, "elapsed_time": "0:02:11", "remaining_time": "0:16:15", "throughput": 1493.24, "total_tokens": 196928}
+{"current_steps": 25, "total_steps": 168, "loss": 0.683, "learning_rate": 4.731745523109029e-05, "epoch": 0.4444444444444444, "percentage": 14.88, "elapsed_time": "0:02:41", "remaining_time": "0:15:24", "throughput": 1513.54, "total_tokens": 244592}
+{"current_steps": 30, "total_steps": 168, "loss": 0.5651, "learning_rate": 4.6168104980707107e-05, "epoch": 0.5333333333333333, "percentage": 17.86, "elapsed_time": "0:03:11", "remaining_time": "0:14:40", "throughput": 1533.3, "total_tokens": 293488}
+{"current_steps": 35, "total_steps": 168, "loss": 0.5411, "learning_rate": 4.4833833507280884e-05, "epoch": 0.6222222222222222, "percentage": 20.83, "elapsed_time": "0:03:42", "remaining_time": "0:14:05", "throughput": 1520.29, "total_tokens": 338160}
+{"current_steps": 40, "total_steps": 168, "loss": 0.572, "learning_rate": 4.332629679574566e-05, "epoch": 0.7111111111111111, "percentage": 23.81, "elapsed_time": "0:04:11", "remaining_time": "0:13:24", "throughput": 1523.32, "total_tokens": 382768}
+{"current_steps": 45, "total_steps": 168, "loss": 0.6106, "learning_rate": 4.16586644488001e-05, "epoch": 0.8, "percentage": 26.79, "elapsed_time": "0:04:41", "remaining_time": "0:12:50", "throughput": 1528.21, "total_tokens": 430928}
+{"current_steps": 50, "total_steps": 168, "loss": 0.5552, "learning_rate": 3.9845504639337535e-05, "epoch": 0.8888888888888888, "percentage": 29.76, "elapsed_time": "0:05:12", "remaining_time": "0:12:16", "throughput": 1529.26, "total_tokens": 477520}
+{"current_steps": 55, "total_steps": 168, "loss": 0.5659, "learning_rate": 3.790265684518767e-05, "epoch": 0.9777777777777777, "percentage": 32.74, "elapsed_time": "0:05:44", "remaining_time": "0:11:47", "throughput": 1528.57, "total_tokens": 526640}
+{"current_steps": 60, "total_steps": 168, "loss": 0.58, "learning_rate": 3.5847093477938956e-05, "epoch": 1.0666666666666667, "percentage": 35.71, "elapsed_time": "0:06:17", "remaining_time": "0:11:20", "throughput": 1527.01, "total_tokens": 577184}
+{"current_steps": 65, "total_steps": 168, "loss": 0.5429, "learning_rate": 3.369677161463068e-05, "epoch": 1.1555555555555554, "percentage": 38.69, "elapsed_time": "0:06:48", "remaining_time": "0:10:47", "throughput": 1529.65, "total_tokens": 625344}
+{"current_steps": 70, "total_steps": 168, "loss": 0.4384, "learning_rate": 3.147047612756302e-05, "epoch": 1.2444444444444445, "percentage": 41.67, "elapsed_time": "0:07:17", "remaining_time": "0:10:12", "throughput": 1532.32, "total_tokens": 670224}
+{"current_steps": 75, "total_steps": 168, "loss": 0.5749, "learning_rate": 2.918765558261841e-05, "epoch": 1.3333333333333333, "percentage": 44.64, "elapsed_time": "0:07:48", "remaining_time": "0:09:41", "throughput": 1525.37, "total_tokens": 714880}
+{"current_steps": 80, "total_steps": 168, "loss": 0.4222, "learning_rate": 2.686825233966061e-05, "epoch": 1.4222222222222223, "percentage": 47.62, "elapsed_time": "0:08:17", "remaining_time": "0:09:06", "throughput": 1514.93, "total_tokens": 752944}
+{"current_steps": 85, "total_steps": 168, "loss": 0.556, "learning_rate": 2.4532528339227452e-05, "epoch": 1.511111111111111, "percentage": 50.6, "elapsed_time": "0:08:48", "remaining_time": "0:08:35", "throughput": 1519.0, "total_tokens": 802064}
+{"current_steps": 90, "total_steps": 168, "loss": 0.4963, "learning_rate": 2.2200888097417307e-05, "epoch": 1.6, "percentage": 53.57, "elapsed_time": "0:09:19", "remaining_time": "0:08:05", "throughput": 1522.01, "total_tokens": 852112}
+{"current_steps": 95, "total_steps": 168, "loss": 0.5085, "learning_rate": 1.9893700455257996e-05, "epoch": 1.6888888888888889, "percentage": 56.55, "elapsed_time": "0:09:54", "remaining_time": "0:07:36", "throughput": 1522.94, "total_tokens": 905184}
+{"current_steps": 100, "total_steps": 168, "loss": 0.4726, "learning_rate": 1.7631120639727393e-05, "epoch": 1.7777777777777777, "percentage": 59.52, "elapsed_time": "0:10:23", "remaining_time": "0:07:04", "throughput": 1522.11, "total_tokens": 949424}
+{"current_steps": 100, "total_steps": 168, "eval_loss": 0.39410004019737244, "epoch": 1.7777777777777777, "percentage": 59.52, "elapsed_time": "0:10:35", "remaining_time": "0:07:11", "throughput": 1494.69, "total_tokens": 949424}
+{"current_steps": 105, "total_steps": 168, "loss": 0.4874, "learning_rate": 1.5432914190872757e-05, "epoch": 1.8666666666666667, "percentage": 62.5, "elapsed_time": "0:11:08", "remaining_time": "0:06:40", "throughput": 1490.45, "total_tokens": 996112}
+{"current_steps": 110, "total_steps": 168, "loss": 0.5235, "learning_rate": 1.331828429317345e-05, "epoch": 1.9555555555555557, "percentage": 65.48, "elapsed_time": "0:11:42", "remaining_time": "0:06:10", "throughput": 1491.79, "total_tokens": 1047520}
+{"current_steps": 115, "total_steps": 168, "loss": 0.4984, "learning_rate": 1.130570401955322e-05, "epoch": 2.0444444444444443, "percentage": 68.45, "elapsed_time": "0:12:14", "remaining_time": "0:05:38", "throughput": 1493.02, "total_tokens": 1096832}
+{"current_steps": 120, "total_steps": 168, "loss": 0.4777, "learning_rate": 9.412754953531663e-06, "epoch": 2.1333333333333333, "percentage": 71.43, "elapsed_time": "0:12:48", "remaining_time": "0:05:07", "throughput": 1498.71, "total_tokens": 1151104}
+{"current_steps": 125, "total_steps": 168, "loss": 0.4918, "learning_rate": 7.65597359928646e-06, "epoch": 2.2222222222222223, "percentage": 74.4, "elapsed_time": "0:13:17", "remaining_time": "0:04:34", "throughput": 1498.79, "total_tokens": 1194592}
+{"current_steps": 130, "total_steps": 168, "loss": 0.4573, "learning_rate": 6.050706921363672e-06, "epoch": 2.311111111111111, "percentage": 77.38, "elapsed_time": "0:13:46", "remaining_time": "0:04:01", "throughput": 1494.08, "total_tokens": 1235104}
+{"current_steps": 135, "total_steps": 168, "loss": 0.4549, "learning_rate": 4.610978276018496e-06, "epoch": 2.4, "percentage": 80.36, "elapsed_time": "0:14:16", "remaining_time": "0:03:29", "throughput": 1494.64, "total_tokens": 1280560}
+{"current_steps": 140, "total_steps": 168, "loss": 0.5422, "learning_rate": 3.3493649053890326e-06, "epoch": 2.488888888888889, "percentage": 83.33, "elapsed_time": "0:14:48", "remaining_time": "0:02:57", "throughput": 1497.43, "total_tokens": 1329792}
+{"current_steps": 145, "total_steps": 168, "loss": 0.4968, "learning_rate": 2.2768880646947268e-06, "epoch": 2.5777777777777775, "percentage": 86.31, "elapsed_time": "0:15:20", "remaining_time": "0:02:26", "throughput": 1500.22, "total_tokens": 1380976}
+{"current_steps": 150, "total_steps": 168, "loss": 0.4361, "learning_rate": 1.4029167422908107e-06, "epoch": 2.6666666666666665, "percentage": 89.29, "elapsed_time": "0:15:50", "remaining_time": "0:01:54", "throughput": 1497.23, "total_tokens": 1423712}
+{"current_steps": 155, "total_steps": 168, "loss": 0.4517, "learning_rate": 7.350858136652261e-07, "epoch": 2.7555555555555555, "percentage": 92.26, "elapsed_time": "0:16:21", "remaining_time": "0:01:22", "throughput": 1498.31, "total_tokens": 1470000}
+{"current_steps": 160, "total_steps": 168, "loss": 0.4099, "learning_rate": 2.7922934437178695e-07, "epoch": 2.8444444444444446, "percentage": 95.24, "elapsed_time": "0:16:51", "remaining_time": "0:00:50", "throughput": 1498.38, "total_tokens": 1515168}
+{"current_steps": 165, "total_steps": 168, "loss": 0.4861, "learning_rate": 3.9329624554584884e-08, "epoch": 2.9333333333333336, "percentage": 98.21, "elapsed_time": "0:17:21", "remaining_time": "0:00:18", "throughput": 1504.61, "total_tokens": 1566432}
+{"current_steps": 168, "total_steps": 168, "epoch": 2.986666666666667, "percentage": 100.0, "elapsed_time": "0:17:43", "remaining_time": "0:00:00", "throughput": 1501.4, "total_tokens": 1596080}

trainer_state.json

@@ -0,0 +1,316 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.986666666666667,
+  "eval_steps": 100,
+  "global_step": 168,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.7499012351036072,
+      "learning_rate": 4.989080197352834e-05,
+      "loss": 0.9542,
+      "num_input_tokens_seen": 47168,
+      "step": 5
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.6703861355781555,
+      "learning_rate": 4.956416183083221e-05,
+      "loss": 0.7834,
+      "num_input_tokens_seen": 100640,
+      "step": 10
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.6384798884391785,
+      "learning_rate": 4.9022933048627496e-05,
+      "loss": 0.7296,
+      "num_input_tokens_seen": 150048,
+      "step": 15
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.36834755539894104,
+      "learning_rate": 4.827184371610511e-05,
+      "loss": 0.6653,
+      "num_input_tokens_seen": 196928,
+      "step": 20
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.2962658405303955,
+      "learning_rate": 4.731745523109029e-05,
+      "loss": 0.683,
+      "num_input_tokens_seen": 244592,
+      "step": 25
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.33365580439567566,
+      "learning_rate": 4.6168104980707107e-05,
+      "loss": 0.5651,
+      "num_input_tokens_seen": 293488,
+      "step": 30
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.4523426294326782,
+      "learning_rate": 4.4833833507280884e-05,
+      "loss": 0.5411,
+      "num_input_tokens_seen": 338160,
+      "step": 35
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.41214215755462646,
+      "learning_rate": 4.332629679574566e-05,
+      "loss": 0.572,
+      "num_input_tokens_seen": 382768,
+      "step": 40
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3807964026927948,
+      "learning_rate": 4.16586644488001e-05,
+      "loss": 0.6106,
+      "num_input_tokens_seen": 430928,
+      "step": 45
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.45064786076545715,
+      "learning_rate": 3.9845504639337535e-05,
+      "loss": 0.5552,
+      "num_input_tokens_seen": 477520,
+      "step": 50
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.4501620829105377,
+      "learning_rate": 3.790265684518767e-05,
+      "loss": 0.5659,
+      "num_input_tokens_seen": 526640,
+      "step": 55
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.24308599531650543,
+      "learning_rate": 3.5847093477938956e-05,
+      "loss": 0.58,
+      "num_input_tokens_seen": 577184,
+      "step": 60
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 0.33691608905792236,
+      "learning_rate": 3.369677161463068e-05,
+      "loss": 0.5429,
+      "num_input_tokens_seen": 625344,
+      "step": 65
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 0.2771952748298645,
+      "learning_rate": 3.147047612756302e-05,
+      "loss": 0.4384,
+      "num_input_tokens_seen": 670224,
+      "step": 70
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4307265281677246,
+      "learning_rate": 2.918765558261841e-05,
+      "loss": 0.5749,
+      "num_input_tokens_seen": 714880,
+      "step": 75
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 0.4149227440357208,
+      "learning_rate": 2.686825233966061e-05,
+      "loss": 0.4222,
+      "num_input_tokens_seen": 752944,
+      "step": 80
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 0.44140687584877014,
+      "learning_rate": 2.4532528339227452e-05,
+      "loss": 0.556,
+      "num_input_tokens_seen": 802064,
+      "step": 85
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.376556396484375,
+      "learning_rate": 2.2200888097417307e-05,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 852112,
+      "step": 90
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 0.3596157431602478,
+      "learning_rate": 1.9893700455257996e-05,
+      "loss": 0.5085,
+      "num_input_tokens_seen": 905184,
+      "step": 95
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.4776301085948944,
+      "learning_rate": 1.7631120639727393e-05,
+      "loss": 0.4726,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "eval_loss": 0.39410004019737244,
+      "eval_runtime": 11.4337,
+      "eval_samples_per_second": 8.746,
+      "eval_steps_per_second": 4.373,
+      "num_input_tokens_seen": 949424,
+      "step": 100
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.26946938037872314,
+      "learning_rate": 1.5432914190872757e-05,
+      "loss": 0.4874,
+      "num_input_tokens_seen": 996112,
+      "step": 105
+    },
+    {
+      "epoch": 1.9555555555555557,
+      "grad_norm": 0.37138107419013977,
+      "learning_rate": 1.331828429317345e-05,
+      "loss": 0.5235,
+      "num_input_tokens_seen": 1047520,
+      "step": 110
+    },
+    {
+      "epoch": 2.0444444444444443,
+      "grad_norm": 0.33912232518196106,
+      "learning_rate": 1.130570401955322e-05,
+      "loss": 0.4984,
+      "num_input_tokens_seen": 1096832,
+      "step": 115
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.4000737965106964,
+      "learning_rate": 9.412754953531663e-06,
+      "loss": 0.4777,
+      "num_input_tokens_seen": 1151104,
+      "step": 120
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.5536447167396545,
+      "learning_rate": 7.65597359928646e-06,
+      "loss": 0.4918,
+      "num_input_tokens_seen": 1194592,
+      "step": 125
+    },
+    {
+      "epoch": 2.311111111111111,
+      "grad_norm": 0.42406854033470154,
+      "learning_rate": 6.050706921363672e-06,
+      "loss": 0.4573,
+      "num_input_tokens_seen": 1235104,
+      "step": 130
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.37090376019477844,
+      "learning_rate": 4.610978276018496e-06,
+      "loss": 0.4549,
+      "num_input_tokens_seen": 1280560,
+      "step": 135
+    },
+    {
+      "epoch": 2.488888888888889,
+      "grad_norm": 0.3070351779460907,
+      "learning_rate": 3.3493649053890326e-06,
+      "loss": 0.5422,
+      "num_input_tokens_seen": 1329792,
+      "step": 140
+    },
+    {
+      "epoch": 2.5777777777777775,
+      "grad_norm": 0.3824547529220581,
+      "learning_rate": 2.2768880646947268e-06,
+      "loss": 0.4968,
+      "num_input_tokens_seen": 1380976,
+      "step": 145
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.4585011899471283,
+      "learning_rate": 1.4029167422908107e-06,
+      "loss": 0.4361,
+      "num_input_tokens_seen": 1423712,
+      "step": 150
+    },
+    {
+      "epoch": 2.7555555555555555,
+      "grad_norm": 0.3712801933288574,
+      "learning_rate": 7.350858136652261e-07,
+      "loss": 0.4517,
+      "num_input_tokens_seen": 1470000,
+      "step": 155
+    },
+    {
+      "epoch": 2.8444444444444446,
+      "grad_norm": 0.4640035629272461,
+      "learning_rate": 2.7922934437178695e-07,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 1515168,
+      "step": 160
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.24545951187610626,
+      "learning_rate": 3.9329624554584884e-08,
+      "loss": 0.4861,
+      "num_input_tokens_seen": 1566432,
+      "step": 165
+    },
+    {
+      "epoch": 2.986666666666667,
+      "num_input_tokens_seen": 1596080,
+      "step": 168,
+      "total_flos": 1.467473931042816e+16,
+      "train_loss": 0.5455739086582547,
+      "train_runtime": 1063.0707,
+      "train_samples_per_second": 2.54,
+      "train_steps_per_second": 0.158
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 168,
+  "num_input_tokens_seen": 1596080,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.467473931042816e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3714cd435343e43fff99200ac2402d2ab09a63941e76fb7bd73195c6e4f4472
+size 5368

training_args.yaml

@@ -0,0 +1,36 @@
+cutoff_len: 1024
+dataset: glaive_toolcall_en
+dataset_dir: data
+ddp_timeout: 180000000
+do_train: true
+eval_steps: 100
+eval_strategy: steps
+finetuning_type: lora
+flash_attn: auto
+fp16: true
+gradient_accumulation_steps: 8
+include_num_input_tokens_seen: true
+learning_rate: 5.0e-05
+logging_steps: 5
+lora_alpha: 16
+lora_dropout: 0
+lora_rank: 8
+lora_target: all
+lr_scheduler_type: cosine
+max_grad_norm: 1.0
+max_samples: 100000
+model_name_or_path: Qwen/Qwen-1_8B
+num_train_epochs: 3.0
+optim: adamw_torch
+output_dir: saves\Qwen-1.8B\lora\train_2024-09-02-15-46-54
+packing: false
+per_device_eval_batch_size: 2
+per_device_train_batch_size: 2
+plot_loss: true
+preprocessing_num_workers: 16
+report_to: none
+save_steps: 100
+stage: sft
+template: default
+val_size: 0.1
+warmup_steps: 0