add v4

LLM-Detector-V4-11w/README.md

@@ -0,0 +1,56 @@
+---
+license: other
+base_model: ./Qwen-1_8B-Chat
+tags:
+- llama-factory
+- lora
+- generated_from_trainer
+model-index:
+- name: qwen-1.8b-1
+  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. -->
+
+# qwen-1.8b-1
+
+This model is a fine-tuned version of [./Qwen-1_8B-Chat](https://huggingface.co/./Qwen-1_8B-Chat) on the ta, the tb, the tc, the td, the te, the tf, the tg and the th datasets.
+
+## 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: 8
+- eval_batch_size: 8
+- seed: 42
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 32
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- num_epochs: 1.0
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.33.0
+- Pytorch 2.1.1+cu121
+- Datasets 2.14.7
+- Tokenizers 0.13.3

LLM-Detector-V4-11w/adapter_config.json

@@ -0,0 +1,22 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "./Qwen-1_8B-Chat",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16.0,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn"
+  ],
+  "task_type": "CAUSAL_LM"
+}

LLM-Detector-V4-11w/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6965aafad356ecc98e45461ada6572c6cd6eacd624f7f2001ff6a47130d5d0c
+size 6308670

LLM-Detector-V4-11w/all_results.json

@@ -0,0 +1,7 @@
+{
+    "epoch": 1.0,
+    "train_loss": 0.09819089700903973,
+    "train_runtime": 7694.3429,
+    "train_samples_per_second": 15.456,
+    "train_steps_per_second": 0.483
+}

LLM-Detector-V4-11w/checkpoint-1000/README.md

@@ -0,0 +1,207 @@
+---
+library_name: peft
+base_model: ./Qwen-1_8B-Chat
+---
+
+# 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]
+- **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 Data 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 Data 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]
+
+
+## Training procedure
+
+
+### Framework versions
+
+
+- PEFT 0.6.2

LLM-Detector-V4-11w/checkpoint-1000/adapter_config.json

@@ -0,0 +1,22 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "./Qwen-1_8B-Chat",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16.0,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn"
+  ],
+  "task_type": "CAUSAL_LM"
+}

LLM-Detector-V4-11w/checkpoint-1000/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:124dd6f8fccbde92448f1727138226664a814afabc1e3e10090af0abc318aa1e
+size 6308670

LLM-Detector-V4-11w/checkpoint-1000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:620e76eef0903318cf917c4b7bb18eecaec0acfa968aeca9e3345c8bc0ff0428
+size 12623610

LLM-Detector-V4-11w/checkpoint-1000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6b2411bd26912ef7914bec8dace4f47f701c3b0d9fd4166ad1988f8b10967c8
+size 14244

LLM-Detector-V4-11w/checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:913bff447363333d34842812791634ae8b8782dead63524bc6f9a5994b43ae65
+size 1064

LLM-Detector-V4-11w/checkpoint-1000/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "additional_special_tokens": [
+    "<|im_end|>"
+  ],
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|endoftext|>"
+}

LLM-Detector-V4-11w/checkpoint-1000/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)

LLM-Detector-V4-11w/checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 8192,
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

LLM-Detector-V4-11w/checkpoint-1000/trainer_state.json

@@ -0,0 +1,79 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.26907036189963673,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991071065046783e-05,
+      "loss": 3.0435,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643480408906496e-05,
+      "loss": 0.0382,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920021814047156e-05,
+      "loss": 0.0227,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858409013313266e-05,
+      "loss": 0.0258,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7799497480410125e-05,
+      "loss": 0.0256,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685204464371269e-05,
+      "loss": 0.0312,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574849941884044e-05,
+      "loss": 0.0212,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449674459261804e-05,
+      "loss": 0.0271,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310572163498205e-05,
+      "loss": 0.0203,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158536682873821e-05,
+      "loss": 0.0172,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 3716,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 8.230715481076531e+16,
+  "trial_name": null,
+  "trial_params": null
+}

LLM-Detector-V4-11w/checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bc4e49fff68420e89c2879345b8b8f748c6b8c4a56351c173150f7352012454
+size 4664

LLM-Detector-V4-11w/checkpoint-2000/README.md

@@ -0,0 +1,207 @@
+---
+library_name: peft
+base_model: ./Qwen-1_8B-Chat
+---
+
+# 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]
+- **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 Data 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 Data 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]
+
+
+## Training procedure
+
+
+### Framework versions
+
+
+- PEFT 0.6.2

LLM-Detector-V4-11w/checkpoint-2000/adapter_config.json

@@ -0,0 +1,22 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "./Qwen-1_8B-Chat",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16.0,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn"
+  ],
+  "task_type": "CAUSAL_LM"
+}

LLM-Detector-V4-11w/checkpoint-2000/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1986936875e1b229164da9982fb915abcb41eb04c84a17c11873032bdc7916eb
+size 6308670

LLM-Detector-V4-11w/checkpoint-2000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a0cb946c801493b87dd628891a62c65a4096d3599d4de1536f8d115192d4619
+size 12623610

LLM-Detector-V4-11w/checkpoint-2000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:293df5460fc572910480c53776d11674c9b37961d0e3a23477756a518f2153ef
+size 14244

LLM-Detector-V4-11w/checkpoint-2000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74bac387f60615f990f0d8cb5bf4d07c8fad3441e7fa74f3ac632d281ac044e5
+size 1064

LLM-Detector-V4-11w/checkpoint-2000/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "additional_special_tokens": [
+    "<|im_end|>"
+  ],
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|endoftext|>"
+}

LLM-Detector-V4-11w/checkpoint-2000/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)

LLM-Detector-V4-11w/checkpoint-2000/tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 8192,
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

LLM-Detector-V4-11w/checkpoint-2000/trainer_state.json

@@ -0,0 +1,139 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5381407237992735,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991071065046783e-05,
+      "loss": 3.0435,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643480408906496e-05,
+      "loss": 0.0382,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920021814047156e-05,
+      "loss": 0.0227,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858409013313266e-05,
+      "loss": 0.0258,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7799497480410125e-05,
+      "loss": 0.0256,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685204464371269e-05,
+      "loss": 0.0312,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574849941884044e-05,
+      "loss": 0.0212,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449674459261804e-05,
+      "loss": 0.0271,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310572163498205e-05,
+      "loss": 0.0203,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158536682873821e-05,
+      "loss": 0.0172,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.994654029322313e-05,
+      "loss": 0.0241,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8200948408864986e-05,
+      "loss": 0.0167,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.636106019677602e-05,
+      "loss": 0.0134,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4440018250689767e-05,
+      "loss": 0.0145,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2451544857469436e-05,
+      "loss": 0.0127,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.040984397678245e-05,
+      "loss": 0.0164,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8329499780114865e-05,
+      "loss": 0.0187,
+      "step": 1700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6225372473876565e-05,
+      "loss": 0.0147,
+      "step": 1800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.41124921507481e-05,
+      "loss": 0.0156,
+      "step": 1900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2005951427504784e-05,
+      "loss": 0.0184,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 3716,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.6501750008604262e+17,
+  "trial_name": null,
+  "trial_params": null
+}

LLM-Detector-V4-11w/checkpoint-2000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bc4e49fff68420e89c2879345b8b8f748c6b8c4a56351c173150f7352012454
+size 4664

LLM-Detector-V4-11w/checkpoint-3000/README.md

@@ -0,0 +1,207 @@
+---
+library_name: peft
+base_model: ./Qwen-1_8B-Chat
+---
+
+# 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]
+- **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 Data 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 Data 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]
+
+
+## Training procedure
+
+
+### Framework versions
+
+
+- PEFT 0.6.2

LLM-Detector-V4-11w/checkpoint-3000/adapter_config.json

@@ -0,0 +1,22 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "./Qwen-1_8B-Chat",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16.0,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "c_attn"
+  ],
+  "task_type": "CAUSAL_LM"
+}

LLM-Detector-V4-11w/checkpoint-3000/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:028af04c0764159e23c5c6ec745e763a194d72ee12668faea997060fe0adcdb7
+size 6308670

LLM-Detector-V4-11w/checkpoint-3000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d958e6fedf6779f5f6ff24e610500ffe7b00b1fc599511ff25458b9832ff9cf4
+size 12623610

LLM-Detector-V4-11w/checkpoint-3000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:116c8e4fc5473807936d8062da5727b2e081708525f159b7c34cafb7a954573e
+size 14244

LLM-Detector-V4-11w/checkpoint-3000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cd44a3ee0f872669b221a2176a2080abe02eb59a5fdb46ac96d95b4fbe14fcf
+size 1064

LLM-Detector-V4-11w/checkpoint-3000/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "additional_special_tokens": [
+    "<|im_end|>"
+  ],
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|endoftext|>"
+}

LLM-Detector-V4-11w/checkpoint-3000/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)

LLM-Detector-V4-11w/checkpoint-3000/tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 8192,
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

LLM-Detector-V4-11w/checkpoint-3000/trainer_state.json

@@ -0,0 +1,199 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8072110856989103,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991071065046783e-05,
+      "loss": 3.0435,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643480408906496e-05,
+      "loss": 0.0382,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920021814047156e-05,
+      "loss": 0.0227,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858409013313266e-05,
+      "loss": 0.0258,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7799497480410125e-05,
+      "loss": 0.0256,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685204464371269e-05,
+      "loss": 0.0312,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574849941884044e-05,
+      "loss": 0.0212,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449674459261804e-05,
+      "loss": 0.0271,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310572163498205e-05,
+      "loss": 0.0203,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158536682873821e-05,
+      "loss": 0.0172,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.994654029322313e-05,
+      "loss": 0.0241,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8200948408864986e-05,
+      "loss": 0.0167,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.636106019677602e-05,
+      "loss": 0.0134,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4440018250689767e-05,
+      "loss": 0.0145,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2451544857469436e-05,
+      "loss": 0.0127,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.040984397678245e-05,
+      "loss": 0.0164,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8329499780114865e-05,
+      "loss": 0.0187,
+      "step": 1700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6225372473876565e-05,
+      "loss": 0.0147,
+      "step": 1800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.41124921507481e-05,
+      "loss": 0.0156,
+      "step": 1900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2005951427504784e-05,
+      "loss": 0.0184,
+      "step": 2000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9920797636221933e-05,
+      "loss": 0.0119,
+      "step": 2100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7871925338955412e-05,
+      "loss": 0.0144,
+      "step": 2200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5873969933681e-05,
+      "loss": 0.0124,
+      "step": 2300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3941203111481194e-05,
+      "loss": 0.0129,
+      "step": 2400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2087430911744144e-05,
+      "loss": 0.0067,
+      "step": 2500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0325895103581462e-05,
+      "loss": 0.0161,
+      "step": 2600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.669178597912217e-06,
+      "loss": 0.0124,
+      "step": 2700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.129115565868463e-06,
+      "loss": 0.0161,
+      "step": 2800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.716706905559768e-06,
+      "loss": 0.0145,
+      "step": 2900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4420416610303325e-06,
+      "loss": 0.0088,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 3716,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.4736072211654246e+17,
+  "trial_name": null,
+  "trial_params": null
+}

LLM-Detector-V4-11w/checkpoint-3000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bc4e49fff68420e89c2879345b8b8f748c6b8c4a56351c173150f7352012454
+size 4664

LLM-Detector-V4-11w/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "additional_special_tokens": [
+    "<|im_end|>"
+  ],
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|endoftext|>"
+}

LLM-Detector-V4-11w/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)

LLM-Detector-V4-11w/tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 8192,
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "QWenTokenizer"
+}

LLM-Detector-V4-11w/train_results.json

@@ -0,0 +1,7 @@
+{
+    "epoch": 1.0,
+    "train_loss": 0.09819089700903973,
+    "train_runtime": 7694.3429,
+    "train_samples_per_second": 15.456,
+    "train_steps_per_second": 0.483
+}

LLM-Detector-V4-11w/trainer_log.jsonl

@@ -0,0 +1,38 @@
+{"current_steps": 100, "total_steps": 3716, "loss": 3.0435, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.991071065046783e-05, "epoch": 0.03, "percentage": 2.69, "elapsed_time": "0:03:29", "remaining_time": "2:06:21"}
+{"current_steps": 200, "total_steps": 3716, "loss": 0.0382, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9643480408906496e-05, "epoch": 0.05, "percentage": 5.38, "elapsed_time": "0:07:00", "remaining_time": "2:03:04"}
+{"current_steps": 300, "total_steps": 3716, "loss": 0.0227, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.920021814047156e-05, "epoch": 0.08, "percentage": 8.07, "elapsed_time": "0:10:21", "remaining_time": "1:57:59"}
+{"current_steps": 400, "total_steps": 3716, "loss": 0.0258, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.858409013313266e-05, "epoch": 0.11, "percentage": 10.76, "elapsed_time": "0:13:45", "remaining_time": "1:54:02"}
+{"current_steps": 500, "total_steps": 3716, "loss": 0.0256, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.7799497480410125e-05, "epoch": 0.13, "percentage": 13.46, "elapsed_time": "0:17:14", "remaining_time": "1:50:50"}
+{"current_steps": 600, "total_steps": 3716, "loss": 0.0312, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.685204464371269e-05, "epoch": 0.16, "percentage": 16.15, "elapsed_time": "0:20:42", "remaining_time": "1:47:31"}
+{"current_steps": 700, "total_steps": 3716, "loss": 0.0212, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.574849941884044e-05, "epoch": 0.19, "percentage": 18.84, "elapsed_time": "0:24:08", "remaining_time": "1:44:01"}
+{"current_steps": 800, "total_steps": 3716, "loss": 0.0271, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.449674459261804e-05, "epoch": 0.22, "percentage": 21.53, "elapsed_time": "0:27:32", "remaining_time": "1:40:23"}
+{"current_steps": 900, "total_steps": 3716, "loss": 0.0203, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.310572163498205e-05, "epoch": 0.24, "percentage": 24.22, "elapsed_time": "0:31:00", "remaining_time": "1:37:02"}
+{"current_steps": 1000, "total_steps": 3716, "loss": 0.0172, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.158536682873821e-05, "epoch": 0.27, "percentage": 26.91, "elapsed_time": "0:34:29", "remaining_time": "1:33:41"}
+{"current_steps": 1100, "total_steps": 3716, "loss": 0.0241, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.994654029322313e-05, "epoch": 0.3, "percentage": 29.6, "elapsed_time": "0:37:53", "remaining_time": "1:30:07"}
+{"current_steps": 1200, "total_steps": 3716, "loss": 0.0167, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.8200948408864986e-05, "epoch": 0.32, "percentage": 32.29, "elapsed_time": "0:41:28", "remaining_time": "1:26:58"}
+{"current_steps": 1300, "total_steps": 3716, "loss": 0.0134, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.636106019677602e-05, "epoch": 0.35, "percentage": 34.98, "elapsed_time": "0:44:54", "remaining_time": "1:23:27"}
+{"current_steps": 1400, "total_steps": 3716, "loss": 0.0145, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.4440018250689767e-05, "epoch": 0.38, "percentage": 37.67, "elapsed_time": "0:48:21", "remaining_time": "1:20:00"}
+{"current_steps": 1500, "total_steps": 3716, "loss": 0.0127, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.2451544857469436e-05, "epoch": 0.4, "percentage": 40.37, "elapsed_time": "0:51:46", "remaining_time": "1:16:28"}
+{"current_steps": 1600, "total_steps": 3716, "loss": 0.0164, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.040984397678245e-05, "epoch": 0.43, "percentage": 43.06, "elapsed_time": "0:55:13", "remaining_time": "1:13:02"}
+{"current_steps": 1700, "total_steps": 3716, "loss": 0.0187, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.8329499780114865e-05, "epoch": 0.46, "percentage": 45.75, "elapsed_time": "0:58:37", "remaining_time": "1:09:31"}
+{"current_steps": 1800, "total_steps": 3716, "loss": 0.0147, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.6225372473876565e-05, "epoch": 0.48, "percentage": 48.44, "elapsed_time": "1:02:04", "remaining_time": "1:06:04"}
+{"current_steps": 1900, "total_steps": 3716, "loss": 0.0156, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.41124921507481e-05, "epoch": 0.51, "percentage": 51.13, "elapsed_time": "1:05:29", "remaining_time": "1:02:35"}
+{"current_steps": 2000, "total_steps": 3716, "loss": 0.0184, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2005951427504784e-05, "epoch": 0.54, "percentage": 53.82, "elapsed_time": "1:09:00", "remaining_time": "0:59:12"}
+{"current_steps": 2100, "total_steps": 3716, "loss": 0.0119, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.9920797636221933e-05, "epoch": 0.57, "percentage": 56.51, "elapsed_time": "1:12:21", "remaining_time": "0:55:40"}
+{"current_steps": 2200, "total_steps": 3716, "loss": 0.0144, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.7871925338955412e-05, "epoch": 0.59, "percentage": 59.2, "elapsed_time": "1:15:50", "remaining_time": "0:52:15"}
+{"current_steps": 2300, "total_steps": 3716, "loss": 0.0124, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.5873969933681e-05, "epoch": 0.62, "percentage": 61.89, "elapsed_time": "1:19:14", "remaining_time": "0:48:47"}
+{"current_steps": 2400, "total_steps": 3716, "loss": 0.0129, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3941203111481194e-05, "epoch": 0.65, "percentage": 64.59, "elapsed_time": "1:22:41", "remaining_time": "0:45:20"}
+{"current_steps": 2500, "total_steps": 3716, "loss": 0.0067, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2087430911744144e-05, "epoch": 0.67, "percentage": 67.28, "elapsed_time": "1:26:07", "remaining_time": "0:41:53"}
+{"current_steps": 2600, "total_steps": 3716, "loss": 0.0161, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0325895103581462e-05, "epoch": 0.7, "percentage": 69.97, "elapsed_time": "1:29:32", "remaining_time": "0:38:26"}
+{"current_steps": 2700, "total_steps": 3716, "loss": 0.0124, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.669178597912217e-06, "epoch": 0.73, "percentage": 72.66, "elapsed_time": "1:33:00", "remaining_time": "0:34:59"}
+{"current_steps": 2800, "total_steps": 3716, "loss": 0.0161, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.129115565868463e-06, "epoch": 0.75, "percentage": 75.35, "elapsed_time": "1:36:30", "remaining_time": "0:31:34"}
+{"current_steps": 2900, "total_steps": 3716, "loss": 0.0145, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.716706905559768e-06, "epoch": 0.78, "percentage": 78.04, "elapsed_time": "1:39:54", "remaining_time": "0:28:06"}
+{"current_steps": 3000, "total_steps": 3716, "loss": 0.0088, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.4420416610303325e-06, "epoch": 0.81, "percentage": 80.73, "elapsed_time": "1:43:21", "remaining_time": "0:24:40"}
+{"current_steps": 3100, "total_steps": 3716, "loss": 0.0124, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.314224954724593e-06, "epoch": 0.83, "percentage": 83.42, "elapsed_time": "1:46:56", "remaining_time": "0:21:14"}
+{"current_steps": 3200, "total_steps": 3716, "loss": 0.0154, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.341312948250349e-06, "epoch": 0.86, "percentage": 86.11, "elapsed_time": "1:50:23", "remaining_time": "0:17:48"}
+{"current_steps": 3300, "total_steps": 3716, "loss": 0.0161, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.5302552960244022e-06, "epoch": 0.89, "percentage": 88.81, "elapsed_time": "1:53:55", "remaining_time": "0:14:21"}
+{"current_steps": 3400, "total_steps": 3716, "loss": 0.0118, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.868455028627959e-07, "epoch": 0.91, "percentage": 91.5, "elapsed_time": "1:57:26", "remaining_time": "0:10:54"}
+{"current_steps": 3500, "total_steps": 3716, "loss": 0.0067, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.156795401186381e-07, "epoch": 0.94, "percentage": 94.19, "elapsed_time": "2:00:47", "remaining_time": "0:07:27"}
+{"current_steps": 3600, "total_steps": 3716, "loss": 0.0096, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2012301597874863e-07, "epoch": 0.97, "percentage": 96.88, "elapsed_time": "2:04:14", "remaining_time": "0:04:00"}
+{"current_steps": 3700, "total_steps": 3716, "loss": 0.0113, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2871344263597496e-09, "epoch": 1.0, "percentage": 99.57, "elapsed_time": "2:07:40", "remaining_time": "0:00:33"}
+{"current_steps": 3716, "total_steps": 3716, "loss": null, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "2:08:14", "remaining_time": "0:00:00"}

LLM-Detector-V4-11w/trainer_state.json

@@ -0,0 +1,250 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998654648190501,
+  "eval_steps": 500,
+  "global_step": 3716,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991071065046783e-05,
+      "loss": 3.0435,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643480408906496e-05,
+      "loss": 0.0382,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920021814047156e-05,
+      "loss": 0.0227,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858409013313266e-05,
+      "loss": 0.0258,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7799497480410125e-05,
+      "loss": 0.0256,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685204464371269e-05,
+      "loss": 0.0312,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574849941884044e-05,
+      "loss": 0.0212,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449674459261804e-05,
+      "loss": 0.0271,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310572163498205e-05,
+      "loss": 0.0203,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158536682873821e-05,
+      "loss": 0.0172,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.994654029322313e-05,
+      "loss": 0.0241,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8200948408864986e-05,
+      "loss": 0.0167,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.636106019677602e-05,
+      "loss": 0.0134,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4440018250689767e-05,
+      "loss": 0.0145,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2451544857469436e-05,
+      "loss": 0.0127,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.040984397678245e-05,
+      "loss": 0.0164,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8329499780114865e-05,
+      "loss": 0.0187,
+      "step": 1700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6225372473876565e-05,
+      "loss": 0.0147,
+      "step": 1800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.41124921507481e-05,
+      "loss": 0.0156,
+      "step": 1900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2005951427504784e-05,
+      "loss": 0.0184,
+      "step": 2000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9920797636221933e-05,
+      "loss": 0.0119,
+      "step": 2100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7871925338955412e-05,
+      "loss": 0.0144,
+      "step": 2200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5873969933681e-05,
+      "loss": 0.0124,
+      "step": 2300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3941203111481194e-05,
+      "loss": 0.0129,
+      "step": 2400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2087430911744144e-05,
+      "loss": 0.0067,
+      "step": 2500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0325895103581462e-05,
+      "loss": 0.0161,
+      "step": 2600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.669178597912217e-06,
+      "loss": 0.0124,
+      "step": 2700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.129115565868463e-06,
+      "loss": 0.0161,
+      "step": 2800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.716706905559768e-06,
+      "loss": 0.0145,
+      "step": 2900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4420416610303325e-06,
+      "loss": 0.0088,
+      "step": 3000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.314224954724593e-06,
+      "loss": 0.0124,
+      "step": 3100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.341312948250349e-06,
+      "loss": 0.0154,
+      "step": 3200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5302552960244022e-06,
+      "loss": 0.0161,
+      "step": 3300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.868455028627959e-07,
+      "loss": 0.0118,
+      "step": 3400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.156795401186381e-07,
+      "loss": 0.0067,
+      "step": 3500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2012301597874863e-07,
+      "loss": 0.0096,
+      "step": 3600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2871344263597496e-09,
+      "loss": 0.0113,
+      "step": 3700
+    },
+    {
+      "epoch": 1.0,
+      "step": 3716,
+      "total_flos": 3.0689685240938496e+17,
+      "train_loss": 0.09819089700903973,
+      "train_runtime": 7694.3429,
+      "train_samples_per_second": 15.456,
+      "train_steps_per_second": 0.483
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 3716,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 3.0689685240938496e+17,
+  "trial_name": null,
+  "trial_params": null
+}

LLM-Detector-V4-11w/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bc4e49fff68420e89c2879345b8b8f748c6b8c4a56351c173150f7352012454
+size 4664