cs-552-2026-theattentionseekers/general_knowledge_model
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: cs-552-2026-theattentionseekers/general_knowledge_model
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-06-13 10:12:38 +08:00
commit bca1f29769
31 changed files with 153764 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
.gitattributes vendored Normal file
View File

@@ -0,0 +1,36 @@
*.7z filter=lfs diff=lfs merge=lfs -text
*.arrow filter=lfs diff=lfs merge=lfs -text
*.bin filter=lfs diff=lfs merge=lfs -text
*.bz2 filter=lfs diff=lfs merge=lfs -text
*.ckpt filter=lfs diff=lfs merge=lfs -text
*.ftz filter=lfs diff=lfs merge=lfs -text
*.gz filter=lfs diff=lfs merge=lfs -text
*.h5 filter=lfs diff=lfs merge=lfs -text
*.joblib filter=lfs diff=lfs merge=lfs -text
*.lfs.* filter=lfs diff=lfs merge=lfs -text
*.mlmodel filter=lfs diff=lfs merge=lfs -text
*.model filter=lfs diff=lfs merge=lfs -text
*.msgpack filter=lfs diff=lfs merge=lfs -text
*.npy filter=lfs diff=lfs merge=lfs -text
*.npz filter=lfs diff=lfs merge=lfs -text
*.onnx filter=lfs diff=lfs merge=lfs -text
*.ot filter=lfs diff=lfs merge=lfs -text
*.parquet filter=lfs diff=lfs merge=lfs -text
*.pb filter=lfs diff=lfs merge=lfs -text
*.pickle filter=lfs diff=lfs merge=lfs -text
*.pkl filter=lfs diff=lfs merge=lfs -text
*.pt filter=lfs diff=lfs merge=lfs -text
*.pth filter=lfs diff=lfs merge=lfs -text
*.rar filter=lfs diff=lfs merge=lfs -text
*.safetensors filter=lfs diff=lfs merge=lfs -text
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
*.tar.* filter=lfs diff=lfs merge=lfs -text
*.tar filter=lfs diff=lfs merge=lfs -text
*.tflite filter=lfs diff=lfs merge=lfs -text
*.tgz filter=lfs diff=lfs merge=lfs -text
*.wasm filter=lfs diff=lfs merge=lfs -text
*.xz filter=lfs diff=lfs merge=lfs -text
*.zip filter=lfs diff=lfs merge=lfs -text
*.zst filter=lfs diff=lfs merge=lfs -text
*tfevents* filter=lfs diff=lfs merge=lfs -text
tokenizer.json filter=lfs diff=lfs merge=lfs -text

6
.ipynb_checkpoints/chat_template-checkpoint.jinja Normal file
View File

@@ -0,0 +1,6 @@
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}<|im_start|>system
You are solving a general knowledge multiple-choice benchmark. Choose the single best option letter from the listed options. Keep your reasoning strictly concise (max 4000 tokens). Your entire response must begin with a thinking process wrapped in <think>...</think> tags, followed by exactly one LaTeX box like \boxed{A} containing the correct option letter.<|im_end|>
{% endif %}<|im_start|>{{ message['role'] }}
{{ message['content'] }}<|im_end|>
{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
{% endif %}

30
.ipynb_checkpoints/tokenizer_config-checkpoint.json Normal file
View File

@@ -0,0 +1,30 @@
{
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": null,
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"errors": "replace",
"extra_special_tokens": [
"<|im_start|>",
"<|im_end|>",
"<|object_ref_start|>",
"<|object_ref_end|>",
"<|box_start|>",
"<|box_end|>",
"<|quad_start|>",
"<|quad_end|>",
"<|vision_start|>",
"<|vision_end|>",
"<|vision_pad|>",
"<|image_pad|>",
"<|video_pad|>"
],
"is_local": false,
"local_files_only": false,
"model_max_length": 131072,
"pad_token": "<|endoftext|>",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
"unk_token": null
}

66
README.md Normal file
View File

@@ -0,0 +1,66 @@
---
library_name: transformers
model_name: generalknowledge_grpo
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for generalknowledge_grpo
This model is a fine-tuned version of [None](https://huggingface.co/None).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="None", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 1.3.0
- Transformers: 5.7.0
- Pytorch: 2.10.0+cu128
- Datasets: 4.8.5
- Tokenizers: 0.22.2
## Citations
Cite GRPO as:
```bibtex
@article{shao2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@software{vonwerra2020trl,
title = {{TRL: Transformers Reinforcement Learning}},
author = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
license = {Apache-2.0},
url = {https://github.com/huggingface/trl},
year = {2020}
}
```

8
chat_template.jinja Normal file
View File

@@ -0,0 +1,8 @@
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}<|im_start|>system
You are answering a multiple-choice general-knowledge question. If the question needs reasoning, think briefly and concisely inside <think>...</think> tags; for simple factual questions you may answer directly. Always finish with exactly one LaTeX box containing the letter of the single best option, e.g. \boxed{A}.<|im_end|>
{% endif %}{% if message['role'] == 'assistant' %}<|im_start|>assistant
{% generation %}{{ message['content'] }}<|im_end|>{% endgeneration %}
{% else %}<|im_start|>{{ message['role'] }}
{{ message['content'] }}<|im_end|>
{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
{% endif %}

6
checkpoint-150/chat_template.jinja Normal file
View File

@@ -0,0 +1,6 @@
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}<|im_start|>system
You are solving a general knowledge multiple-choice benchmark. Choose the single best option from the listed options. If the question requires logic or calculation, you may think step-by-step before answering, with the reasoning wrapped in <think>...</think> tags, but keep your reasoning strictly concise (max 4000 tokens). Your final answer must ALWAYS be wrapped in exactly one LaTeX box, like \boxed{A}.<|im_end|>
{% endif %}<|im_start|>{{ message['role'] }}
{{ message['content'] }}<|im_end|>
{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
{% endif %}

63
checkpoint-150/config.json Normal file
View File

@@ -0,0 +1,63 @@
{
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": null,
"dtype": "bfloat16",
"eos_token_id": 151645,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"intermediate_size": 6144,
"layer_types": [
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention"
],
"max_position_embeddings": 40960,
"max_window_layers": 28,
"model_type": "qwen3",
"num_attention_heads": 16,
"num_hidden_layers": 28,
"num_key_value_heads": 8,
"pad_token_id": 151643,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 1000000,
"rope_type": "default"
},
"sliding_window": null,
"tie_word_embeddings": true,
"transformers_version": "5.7.0",
"use_cache": false,
"use_sliding_window": false,
"vocab_size": 151936
}

12
checkpoint-150/generation_config.json Normal file
View File

@@ -0,0 +1,12 @@
{
"do_sample": true,
"eos_token_id": [
151645,
151643
],
"pad_token_id": 151643,
"temperature": 0.6,
"top_k": 20,
"top_p": 0.95,
"transformers_version": "5.7.0"
}

3
checkpoint-150/rng_state.pth Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:4dcdcfa7549cafec3c0d60363705c65435266932fbd369f76d25dff5a575c5f8
size 14645

3
checkpoint-150/scheduler.pt Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:bd5a53071a20df912367c1e686cfe28e9485a664be8c9fd166e945ca510dfb58
size 1465

3
checkpoint-150/tokenizer.json Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
size 11422650

30
checkpoint-150/tokenizer_config.json Normal file
View File

@@ -0,0 +1,30 @@
{
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": null,
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"errors": "replace",
"extra_special_tokens": [
"<|im_start|>",
"<|im_end|>",
"<|object_ref_start|>",
"<|object_ref_end|>",
"<|box_start|>",
"<|box_end|>",
"<|quad_start|>",
"<|quad_end|>",
"<|vision_start|>",
"<|vision_end|>",
"<|vision_pad|>",
"<|image_pad|>",
"<|video_pad|>"
],
"is_local": true,
"local_files_only": false,
"model_max_length": 131072,
"pad_token": "<|endoftext|>",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
"unk_token": null
}

484
checkpoint-150/trainer_state.json Normal file
View File

@@ -0,0 +1,484 @@
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 150,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0125,
"completions/max_length": 454.5,
"completions/max_terminated_length": 342.3,
"completions/mean_length": 139.7,
"completions/mean_terminated_length": 119.76785888671876,
"completions/min_length": 11.4,
"completions/min_terminated_length": 11.4,
"entropy": 0.45251130862161515,
"epoch": 0.06666666666666667,
"frac_reward_zero_std": 0.55,
"grad_norm": 24.625,
"kl": 0.003444959384390245,
"learning_rate": 9.911436253643443e-07,
"loss": -0.056199127435684205,
"num_tokens": 22424.0,
"reward": 1.4375,
"reward_std": 0.44930775463581085,
"rewards/accuracy_reward_func/mean": 0.45,
"rewards/accuracy_reward_func/std": 0.42546978294849397,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 10,
"step_time": 17.857542922068387
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 405.7,
"completions/max_terminated_length": 405.7,
"completions/mean_length": 104.75,
"completions/mean_terminated_length": 104.75,
"completions/min_length": 11.0,
"completions/min_terminated_length": 11.0,
"entropy": 0.37909337431192397,
"epoch": 0.13333333333333333,
"frac_reward_zero_std": 0.6,
"grad_norm": 33.25,
"kl": 0.0022808335735902572,
"learning_rate": 9.609315757942502e-07,
"loss": -0.12742023468017577,
"num_tokens": 41264.0,
"reward": 1.65,
"reward_std": 0.369350728392601,
"rewards/accuracy_reward_func/mean": 0.675,
"rewards/accuracy_reward_func/std": 0.32949312031269073,
"rewards/format_reward_func/mean": 0.975,
"rewards/format_reward_func/std": 0.046291005611419675,
"step": 20,
"step_time": 15.968829826638103
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0375,
"completions/max_length": 770.7,
"completions/max_terminated_length": 607.1,
"completions/mean_length": 247.05,
"completions/mean_terminated_length": 189.3309555053711,
"completions/min_length": 11.0,
"completions/min_terminated_length": 11.0,
"entropy": 0.4646949103684165,
"epoch": 0.2,
"frac_reward_zero_std": 0.6,
"grad_norm": 5.15625,
"kl": 0.003183109380734095,
"learning_rate": 9.10574604566852e-07,
"loss": 0.23813328742980958,
"num_tokens": 72036.0,
"reward": 1.6125,
"reward_std": 0.5259025603532791,
"rewards/accuracy_reward_func/mean": 0.675,
"rewards/accuracy_reward_func/std": 0.41283679008483887,
"rewards/format_reward_func/mean": 0.9375,
"rewards/format_reward_func/std": 0.15235702097415924,
"step": 30,
"step_time": 28.961046700924634
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.025,
"completions/max_length": 651.1,
"completions/max_terminated_length": 495.6,
"completions/mean_length": 195.975,
"completions/mean_terminated_length": 155.1339324951172,
"completions/min_length": 12.4,
"completions/min_terminated_length": 12.4,
"entropy": 0.42483185429591686,
"epoch": 0.26666666666666666,
"frac_reward_zero_std": 0.5,
"grad_norm": 12.0625,
"kl": 0.002366386147068633,
"learning_rate": 8.422735529643443e-07,
"loss": 0.07951252460479737,
"num_tokens": 98462.0,
"reward": 1.55,
"reward_std": 0.5763359367847443,
"rewards/accuracy_reward_func/mean": 0.625,
"rewards/accuracy_reward_func/std": 0.4401642054319382,
"rewards/format_reward_func/mean": 0.925,
"rewards/format_reward_func/std": 0.15782093703746797,
"step": 40,
"step_time": 24.76931145079434
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 212.6,
"completions/max_terminated_length": 212.6,
"completions/mean_length": 77.7625,
"completions/mean_terminated_length": 77.7625,
"completions/min_length": 11.4,
"completions/min_terminated_length": 11.4,
"entropy": 0.5150532013736665,
"epoch": 0.3333333333333333,
"frac_reward_zero_std": 0.4,
"grad_norm": 24.0,
"kl": 0.012370214833595128,
"learning_rate": 7.590135046865651e-07,
"loss": 0.07622140645980835,
"num_tokens": 115615.0,
"reward": 1.65,
"reward_std": 0.3610740154981613,
"rewards/accuracy_reward_func/mean": 0.6625,
"rewards/accuracy_reward_func/std": 0.32571869492530825,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 50,
"step_time": 9.345398245193064
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 391.9,
"completions/max_terminated_length": 391.9,
"completions/mean_length": 140.5125,
"completions/mean_terminated_length": 140.5125,
"completions/min_length": 10.5,
"completions/min_terminated_length": 10.5,
"entropy": 0.4693372020497918,
"epoch": 0.4,
"frac_reward_zero_std": 0.55,
"grad_norm": 0.4921875,
"kl": 0.003207334756950786,
"learning_rate": 6.644333233692916e-07,
"loss": -0.014827732741832734,
"num_tokens": 137696.0,
"reward": 1.6375,
"reward_std": 0.37656808495521543,
"rewards/accuracy_reward_func/mean": 0.6625,
"rewards/accuracy_reward_func/std": 0.3202626138925552,
"rewards/format_reward_func/mean": 0.975,
"rewards/format_reward_func/std": 0.07071067690849304,
"step": 60,
"step_time": 15.56080833701417
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.025,
"completions/max_length": 588.2,
"completions/max_terminated_length": 532.9,
"completions/mean_length": 150.75,
"completions/mean_terminated_length": 106.68928833007813,
"completions/min_length": 11.1,
"completions/min_terminated_length": 11.1,
"entropy": 0.4096149187302217,
"epoch": 0.4666666666666667,
"frac_reward_zero_std": 0.5,
"grad_norm": 14.75,
"kl": 0.0029691428710791,
"learning_rate": 5.626666167821521e-07,
"loss": -0.02979586124420166,
"num_tokens": 161620.0,
"reward": 1.525,
"reward_std": 0.48160980343818666,
"rewards/accuracy_reward_func/mean": 0.5625,
"rewards/accuracy_reward_func/std": 0.4115981698036194,
"rewards/format_reward_func/mean": 0.9625,
"rewards/format_reward_func/std": 0.10606601536273956,
"step": 70,
"step_time": 22.58805529791862
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 296.8,
"completions/max_terminated_length": 296.8,
"completions/mean_length": 94.425,
"completions/mean_terminated_length": 94.425,
"completions/min_length": 11.5,
"completions/min_terminated_length": 11.5,
"entropy": 0.3293402023613453,
"epoch": 0.5333333333333333,
"frac_reward_zero_std": 0.75,
"grad_norm": 1.1953125,
"kl": 0.0025371614843209046,
"learning_rate": 4.5816107833384233e-07,
"loss": -0.060630393028259275,
"num_tokens": 180298.0,
"reward": 1.8,
"reward_std": 0.3336638957262039,
"rewards/accuracy_reward_func/mean": 0.8125,
"rewards/accuracy_reward_func/std": 0.3055525213479996,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 80,
"step_time": 12.384980474878102
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0125,
"completions/max_length": 571.0,
"completions/max_terminated_length": 396.8,
"completions/mean_length": 169.8875,
"completions/mean_terminated_length": 145.42857208251954,
"completions/min_length": 11.6,
"completions/min_terminated_length": 11.6,
"entropy": 0.32843760270625355,
"epoch": 0.6,
"frac_reward_zero_std": 0.75,
"grad_norm": 0.265625,
"kl": 0.006397408900011214,
"learning_rate": 3.554841015277641e-07,
"loss": 0.06048367619514465,
"num_tokens": 204697.0,
"reward": 1.725,
"reward_std": 0.2411646068096161,
"rewards/accuracy_reward_func/mean": 0.7375,
"rewards/accuracy_reward_func/std": 0.20580926835536956,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 90,
"step_time": 21.92367275925353
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0125,
"completions/max_length": 561.9,
"completions/max_terminated_length": 436.8,
"completions/mean_length": 148.8625,
"completions/mean_terminated_length": 124.88392944335938,
"completions/min_length": 11.8,
"completions/min_terminated_length": 11.8,
"entropy": 0.385774496383965,
"epoch": 0.6666666666666666,
"frac_reward_zero_std": 0.7,
"grad_norm": 0.05419921875,
"kl": 0.006441528908105454,
"learning_rate": 2.591231629491423e-07,
"loss": -0.09045490622520447,
"num_tokens": 227338.0,
"reward": 1.6375,
"reward_std": 0.4069852590560913,
"rewards/accuracy_reward_func/mean": 0.65,
"rewards/accuracy_reward_func/std": 0.38802950978279116,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 100,
"step_time": 21.567665427271276
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 276.2,
"completions/max_terminated_length": 276.2,
"completions/mean_length": 65.1375,
"completions/mean_terminated_length": 65.1375,
"completions/min_length": 10.5,
"completions/min_terminated_length": 10.5,
"entropy": 0.3241632700897753,
"epoch": 0.7333333333333333,
"frac_reward_zero_std": 0.5,
"grad_norm": 0.98046875,
"kl": 0.003847465490668345,
"learning_rate": 1.7328969800494726e-07,
"loss": 0.09001727104187011,
"num_tokens": 242669.0,
"reward": 1.7375,
"reward_std": 0.4489682465791702,
"rewards/accuracy_reward_func/mean": 0.75,
"rewards/accuracy_reward_func/std": 0.428018119931221,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 110,
"step_time": 11.50382228847593
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0375,
"completions/max_length": 851.4,
"completions/max_terminated_length": 603.7,
"completions/mean_length": 275.25,
"completions/mean_terminated_length": 206.3190490722656,
"completions/min_length": 11.0,
"completions/min_terminated_length": 11.0,
"entropy": 0.41810003817081454,
"epoch": 0.8,
"frac_reward_zero_std": 0.7,
"grad_norm": 1.28125,
"kl": 0.011894569395826693,
"learning_rate": 1.0173504098790186e-07,
"loss": 0.12610946893692015,
"num_tokens": 275633.0,
"reward": 1.6,
"reward_std": 0.45666705071926117,
"rewards/accuracy_reward_func/mean": 0.6375,
"rewards/accuracy_reward_func/std": 0.4069852650165558,
"rewards/format_reward_func/mean": 0.9625,
"rewards/format_reward_func/std": 0.0816463440656662,
"step": 120,
"step_time": 31.933318979479374
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0125,
"completions/max_length": 568.3,
"completions/max_terminated_length": 374.9,
"completions/mean_length": 130.45,
"completions/mean_terminated_length": 105.21250019073486,
"completions/min_length": 11.2,
"completions/min_terminated_length": 11.2,
"entropy": 0.45407755924388765,
"epoch": 0.8666666666666667,
"frac_reward_zero_std": 0.55,
"grad_norm": 51.75,
"kl": 0.008640645261036185,
"learning_rate": 4.758647376699032e-08,
"loss": 0.14615154266357422,
"num_tokens": 297813.0,
"reward": 1.5625,
"reward_std": 0.3699325889348984,
"rewards/accuracy_reward_func/mean": 0.5875,
"rewards/accuracy_reward_func/std": 0.316480353474617,
"rewards/format_reward_func/mean": 0.975,
"rewards/format_reward_func/std": 0.07071067690849304,
"step": 130,
"step_time": 21.87842119038105
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 117.7,
"completions/max_terminated_length": 117.7,
"completions/mean_length": 32.7375,
"completions/mean_terminated_length": 32.7375,
"completions/min_length": 10.8,
"completions/min_terminated_length": 10.8,
"entropy": 0.40678847576491534,
"epoch": 0.9333333333333333,
"frac_reward_zero_std": 0.7,
"grad_norm": 1.03125,
"kl": 0.005609498374019495,
"learning_rate": 1.3210548563419855e-08,
"loss": 0.04860760569572449,
"num_tokens": 311608.0,
"reward": 1.6,
"reward_std": 0.31909499764442445,
"rewards/accuracy_reward_func/mean": 0.6125,
"rewards/accuracy_reward_func/std": 0.30129902362823485,
"rewards/format_reward_func/mean": 0.9875,
"rewards/format_reward_func/std": 0.03535533845424652,
"step": 140,
"step_time": 6.0708080609329045
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 390.7,
"completions/max_terminated_length": 390.7,
"completions/mean_length": 116.95,
"completions/mean_terminated_length": 116.95,
"completions/min_length": 10.9,
"completions/min_terminated_length": 10.9,
"entropy": 0.41855254471302034,
"epoch": 1.0,
"frac_reward_zero_std": 0.55,
"grad_norm": 0.1533203125,
"kl": 0.0024387751711611116,
"learning_rate": 1.0965826257725019e-10,
"loss": -0.009271913766860962,
"num_tokens": 332084.0,
"reward": 1.4875,
"reward_std": 0.4244658529758453,
"rewards/accuracy_reward_func/mean": 0.5125,
"rewards/accuracy_reward_func/std": 0.36824311017990113,
"rewards/format_reward_func/mean": 0.975,
"rewards/format_reward_func/std": 0.07071067690849304,
"step": 150,
"step_time": 15.536289924476296
}
],
"logging_steps": 10,
"max_steps": 150,
"num_input_tokens_seen": 332084,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}

3
checkpoint-150/training_args.bin Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:de9e0fa52b52fd893dc9365789e6afe10494029178f185a645e456c0d5ecf70e
size 7185

6
checkpoint-450/chat_template.jinja Normal file
View File

@@ -0,0 +1,6 @@
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}<|im_start|>system
You are solving a general knowledge multiple-choice benchmark. Choose the single best option from the listed options. If the question requires logic or calculation, you may think step-by-step before answering, with the reasoning wrapped in <think>...</think> tags, but keep your reasoning strictly concise (max 4000 tokens). Your final answer must ALWAYS be wrapped in exactly one LaTeX box, like \boxed{A}.<|im_end|>
{% endif %}<|im_start|>{{ message['role'] }}
{{ message['content'] }}<|im_end|>
{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
{% endif %}

63
checkpoint-450/config.json Normal file
View File

@@ -0,0 +1,63 @@
{
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": null,
"dtype": "bfloat16",
"eos_token_id": 151645,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"intermediate_size": 6144,
"layer_types": [
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention"
],
"max_position_embeddings": 40960,
"max_window_layers": 28,
"model_type": "qwen3",
"num_attention_heads": 16,
"num_hidden_layers": 28,
"num_key_value_heads": 8,
"pad_token_id": 151643,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 1000000,
"rope_type": "default"
},
"sliding_window": null,
"tie_word_embeddings": true,
"transformers_version": "5.7.0",
"use_cache": false,
"use_sliding_window": false,
"vocab_size": 151936
}

12
checkpoint-450/generation_config.json Normal file
View File

@@ -0,0 +1,12 @@
{
"do_sample": true,
"eos_token_id": [
151645,
151643
],
"pad_token_id": 151643,
"temperature": 0.6,
"top_k": 20,
"top_p": 0.95,
"transformers_version": "5.7.0"
}

3
checkpoint-450/rng_state.pth Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:657e0330141404bd24497f59371ac4c26a5c17060ed936305c7e9b6105467f06
size 14645

3
checkpoint-450/scheduler.pt Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:94945cd9e703c68a467d55ede62dcc6b791170b2da980d03b8779fe8333c731c
size 1465

3
checkpoint-450/tokenizer.json Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
size 11422650

30
checkpoint-450/tokenizer_config.json Normal file
View File

@@ -0,0 +1,30 @@
{
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": null,
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"errors": "replace",
"extra_special_tokens": [
"<|im_start|>",
"<|im_end|>",
"<|object_ref_start|>",
"<|object_ref_end|>",
"<|box_start|>",
"<|box_end|>",
"<|quad_start|>",
"<|quad_end|>",
"<|vision_start|>",
"<|vision_end|>",
"<|vision_pad|>",
"<|image_pad|>",
"<|video_pad|>"
],
"is_local": true,
"local_files_only": false,
"model_max_length": 131072,
"pad_token": "<|endoftext|>",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
"unk_token": null
}

1384
checkpoint-450/trainer_state.json Normal file
View File

File diff suppressed because it is too large Load Diff

3
checkpoint-450/training_args.bin Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:de9e0fa52b52fd893dc9365789e6afe10494029178f185a645e456c0d5ecf70e
size 7185

63
config.json Normal file
View File

@@ -0,0 +1,63 @@
{
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": null,
"dtype": "bfloat16",
"eos_token_id": 151645,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"intermediate_size": 6144,
"layer_types": [
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention"
],
"max_position_embeddings": 40960,
"max_window_layers": 28,
"model_type": "qwen3",
"num_attention_heads": 16,
"num_hidden_layers": 28,
"num_key_value_heads": 8,
"pad_token_id": 151643,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 1000000,
"rope_type": "default"
},
"sliding_window": null,
"tie_word_embeddings": true,
"transformers_version": "5.10.2",
"use_cache": false,
"use_sliding_window": false,
"vocab_size": 151936
}

13
generation_config.json Normal file
View File

@@ -0,0 +1,13 @@
{
"bos_token_id": 151643,
"do_sample": true,
"eos_token_id": [
151645,
151643
],
"pad_token_id": 151643,
"temperature": 0.6,
"top_k": 20,
"top_p": 0.95,
"transformers_version": "5.10.2"
}

151388
merges.txt Normal file
View File

File diff suppressed because it is too large Load Diff

3
model.safetensors Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:aa31821dc9bea0c698daddf243588e2f0f32c645dd108e9325dd314800ea9470
size 3441185608

3
tokenizer.json Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
size 11422650

30
tokenizer_config.json Normal file
View File

@@ -0,0 +1,30 @@
{
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": null,
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"errors": "replace",
"extra_special_tokens": [
"<|im_start|>",
"<|im_end|>",
"<|object_ref_start|>",
"<|object_ref_end|>",
"<|box_start|>",
"<|box_end|>",
"<|quad_start|>",
"<|quad_end|>",
"<|vision_start|>",
"<|vision_end|>",
"<|vision_pad|>",
"<|image_pad|>",
"<|video_pad|>"
],
"is_local": false,
"local_files_only": false,
"model_max_length": 131072,
"pad_token": "<|endoftext|>",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
"unk_token": null
}

3
training_args.bin Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9e46f4598ce4fbab39d6227c63649ff9be540876e089ff6fac7e8597c26f5e52
size 5649

1
vocab.json Normal file
View File

File diff suppressed because one or more lines are too long