QwenLM/Qwen2.5

Issues

• [Bug]: 对lora merge后的模型量化，量化后模型输出一直出现human：

  #1029 opened 20 days ago by shenshaowei
  4

• [Bug]: qwen2.5 tools输出格式问题

  #1072 opened a month ago by kwjlhh
  4

• [QwQ-Preview]: LiveCodeBench复现

  #1121 opened 21 days ago by mattgithub1919
  0

• [QwQ-Preview]: 请问Qwen2.5-7B-Instruct默认是FP32还是FP16

  #1120 opened 20 days ago by lckj2009
  1

• 您好，请问QwQ-32B-Preview发布的榜单中LiveCodeBench测试集中2024.08-2024.11数据集是从哪里的？LiveCodeBench最新开源数据集只截至到2024.09

  #1112 opened a month ago by 13416157913
  5

• [Badcase]: Generate json format error

  #1095 opened a month ago by jxrjlxc02
  2

• [Bug]: 在 4 卡 16GB V100 机器上采用 lmdeploy 部署 qwen2.5-32b-instruct-gptq-int4 模型，最高输出速度只有 80token/s ，请问这个速度正常吗？

  #1023 opened 2 months ago by SolomonLeon
  4

• [Badcase]: Qwen2.5-7B-Instruct 中德、中意翻译 不遵循指令、code switch

  #1097 opened a month ago by pio57019
  0

• [Bug]: AttributeError: Model Qwen2ForCausalLM does not support BitsAndBytes quantization yet.

  #1049 opened 22 days ago by yananchen1989
  2

• [Bug]: Qwen 2.5 32B/14B keep repeating input prompt

  #1089 opened 22 days ago by youyouisme
  1

• [BUG]: autoawq 0.2.7量化得到模型，无法使用vllm进行推理

  #1117 opened 22 days ago by dcdmm
  5

• "llama_model_load: error loading model: check_tensor_dims: tensor 'token_embd.weight' not found" after using "Quantizing the GGUF with AWQ Scale"

  #1101 opened 22 days ago by Autism-al
  4

• [Bug]: 使用官方文档提供的方法:LLaMA-Factory微调Qwen2.5-7B-Instruct出现no_sync兼容问题?

  #1118 opened 22 days ago by paulxs
  2

• [Bug]: Poor support for Hebrew

  #1114 opened 25 days ago by shohamyamin
  1

• [Bug]: vLLM got different results with PeftModelForCausalLM

  #1018 opened 25 days ago by chansonzhang
  4

• A simplified version of the inference code ?

  #1057 opened 2 months ago by weizhenhuan
  2

• [Badcase]: openai.BadRequestError: Error code: 400 - {'error': {'message': 'unexpected EOF', 'type': 'invalid_request_error', 'param': None, 'code': None}}

  #1026 opened a month ago by XyLove0223
  6

• [Bug]: Qwen2.5-14B-Instruct-GPTQ-Int4存在严重的复读机和幻觉现象

  #1024 opened a month ago by yang-collect
  8

• [Badcase]:

  #1027 opened a month ago by zhuzcalex
  1

• [Bug]: vllm infer Qwen2.5-32B-Instruct-AWQ with 2 * Nvidia-L20, output repeat !!!!

  #1090 opened a month ago by RoyaltyLJW
  16

• [Bug]: Qwen2.5-32b-int4用vllm跑 好像只会生成感叹号

  #1103 opened a month ago by ciaoyizhen
  4

• [Badcase]: Twenty-four game, how to get 24 from 23, 3, 11, 16 Not possible with the given constraints

  #1109 opened a month ago by novohool
  0

• 您好，为什么QwQ-32B-Preview模型测试GPQA数据集时，模型会回答中文，而且模型回答好像停止不了，一直回答到token限制为止

  #1111 opened a month ago by 13416157913
  0

• [Bug]: 模型回答时极大概率出现�

  #1039 opened 2 months ago by GitHub-lql
  3

• [Bug]: Inference code that works locally (M1 Mac) fails on cloud infrastructure (Linux box)

  #1102 opened a month ago by nathancarter
  4

• [Bug]: 使用 Xinference vLLM 启动 qwen2.5-32b-instruct 推理结果都是感叹号

  #1038 opened 2 months ago by andylzming
  3

• [Bug]: 使用vllm进行推理时，设置parallel_tool_calls似乎不生效。我想实现单工具调用，应该怎么设置呢

  #1092 opened a month ago by RayneSun
  5

• [Badcase]: loss unstable

  #1074 opened a month ago by Solo4working
  4

• [Bug]: 在有gpu的机器上，docker中使用ollama run qwen2.5:1.5时，报错server cpu not listed

  #1081 opened a month ago by 151812117
  2

• 您好，请问一下为什么Qwen2.5官方发布的榜单中72B和32B的LiveCodeBench分数跟LiveCodeBench作者公开72B和32B的分数差异这么大？

  #1079 opened a month ago by 13416157913
  3

• [Badcase]: 使用Reasoning数据微调2.0成功，但是2.5失败

  #1085 opened a month ago by UESTCthb
  1

• [Badcase]: 用mlx框架把Qwen2.5-3B-Instruct转换为mlx版本后，又用mlx_lm.convert把mlx版本的模型进行int4量化，最后进行推理的时候报错：ValueError: [dequantize] The matrix should be given as a uint32

  #1041 opened 2 months ago by ghoshadow
  3

• [Badcase]: 使用llamafactory量化gptq时，使用2个GPU时出错

  #1084 opened a month ago by czhcc
  1

• 您好，请问一下在Qwen2.5的发布榜单中Arena-Hard评测中，裁判模型是使用哪个？基线模型是使用哪个？

  #1086 opened a month ago by 13416157913
  0

• [Badcase]: 抢占式实例部署qwen2.5-72B成功，调用失败

  #1088 opened a month ago by ZX1998-12
  1

• [Badcase]: 相同的微调数据，Qwen1.5 14B准确率比Qwen2.5 14B高20%左右，这是什么原因

  #1016 opened 2 months ago by Jayc-Z
  2

• [Bug]: The final reason why you will get a model that cannot stop generation when you fine-tune the Qwen2.5-7b-base use Lora and a non-<|endoftext|> token as eos_token.

  #1064 opened 2 months ago by hxs91
  9

• [Badcase]: 函数调用结果和目标预期不符

  #1075 opened a month ago by akai-shuuichi
  2

• [Bug]: transformers[4.46.2] error in multi-gpu when training.

  #1070 opened 2 months ago by tanypullin
  1

• [Bug]: 使用Qwen2.5-72B-Instruct-AWQ和Qwen2.5-32B-Instruct-AWQ官方给的演示例子，运行的时候提示：returned non-zero exit status 1

  #1059 opened 2 months ago by SuSuStarSmile
  1

• [Badcase]: Qwen2推理加速没有效果（accelerate、flash_attention_2、deepspeed）

  #1078 opened a month ago by kofuya
  1

• [Bug]: qwen2.5-72b-insruct math 自测分数和榜单分数差异较大

  #1020 opened 2 months ago by tianshiyisi
  7

• [Bug]: Use of the term "open source" to describe qwen when the training data is not open

  #1055 opened 2 months ago by phly95
  3

• [REQUEST]:

  #1040 opened 2 months ago by DAAworld
  1

• [Bug]: vllm启动大模型，超过一定的上下文长度导致大模型回答答非所问

  #1031 opened 2 months ago by Ave-Maria
  3

• [Bug]: Model name error in vllm deployment

  #1052 opened 2 months ago by JulioZhao97
  4

• [Bug]: can not deploy qwen2.5 with vllm

  #1048 opened 2 months ago by joyyyhuang
  1

• [Badcase]: qwen2.5一定概率生成\\n

  #1021 opened 2 months ago by 520jefferson
  1

• [Badcase]: Qwen2.5-7B-Instruct few shot能力很差，具体MedBench的SMDoc数据集

  #1032 opened 2 months ago by tzyodear
  0

• Qwen2.5-1.5b用LLaMA-factory lora微调后，vLLM加载模型报错，求教~

  #1022 opened 2 months ago by Jimmy-L99
  0