modelscope/ms-swift

Issues

• “训练推理界面”后点击“通过 API 使用”时报错。

  #1932 opened 9 days ago by KnightLancelot
  2

• NPU qwen2模型推理报错

  #1951 opened 8 days ago by JiayuQiao
  1

• swift 中如何设置模型内部参数，如internvl中的max_dynamic_patch

  #1953 opened 8 days ago by zws-2019
  2

• internvl2-llama3-76b 微调报错

  #1892 opened 11 days ago by zhangfan-algo
  6

• dpo internvl2存在mismatch

  #1930 opened 9 days ago by Ranking666
  5

• Training stops for `KTO` after model loads into memory.

  #1938 opened 8 days ago by Aunali321
  4

• qwen2_audio_7b_instruct利用VLLM推理错误

  #1937 opened 8 days ago by huangzj421
  2

• 关于 Qwen2_VL-2B 微调时显存不足的问题

  #1935 opened 9 days ago by Betty-J
  4

• 微调速度非常慢

  #1940 opened 8 days ago by gengpeip
  9

• Qwen2-VL-7B-instruct 微调报错：RuntimeError: CUDA error: too many resources requested for launch

  #1927 opened 8 days ago by xiajinxiong
  1

• TypeError: Qwen2ForCausalLM.forward() got an unexpected keyword argument '_data'

  #1929 opened 8 days ago by xiamaozi11
  5

• Qwen2-VL-7B-Instruct Video inference

  #1920 opened 8 days ago by wangli68
  1

• swift infer的时候，传递do_sample参数不起作用

  #1943 opened 8 days ago by baibaiw5
  3

• 使用自定数据集DPO mllm时报错KeyError: 'prompt'

  #1922 opened 8 days ago by SparrowZheyuan18
  2

• 请问是否支持在NPU上训练多模态大模型

  #1941 opened 8 days ago by ChingKwanCheung
  2

• 微调glm4v, 给glm4v的视觉部分都添加了checkpoint, 但是还是显存溢出（lora_target_modules 设置为'ALL' ）

  #1934 opened 9 days ago by samaritan1998
  4

• Cannot get model_type from the deploy service

  #1904 opened 11 days ago by Harry-zzh
  1

• BUG：init_lora lead to the wrong distribute？

  #1944 opened 8 days ago by bonre
  0

• 两张V100微调qwen2-7b，单卡微调正常，双卡微调出现RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.

  #1882 opened 8 days ago by kksmi
  3

• MooER audio support request

  #1919 opened 10 days ago by seetimee
  0

• qwen2-vl-2b-instruct微调报错ValueError: push_best is not a valid HubStrategy, please select one of ['end', 'every_save', 'checkpoint', 'all_checkpoints']

  #1917 opened 8 days ago by learn01one
  4

• GLM4V-9B微调后，直接调用微调的模型出错：ValueError: The following `model_kwargs` are not used by the model: ['images'] (note: typos in the generate arguments will also show up in this list)

  #1902 opened 8 days ago by tw-repository
  1

• Merge LoRA & 量化部分 支持bnb量化嘛

  #1895 opened 8 days ago by kelenlv
  1

• 如何调整训练损失函数类型？自定义损失函数？(Custom loss function)

  #1891 opened 11 days ago by XiaoMaGe-hero
  3

• qwen2-vl微调使用flash_attn报错

  #1887 opened 11 days ago by zhangfan-algo
  6

• dpo微调internvl2

  #1886 opened 8 days ago by Ranking666
  1

• 请教一下，图片、视频等多模态数据超长应该怎么截断？

  #1876 opened 8 days ago by HuiResearch
  4

• CUDA error: too many resources requested for launch (V100, qwen2-vl)

  #1867 opened 14 days ago by Jiax323
  15

• 部分数据集处理时出现了超过max length的warining，但实际数据貌似并没有超过，前几个版本的分支没有这个问题

  #1865 opened 8 days ago by changqingla
  1

• minicpm-V-2最佳实践,执行推理时,模型不输出任何结果

  #1863 opened 14 days ago by hxzl-98
  4

• streaming模式读取数据，显存利用率很低

  #1939 opened 8 days ago by guozhiyao
  0

• lr_scheduler_type

  #1894 opened 9 days ago by mc-lan
  1

• qwen2-vl-chat-instruct示例数据格式

  #1896 opened 11 days ago by Guangming92
  0

• dpo微调与zero3不兼容

  #1899 opened 11 days ago by zhangfan-algo
  1

• llava-llama-3-8b-v1_1 AttributeError: 'NoneType' object has no attribute 'get_output_embeddings'

  #1911 opened 10 days ago by thisiskofi
  2

• 【新增功能需求】 Internvl2模型+VLLM 后端实现 异步客户端请求的Video 推理功能

  #1921 opened 10 days ago by PancakeAwesome
  1

• 视觉模块支持gradient_checkpointing

  #1928 opened 9 days ago by samaritan1998
  0

• qwen2-vl-2b-instruct使用自定义数据集微调出现DatasetGenerationError

  #1918 opened 9 days ago by lgy0404
  0

• minicpm-v-v2.6评测出现结果为0情况

  #1924 opened 10 days ago by zhudongmei123
  0

• Support for Fine-Tuning Best Practices with LLaVA-OV

  #1923 opened 10 days ago by YoungjaeDev
  0

• internvl2-26b多卡训练报错“Expected all tensors to be on the same device...”

  #1910 opened 10 days ago by tzw451721677
  3

• qwen2-vl-2b-instruct微调报错：importlib.metadata.PackageNotFoundError: No package metadata was found for The 'qwen_vl_utils' distribution was not found and is required by this application.

  #1905 opened 10 days ago by lgy0404
  2

• AssertionError: DeepSpeed does not recognize LR scheduler WarmupCosineLR

  #1906 opened 11 days ago by Jintao-Huang
  0

• Failed to import swift.llm.sft because of the following error

  #1898 opened 11 days ago by orzgugu
  15

• ImportError: cannot import name 'LlavaOnevisionForConditionalGeneration' from 'transformers'

  #1878 opened 13 days ago by Lopa07
  2

• internvl-40b模型微调后推理时报错

  #1881 opened 12 days ago by ymlab
  2

• How to freeze the ViT part during full parameter fine-tuning of Qwen2-vl?

  #1879 opened 13 days ago by Jintao-Huang
  1

• Memory(not GPU RAM) exceeds when using 'swift deploy'

  #1866 opened 14 days ago by VenusHui
  0

• qwen2-vl fine-tuning error: module 'torch.nn' has no attribute 'RMSNorm'

  #1870 opened 14 days ago by Jintao-Huang
  0

• deepspeed-zero3: Input should be a valid integer, got a number with a fractional part [type=int_from_float, input_value=11560550.4, input_type=float]

  #1862 opened 14 days ago by Jintao-Huang
  0