modelscope/swift

Issues

• 如何批量导出MS数据集为swift能加载的格式

  #1041 opened 9 days ago by WSC741606
  0

• 是否支持idm-vton模型的微调呢？

  #1040 opened 9 days ago by awzhgw
  0

• None of the inputs have requires_grad=True. Gradients will be None

  #1027 opened 9 days ago by Anorid
  2

• 目前支持多模态模型部署了吗

  #987 opened 9 days ago by LRHstudy
  3

• 自定义数据微调MiniCPM-Llama3-V-2_5报错

  #1030 opened 9 days ago by zhudongwork
  4

• qwen1.5-32b-chat 使用vllm推理很慢

  #986 opened 19 days ago by zhangfan-algo
  5

• swift微调多模态大模型后，比如Intern VL 1.5，可以使用lmdeploy部署吗？

  #1033 opened 10 days ago by wangdong1992
  2

• 自定义数据集报错

  #980 opened 20 days ago by Vindicator645
  2

• 单机多卡微调报错： Expected all tensors to be on the same device, but found at least two devices, cuda:2 and cuda:3! is:closed

  #1022 opened 11 days ago by Elissa0723
  0

• 使用appui功能无法指定ip和端口,还是会使用默认配置

  #1034 opened 10 days ago by zhangfan-algo
  0

• NPU微调报错：ValueError: Your assigned backend {original_backend} is not avaliable, please use {backend}

  #1020 opened 11 days ago by FlynnShi
  1

• Support RLAIF-V

  #981 opened 20 days ago by choyakawa
  0

• bitsandbytes was compiled without GPU support

  #1031 opened 11 days ago by stellarxxu
  0

• react模板放在system和user的区别？

  #1025 opened 11 days ago by nauyiahc
  1

• 可以支持一下InternLM2-Math-Plus-Mixtral8x22B的微调吗

  #1019 opened 12 days ago by zhangfan-algo
  1

• Can I deploy SWIFT-trained models without using SWIFT inference?

  #989 opened 18 days ago by babla9
  2

• model_type: 'minicpm-v-v2_5-chat' is not registered.

  #977 opened 12 days ago by zfy1041264242
  6

• infer 怎么指定输出文件夹？

  #957 opened 12 days ago by AlexJJJChen
  2

• infer 无法跑完所有data

  #1006 opened 15 days ago by AlexJJJChen
  12

• 多模态模型（qwen_vl_chat）量化失败

  #1015 opened 13 days ago by Luccadoremi
  0

• VLLM added support for MultiModal LLaVa - can SWIFT support LLaVa via API?

  #990 opened 18 days ago by babla9
  1

• 可以支持一下qwen1.5系列模型的预训练吗

  #1018 opened 12 days ago by zhangfan-algo
  2

• 模型并行自我认知微调报错

  #1014 opened 12 days ago by wssywh
  1

• 微调qwen后会循环输出

  #1009 opened 14 days ago by sherry085
  8

• 采用 --dtype fp16 的方式DPO训练后，无法推理。

  #1011 opened 13 days ago by stevezhang88
  1

• 使用教程训练会爆cuda错误

  #1012 opened 13 days ago by yangtianyu92
  1

• 多模态提前将图片处理，然后再训练LLM

  #1007 opened 14 days ago by choyakawa
  1

• 微调minicpmv2时cpu占用率超高

  #1008 opened 14 days ago by strawhatboy
  0

• awq 微调后如何推理

  #1002 opened 14 days ago by hehuang139
  3

• Qwen-VL-Chat-Int4 是否可以full finetune？

  #1000 opened 14 days ago by Luccadoremi
  1

• cogvlm2添加history报错

  #993 opened 14 days ago by LRHstudy
  1

• 微调量化后qwen1half-14b-chat-gptq-int8推理时向量报错RuntimeError: probability tensor contains either `inf`, `nan` or element < 0

  #994 opened 18 days ago by AnsongLi
  0

• 自我认知微调失败

  #984 opened 19 days ago by Talbot5
  2

• grok-1载入时间过长

  #988 opened 19 days ago by Di-Zayn
  0

• 支持swift接入数据集的下载和离线训练。

  #982 opened 19 days ago by stitchshaw
  1

• 微调internvl-V1-5一直报warning

  #985 opened 19 days ago by sunzx8
  0

• swift export 指定 --tensor_parallel_size --gpu_memory_utilization 感觉不管用

  #969 opened 21 days ago by LIUKAI0815
  6

• 模型微调后不停止回答<|im_end|>

  #960 opened 19 days ago by againcui
  3

• 是否支持deepseek-v2-chat-lite model ?

  #955 opened 19 days ago by awzhgw
  1

• peft加载qwen1half_72b_chat的lora模型报错

  #975 opened 20 days ago by jhjiang10
  2

• TypeError: Subscripted generics cannot be used with class and instance checks

  #961 opened 22 days ago by Hsu5918
  6

• 能否支持多卡DDP部署

  #956 opened 20 days ago by WSC741606
  2

• 并发

  #963 opened 20 days ago by wuguangshuo
  1

• qwen1half-14b-chat-int4使用lora微调后合并模型报错

  #966 opened 20 days ago by AnsongLi
  1

• 怎么做 batch infer 来提高显卡利用率呢?

  #968 opened 21 days ago by thesby
  1

• New version code issue for Internvl "cannot import name 'BOFTConfig' from 'peft'"

  #976 opened 20 days ago by MVP-D77
  1

• 使用zero3进行多机多卡全量微调，保存的模型权重不完整

  #972 opened 20 days ago by ultrazhl98
  0

• TypeError: unsupported operand type(s) for -: 'NoneType' and 'int'

  #965 opened 20 days ago by mayiming0708
  2

• ValueError: model_type: 'yi-1_5-6b' is not registered

  #959 opened 21 days ago by jacnmm4
  4

• ValueError: model_type: 'yi-1_5-6b' is not registered.

  #958 opened 22 days ago by jacnmm4
  0