modelscope/ms-swift

Issues

• We need to unify the data format for MLLM

  #2174 opened a month ago by YerongLi
  1

• qwen2-vl-72b lora微调 ，不支持纯文本指令和图片指令数据混合训练？

  #2198 opened a month ago by Luccadoremi
  4

• Notebook for Continue Pretraining

  #2190 opened 2 months ago by msamwelmollel
  1

• qwen1.5-A2.7B-moe-chat 训练速度过慢

  #2213 opened 2 months ago by cdxzyc
  5

• qwenvl2-72b-instruct-awq finetune 报错 [rank7]: NotImplementedError: Cannot copy out of meta tensor; no data!

  #2208 opened 2 months ago by ljqnb
  3

• Question about auto_find_batch_size.

  #2209 opened 2 months ago by lluo-Desktop
  1

• 强化学习训练MLLM

  #2212 opened 2 months ago by AnsongLi
  5

• Can't Find How to Apply Inference on Fine-Tuned Qwen2-vl 7B Model

  #2200 opened a month ago by gungorturan
  1

• 能否给一个megatron训练的docker镜像？

  #2214 opened a month ago by jhrsya
  1

• swift.llm.inference_client增加vllm、lmdeploy 部署后端的 request_id参数修改功能，提供用户可以追踪的日志

  #2161 opened 2 months ago by PancakeAwesome
  0

• Audio output is missing from the llama-omni

  #2216 opened 2 months ago by satheeshkola-532
  0

• support for Ovis1.6-Gemma2-9B

  #2183 opened 2 months ago by betterftr
  1

• AttributeError: 'OmniSpeech2SLlamaForCausalLM' object has no attribute '_get_logits_warper'. Did you mean: '_get_logits_processor'?

  #2194 opened 2 months ago by satheeshkola-532
  6

• assert num_new_tags >= 0, f'Number of media: {num_media}, number of media_tags: {num_media_tags}'

  #2164 opened 2 months ago by dgedanke
  1

• qwen2 72B vl部署推理服务报错

  #2204 opened 2 months ago by zhangfan-algo
  6

• How to change the .cache/modelspace location

  #2179 opened 2 months ago by YerongLi
  1

• finetuning internvl2 KeyError: 'architectures'

  #2180 opened 2 months ago by Li-Jicheng
  1

• qwen2-vl-7b-instruct vllm inference assert "factor" in rope_scaling

  #2188 opened 2 months ago by daje0601
  1

• Support for Molmo

  #2199 opened 2 months ago by betterftr
  0

• package installation help: swift with transformers, deepspeed, flash-attention, vllm, etc

  #2202 opened 2 months ago by Li-Jicheng
  1

• Finetuned Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4 cannot be loaded with PEFT

  #2203 opened 2 months ago by ROIM1998
  1

• 是否支持图像的在线数据增强

  #2168 opened 2 months ago by ninghaiywx
  0

• 请问MiniCPM V2_6的视频微调如何设置视频的采样帧数？

  #2196 opened 2 months ago by mycroft1603
  0

• train two models on the two gpus in one mechine at the same time will cause "address already in use".

  #2192 opened 2 months ago by ycwfs
  0

• 请问支持Mac M系列机器吗

  #2163 opened 2 months ago by mobguang
  2

• How to implement weight decay towards the pre-trained model?

  #2189 opened 2 months ago by sedol1339
  0

• llama-3.2-3b instruct doesn't stop writing

  #2184 opened 2 months ago by Aunali321
  5

• media key for custom dataset is not added to empty_row

  #2185 opened 2 months ago by SepehrV
  1

• Does it support grounding, OCR, Video hybrid fine-tuning training?

  #2186 opened 2 months ago by WangRongsheng
  2

• post_encode for internlm-xcomposer2-7b is incorrect

  #2176 opened 2 months ago by YerongLi
  1

• About DataCollatorForCompletionOnlyLM(label)

  #2187 opened 2 months ago by daje0601
  2

• 求助！使用vllm在internvl2评测自制数据集时报错

  #2182 opened 2 months ago by NancyFyong
  0

• 是否可以支持多卡并行的多step inference，以及训练时每一个step online生成训练数据多卡训练模型

  #2181 opened 2 months ago by goodstudent9
  0

• docker로 사용하고 싶은데 어떻게 사용해야하는지 모르겠어요

  #2178 opened 2 months ago by daje0601
  1

• mPlug-Owl3 inference bug report

  #2172 opened 2 months ago by goodstudent9
  0

• qwen2-vl/llama3.2-vl batch_size>1 zero2/zero3 mixed text training error

  #2147 opened 2 months ago by Jintao-Huang
  0

• 新版本的kto无效

  #2167 opened 2 months ago by rtz1998
  1

• Can we support fp 16 and bf16 in training

  #2170 opened 2 months ago by YerongLi
  1

• Image scaling issues with DeepSeek-vl-1.3b

  #2171 opened 2 months ago by xiang-xiang-zhu
  1

• How to SFT with customized datasets

  #2150 opened 2 months ago by YerongLi
  1

• SFT qwen/Qwen2-VL-7B-Instruct did not pass/process the image embedding for feedforwarding

  #2162 opened 2 months ago by YerongLi
  5

• speed reduction when fine-tuning Qwen2.5-7B

  #2166 opened 2 months ago by YingchaoX
  0

• Using customized dataset evaluation ends with lookup error

  #2153 opened 2 months ago by mianbaoji
  2

• GOT-OCR自定义数据

  #2148 opened 2 months ago by cccusername
  8

• 官方DEMO不好使?

  #2159 opened 2 months ago by BenLampson
  4

• mplug-owl3 doing full model sft bug report

  #2158 opened 2 months ago by goodstudent9
  6

• qwen2-audio-7b-instruct推理胡说八道

  #2156 opened 2 months ago by Liufeiran123
  1

• ValueError: Mixed using with peft is not allowed now

  #2152 opened 2 months ago by thesby
  1

• TypeError: ChatGLM4Tokenizer._pad() got an unexpected keyword argument 'padding_side'

  #2151 opened 2 months ago by xdaiycl
  2

• qwen2-vl pretrain supports

  #2145 opened 2 months ago by Jintao-Huang
  0