swift

swift copied to clipboard

# PR type - [ ] Bug Fix - [x] New Feature - [ ] Document Updates - [ ] More Models or Datasets Support

hjh0119

# PR type - [x] Bug Fix - [ ] New Feature - [x] Document Updates - [ ] More Models or Datasets Support # PR information 修复了两个链接

KPCOFGS

Qwen-VL-Chat-Int4 是否可以full finetune？

1

Qwen-VL-Chat-Int4只有lora和qlora的fintune 脚本，请问可以full finetune吗？

Luccadoremi

# PR type - [ ] Bug Fix - [ ] New Feature - [ ] Document Updates - [x] More Models or Datasets Support # PR information Write the...

tastelikefeet

# PR type - [ ] Bug Fix - [ ] New Feature - [ ] Document Updates - [x] More Models or Datasets Support # PR information https://www.modelscope.cn/datasets/AI-ModelScope/ShareGPT4V/summary https://www.modelscope.cn/datasets/AI-ModelScope/LLaVA-Instruct-150K/summary...

hjh0119

**Describe the bug** What the bug is, and how to reproduce, better with screenshots(描述bug以及复现过程，最好有截图) File "/opt/conda/lib/python3.10/site-packages/transformers/generation/utils.py", line 2693, in sample model_inputs = self.prepare_inputs_for_generation(input_ids, **model_kwargs) File "/root/.cache/huggingface/modules/transformers_modules/cogvlm2-llama3-chinese-chat-19B/modeling_cogvlm.py", line 677, in prepare_inputs_for_generation...

LRHstudy

基座模型qwen1half-14b-chat，使用lora微调合并后量化qwen1half-14b-chat-gptq-int8。 目前在推理时报错： RuntimeError: probability tensor contains either `inf`, `nan` or element < 0 在使用你们开源的模型Qwen1___5-14B-Chat-GPTQ-Int4时没有报错，但是同样的Qwen1___5-14B-Chat-GPTQ-Int8也报上述相同的错误。 开源的模型transformer版本为4.37.0，我环境的版本4.39.3，torch 2.2.2。 具体原因是什么？该怎么解决这个问题呢？

AnsongLi

自我认知微调失败

2

微调前 `import os os.environ['CUDA_VISIBLE_DEVICES'] = '0' from swift.llm import ModelType, InferArguments, infer_main infer_args = InferArguments(model_type=ModelType.qwen1half_0_5b_chat) infer_main(infer_args)` 微调 `import os os.environ['CUDA_VISIBLE_DEVICES'] = '0' from swift.llm import DatasetName, ModelType, SftArguments, sft_main sft_args...

Talbot5

# PR type - [ ] Bug Fix - [ ] New Feature - [ ] Document Updates - [ ] More Models or Datasets Support

Jintao-Huang

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

1

Thanks for your work! Since VLLM has support for LLaVa now, is it possible to support LLaVa inference via API endpoint? https://github.com/vllm-project/vllm/blob/main/examples/llava_example.py

babla9