moro0v0
moro0v0
> Hi, thank you for your interest in our work! > > Our experiments are conducted using the A100, so the `dtype` for the LLM can be set to `bfloat16`....
I do need help.brother!!!!!!!!
> I set the parameter gradient_accumulation_steps to 1,bachsize to 1 and use LoRA to make the number of trainable parameters reduce to 3,276,800.However,with two v100(32G),I still can't run this experiment...
hello, I have test olmOCR. But I do fell it perform not good and has more mistake than minerU. Do you have better tools that can make a good perform...
> 目前我已经通过修改代码，能将所有的公式保存成图片了（与image和table一样），但对应的md文件中的公式没有像图片和表格一样有图片链接，我应该如何修改代码完成这个功能、 请问如何将公式保存为图片
> [@Jamly7](https://github.com/Jamly7) After saving the path of equation screenshot, you should modify functions called `pipe_mk_markdown` and `pipe_mk_uni_format` in `magic_pdf/pipe/XXPipe.py` 我在cut_image.py里面也增加了def ocr_cut_image_and_table(spans, page, page_id, pdf_bytes_md5, imageWriter): def return_path(type): return join_path(pdf_bytes_md5, type)...
curl -s -X POST "http://localhost:6768/v1/chat/completions" \ -H "Content-Type: application/json" \ -d '{ "model": "test", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "识别视频里面的内容" }, { "type": "video_url",...
我看了一下代码，为什么现在都在传入本地地址后加一个 try: os.makedirs(SAFE_MEDIA_PATH, exist_ok=True) real_path = os.path.realpath(path) safe_path = os.path.realpath(SAFE_MEDIA_PATH) if not real_path.startswith(safe_path): raise HTTPException( status_code=status.HTTP_403_FORBIDDEN, detail="File access is restricted to the safe media directory." ) 安全地址判断，只让把多媒体内容放在/LLaMA-Factory-main/src/llamafactory/api/safe_media下面
> > 请问这个解决了吗，同遇到这个问题 > > src/llamafactory/train/sft/workflow.py 把代码里面的 gen_kwargs["eos_token_id"] = [tokenizer.eos_token_id] + tokenizer.additional_special_tokens_ids 改成gen_kwargs["eos_token_id"] = [151645, 198] 目前做了一个数据集的推理，已解决。 你好我想请问用api的时候处理audio音频的时候怎么个格式？ { "model": MODEL_NAME, "messages": [ { "role": "system", "content": "" }, {...