Ucas-HaoranWei/Vary-toy

Issues

• Vary-tiny 是否支持中文？

  #41 opened a month ago by Davidwhw
  0

• 请问下，相对于诸如Florence2这类的0.3/0.7B的模型，Vary-tony的性能如何呢？

  #40 opened 2 months ago by gjd2017
  0

• How do I download flash-attn? I have followed the downloading steps completely.

  #39 opened 3 months ago by ranck626
  0

• where the requirements

  #38 opened 3 months ago by Moneorker
  0

• 一张卡train不起来

  #34 opened 7 months ago by fanshuaiyao
  1

• ValueError: Trying to set a tensor of shape torch.Size([257, 1024]) in "weight" (which has shape torch.Size([577, 1024])), this look incorrect.

  #37 opened 6 months ago by willpat1213
  0

• 难以控制生成语言种类

  #36 opened 6 months ago by TekhneC
  1

• 训练loss降为0

  #32 opened 8 months ago by afreestudy
  1

• 路径很奇怪

  #35 opened 6 months ago by qin-yuhao
  3

• 训练数据中的<lb>

  #33 opened 7 months ago by fanshuaiyao
  2

• 请问训练大概需要什么性能的GPU

  #9 opened a year ago by xaswq
  9

• 生成的内容出现问题

  #15 opened a year ago by Nikol-coder
  3

• Enhancement Request: Improve Table Recognition and Markdown Conversion

  #8 opened a year ago by hotwa
  6

• 修改加载 CLIP-VIT-L 模型路径的问题

  #7 opened a year ago by hotwa
  9

• 咨询训练SAM

  #30 opened 9 months ago by zhyj3038
  20

• 训练json的格式

  #31 opened 9 months ago by afreestudy
  0

• 请问修改哪里能在训练模型时，接入opt模型

  #29 opened 9 months ago by LimbCC
  0

• 关于demo运行

  #27 opened 10 months ago by Tiehr2000
  1

• How to handle detected coordinates, how is it normalized？

  #28 opened 10 months ago by LinJM
  2

• 我在进行第一阶段的训练（视觉词汇表）后，测试的时候opt输出错误的坐标位置，无法检测对象

  #26 opened 10 months ago by black1948
  1

• 训练参数 --model_name_or_path

  #17 opened 10 months ago by sixgod-666
  1

• 有什么办法把llm部分切换到hf上的qwen2吗？

  #25 opened 10 months ago by shifan3
  1

• RuntimeError: Input type (c10::Half) and bias type (float) should be the same

  #24 opened 10 months ago by Gary-code
  0

• 路径如何修改

  #14 opened a year ago by lht1605766283
  3

• 请问new vision vocabulary weights是否指的是sam部分的权重？

  #22 opened 10 months ago by whalefa1I
  3

• Error: Downloading models from huggingface

  #20 opened a year ago by chenweilong915
  1

• train errorKeyError: 'data_name1'

  #21 opened a year ago by bsbrother
  1

• 麻烦问一下，qwen 1.8B用的是chat版本的还是非chat版本的？

  #19 opened a year ago by duchenzhuang
  2

• About deployment?

  #18 opened a year ago by CVHub520
  1

• 训练的问题

  #13 opened a year ago by duchenzhuang
  1

• exits with return code = -9 after I delete 'device_map="cuda"', OOM will occur if I keep 'device_map="cuda"'

  #16 opened a year ago by zodiac50
  0

• Trying to set a tensor of shape torch.Size([1024]) in "class_embedding"

  #4 opened a year ago by zhangxyzte
  12

• RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:5 and cuda:6!

  #12 opened a year ago by readyFly
  1

• ValueError: Trying to set a tensor of shape torch.Size([1024, 1024]) in "weight" (which has shape torch.Size([2048, 1024])), this look incorrect.

  #11 opened a year ago by zodiac50
  1

• 有没有开源文档渲染数据代码的计划？

  #6 opened a year ago by yazheng0307
  1

• CUDA out of memory

  #5 opened a year ago by sixgod-666
  8

• 可以更一下 requirements.txt 么

  #1 opened a year ago by tpoisonooo
  7

• 支持 4/8 bit 量化运行

  #3 opened a year ago by yazheng0307
  1

• How to fine-tune Vary-tiny with LoRA or SFT?

  #2 opened a year ago by parap1uie-s
  3