MILVLG/prophet

Issues

• Replacing GPT-3 with other academic LLMs

  #42 opened 2 years ago by wymanCV
  8

• Hello, may I ask how you avoid the problem of large explanatory texts in the answers generated by chat3.5.

  #58 opened 5 months ago by moonpostmen
  2

• Huggingface model

  #57 opened 8 months ago by YongLD
  1

• Naive question on OK-VQA and A-OKVQA evaluation.

  #56 opened a year ago by HenryHZY
  0

• Accuracy does not increased

  #55 opened a year ago by touQerabaS
  1

• Prerequisites Questions

  #3 opened 2 years ago by ningkekeke
  23

• 1

  #54 opened a year ago by 1286169349
  0

• How can I run this model in my custom dataset?

  #53 opened a year ago by RoxasGGBoy
  1

• skip step 1 and go directly to step 2

  #52 opened a year ago by guxiaojing
  1

• 当我在训练stage1时预训练、微调和生成候选答案时报了一样的错OSError: We couldn't connect to 'https://huggingface.co' to load this file, couldn't find it in the

  #46 opened a year ago by Celinewxy
  4

• okvqa-stage1-pretrain

  #50 opened a year ago by 123susu
  3

• Trained model

  #49 opened a year ago by Rkunique
  1

• 当我在训练stage1时预训练、微调和生成候选答案时报了一样的错 TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType 请问该怎么解决呢

  #48 opened a year ago by Celinewxy
  1

• The first candidate answer of your provided candidates_okvqa.json in assets.zip

  #45 opened a year ago by zhongfansun
  3

• OpenAI-Api Cost

  #51 opened a year ago by 123susu
  1

• KeyError: 179520 ？？

  #43 opened a year ago by Andrew82106
  13

• 当我运行bash scripts/extract_img_feats.sh时显示下面内容，但并没有生成coco2014_feats

  #44 opened a year ago by Celinewxy
  7

• 当我运行stage2的命令时，显示错误连接openAI,这个是什么原因呢？

  #47 opened a year ago by Celinewxy
  1

• assets

  #41 opened 2 years ago by Zhaihanlei
  1

• Checkpoints Availability

  #40 opened 2 years ago by lauraminkova
  2

• mcan_530_okvqa.json

  #36 opened 2 years ago by Zhaihanlei
  4

• The process of image caption

  #39 opened 2 years ago by cm-xcju
  0

• The process of image caption

  #38 opened 2 years ago by cm-xcju
  1

• mcan_530_okvqa.json

  #35 opened 2 years ago by Zhaihanlei
  1

• Could you provide a finetuned model for A-OKVQA dataset?

  #23 opened 2 years ago by huangsiyong
  3

• 请问你们跑这几个数据集花了多少钱调用api... 我才刚开始跑一小会就花了5美元。。太贵了

  #30 opened 2 years ago by 805934132
  1

• pretrain.py的代码，当RESUME的时候，start_epoch = self.__C.CKPT_EPOCH是不是有问题呢？

  #28 opened 2 years ago by yanfeng918
  1

• 预训练MCAN模型和在okvqa上微调是一起的吗？应该先预训练MCAN，再去微调。

  #27 opened 2 years ago by yanfeng918
  2

• 请问国内可以直接跑吗？ 跑起来可以直接调到openai的API？

  #26 opened 2 years ago by 805934132
  6

• could you upload file to Baidu netdisk

  #19 opened 2 years ago by verigle
  1

• 您好，关于Caption这块有疑问想请教您一下

  #24 opened 2 years ago by fmdmm
  1

• 运行“bash scripts/extract_img_feats.sh”后提示“RuntimeError: CUDA error: out of memory”

  #25 opened 2 years ago by Gujing-Ace
  2

• Error communicating with OpenAI

  #22 opened 2 years ago by hc0527
  5

• when I skip stage1 and execute stage2, the dataset size is 0

  #20 opened 2 years ago by majixin-li
  2

• OpenAI's apikey

  #18 opened 2 years ago by Zhaihanlei
  1

• could prophet work with LLaMA(from facebook)

  #12 opened 2 years ago by zloop1982
  1

• very good guy and very impressive story

  #16 opened 2 years ago by kaiwang960112
  1

• bug

  #17 opened 2 years ago by Zhaihanlei
  2

• multiple GPUs

  #13 opened 2 years ago by hc0527
  2

• 50GB memory？

  #14 opened 2 years ago by 2201957
  1