RenShuhuai-Andy/TimeChat

Issues

• Ask for reproducing

  #40 opened 6 months ago by HYOJINPARK
  6

• How to resume training?

  #45 opened 3 months ago by GroundMoRe
  0

• Experiments on ActivityNet-Captions

  #44 opened 4 months ago by minjoong507
  0

• q-former和llama的词典不一样，一个是bert，一个是llama。能通用吗？

  #43 opened 4 months ago by guantao18
  1

• Question about the Text Input to the LLM

  #42 opened 5 months ago by ShramanPramanick
  2

• 不能回答中文

  #41 opened 6 months ago by wublubdubdaxml
  1

• Inference with audio

  #29 opened 6 months ago by lakshya-frontera
  2

• Long video test results did not meet expectations

  #38 opened 6 months ago by ffiioonnaa
  2

• Discussion : Steps for swapping Llama 2 with Llama 3

  #37 opened 6 months ago by rahulkrprajapati
  1

• Data type not aligned

  #39 opened 6 months ago by KKKLeon
  1

• 用自己的数据集finetune，如何在train的过程中进行eval？

  #32 opened 7 months ago by changqinyao
  1

• Can this model do qa tasks？

  #26 opened 7 months ago by leexinhao
  2

• Could you test TimeChat on the EgoShema dataset?

  #34 opened 7 months ago by EricLina
  2

• Asking for the Fine-tuned Checkpoint

  #35 opened 7 months ago by minjoong507
  2

• Weight for QA benchmarks

  #36 opened 7 months ago by NIneeeeeem
  3

• Why the result of temporal video grounding is always the multiple of 5?

  #33 opened 7 months ago by zhengrongz
  5

• Question about the output of the time-aware frame encoder

  #28 opened 7 months ago by Mingxiao-Li
  2

• Questions about the provided fine-tuning model parameters

  #30 opened 7 months ago by LanXingXuan
  1

• Based transformers version needed for modifying models/modeling_llama.py

  #31 opened 7 months ago by yeahjack
  3

• Question about fune-tune

  #25 opened 7 months ago by zhengxingmao
  7

• Question about prompt

  #20 opened 8 months ago by Ironieser
  5

• What is the relationship between segment and timetoken？

  #17 opened 9 months ago by sunwhw
  3

• When conducting SFT experiments, setting batch_size_train to 1 or 2 has the same memory usage.

  #27 opened 8 months ago by tiesanguaixia
  0

• For different video datasets, is the frame density always drawn at intervals of 1 second?

  #2 opened a year ago by DuoLong
  5

• the performance is very low on my own dataset.

  #22 opened 8 months ago by onlyonewater
  5

• Subset of YT-Temporal

  #24 opened 8 months ago by patrick-tssn
  1

• Question about batch size

  #23 opened 8 months ago by gyxxyg
  1

• Question about the tokenizer

  #19 opened 9 months ago by gyxxyg
  5

• Experiment-related question

  #21 opened 9 months ago by zhaodongliang678
  3

• RAM and VRAM requirement

  #13 opened 9 months ago by Coronal-Halo
  2

• Question about prompts.

  #18 opened 9 months ago by gyxxyg
  2

• Inquiry on training cost

  #16 opened 9 months ago by HenryHZY
  2

• Demo can‘t show the same desult

  #15 opened 9 months ago by xiaoxiaoli666
  1

• Bad performance of Charades

  #14 opened 9 months ago by soyeonhong
  1

• Do we need to crop the HiREST videos?

  #10 opened 10 months ago by yeliudev
  14

• Seeking Clarification about Fine-tuning Datasets

  #12 opened 10 months ago by ShramanPramanick
  2

• Details of sliding qformer operation

  #11 opened 10 months ago by jihwanp
  1

• the generalization performance is bad when testing on custom videos.

  #8 opened 10 months ago by dragen1860
  1

• torch.load raise TypeError: 'strict' is an invalid keyword argument for Unpickler()

  #9 opened a year ago by wwq66
  4

• Error in loading Video-LLaMA-2-7b_Finetuned

  #7 opened a year ago by dragen1860
  1

• When will the checkpoint and demo scripts be released?

  #5 opened a year ago by Hugh0120
  2

• how to evaluation on activitynet-DVC?

  #6 opened a year ago by TXH-mercury
  3

• UnsatisfiableError

  #4 opened a year ago by LarryLeeee
  4

• Checkpoints to run demo and dataset

  #3 opened a year ago by fazliimam
  1

• A very good video-related work, it is convenient to open source the data set？

  #1 opened a year ago by Xujianzhong
  1