xiaoachen98/Open-LLaVA-NeXT

Issues

• 想学习【MLLM多模态】快加学习交流群

  #19 opened 17 days ago by km1994
  0

• Pretraining with im_start_end token

  #18 opened a month ago by Ali2500
  0

• Is this still active?

  #17 opened 2 months ago by chris-hoertnagl
  1

• loss curve of llava-next-llama3

  #12 opened 3 months ago by simplelifetime
  2

• Inverse Loss Spike issue

  #16 opened 2 months ago by LaBaZh
  1

• Would you plan to adapt it to qwen2-7B?

  #13 opened 3 months ago by Nastu-Ho
  3

• Flashattention issue

  #15 opened 3 months ago by LaBaZh
  1

• srun: unrecognized option '--quotatype=reserved'

  #14 opened 3 months ago by LaBaZh
  1

• loss curve of SFT on vicuna-7b

  #9 opened 3 months ago by Xiaohui9607
  6

• llama3 finetune time

  #11 opened 3 months ago by Xiaohui9607
  1

• Vision Tower

  #10 opened 3 months ago by homiec
  1

• anyres in open-llava-next v.s. s2 in llava

  #8 opened 4 months ago by LIO-H-ZEN
  0

• About the MMMU performance

  #7 opened 4 months ago by LIO-H-ZEN
  0

• Is the param "self.image_newline" only tuned in SFT?

  #6 opened 4 months ago by huangrh99
  1

• data generation code

  #4 opened 4 months ago by trinhvg
  1

• vit not saved

  #5 opened 4 months ago by RifleZhang
  2

• does it support llava-next-video yet?

  #3 opened 4 months ago by dragen1860
  1

• [Question] About finetuning projector

  #2 opened 4 months ago by JY-CCK
  1