Alpha-VLLM/Lumina-mGPT

Issues

• Loss is nan, stopping training

  #37 opened 3 months ago by missTL
  1

• How to load checkpoints from local model path?

  #32 opened 5 months ago by LKAMING97
  4

• 3 x Black and White output

  #11 opened 8 months ago by Excidos
  8

• Can it be fine-tuned on the base of Lumina-mGPT-7B-768-Omni?

  #36 opened 3 months ago by lily-85
  0

• multi-images

  #35 opened 3 months ago by missTL
  0

• 加群

  #31 opened 6 months ago by Henry-Lee-real
  2

• problem with Image editing

  #16 opened 8 months ago by chkmook
  2

• inference_warning

  #34 opened 4 months ago by llllly26
  0

• Improve Hugging Face integration

  #33 opened 4 months ago by NielsRogge
  2

• How to decode image latents with different resolutions?

  #24 opened 6 months ago by JeremyCJM
  3

• WeChat链接更新

  #12 opened 8 months ago by BeBuBu
  3

• Out of memory using the default training configuration

  #28 opened 6 months ago by JacobYuan7
  6

• Floating point exception(core dumped)

  #27 opened 6 months ago by JacobYuan7
  1

• What is the scale of the training data, training machine and training time for the 7B and 30B model

  #6 opened 8 months ago by zhengyunqq
  1

• problem of free demo

  #29 opened 6 months ago by xiexing0916
  1

• training with lora？

  #30 opened 6 months ago by xiexing0916
  3

• Training the model while keeping the chameleon tokenizer

  #26 opened 7 months ago by KaiyueSun98
  0

• No editing Prompt doesn't help most image reconstruction?

  #25 opened 7 months ago by SunzeY
  3

• Possible Error in Patch Size during Tokenization

  #23 opened 7 months ago by JeremyCJM
  1

• VQGAN weights not found and WeChat group QR code expired

  #22 opened 7 months ago by bollossom
  1

• cannot generate 256*256 images?

  #20 opened 7 months ago by zhangzheyu0
  1

• Plan to open source the training data or data processing script?

  #21 opened 7 months ago by densechen
  1

• VQGAN models dont exist

  #5 opened 7 months ago by gjnave
  1

• Trainable Parameters and their precisions

  #17 opened 7 months ago by SxJyJay
  1

• can lumina-mgpt support inference speed up project such as vllm

  #19 opened 7 months ago by yyyouy
  1

• Save ckpt error

  #18 opened 8 months ago by SxJyJay
  3

• Training Scrip Error: AttributeError: 'ChameleonRMSNorm' object has no attribute 'reset_parameters'. Did you mean: 'get_parameter'?

  #14 opened 8 months ago by SunzeY
  5

• Code for fine-tuning Chameleon

  #15 opened 8 months ago by Johanan528
  0

• Potential bug on z-loss calculation

  #13 opened 8 months ago by leloykun
  6

• flash_attn_2_cuda error

  #9 opened 8 months ago by abrakjamson
  2

• Encoding error reading text_tokenizer.json

  #2 opened 8 months ago by throttlekitty
  2

• Has the lumina-mGPT done any quantitative evaluation on generation or VLM benchmark?

  #7 opened 8 months ago by muxizju
  1

• what is the memory requirement to run Lumina-mGPT

  #1 opened 8 months ago by Njasa2k
  1

• Gradio launches public/live by default

  #3 opened 8 months ago by throttlekitty
  2

• Link to 34B model leads to 404 page

  #4 opened 8 months ago by RandomInternetPreson
  3