Alpha-VLLM/LLaMA2-Accessory

Issues

• Eval results do not match the ones in the paper

  #207 opened a month ago by chau25102001
  0

• The model-type of Multi-Modal Full-Parameter Finetuning

  #206 opened a month ago by hekkang
  0

• How to use NPU to fine tune SPHINX

  #205 opened a month ago by XiangdongHe
  0

• Multi-Modal Full-Parameter Finetuning

  #204 opened 3 months ago by hekkang
  0

• SPHINX交流群无法扫码直接加入

  #192 opened 7 months ago by fyting
  2

• ask for help

  #202 opened 4 months ago by zcdliuwei
  2

• 论文中的混合系数β取值是多少

  #203 opened 5 months ago by deepxiaobai
  0

• 有一些问答产生乱码或者空格

  #196 opened 6 months ago by lwtgithublwt
  2

• QPEFT微调后，通过single_turn_mm测试，回答乱码了

  #200 opened 6 months ago by bibibabibo26
  2

• Tensor must be cuda and dense

  #199 opened 6 months ago by bibibabibo26
  1

• 量化微调的时候如果用flashAttention则报错

  #201 opened 6 months ago by bibibabibo26
  0

• The web demo cannot be connected

  #197 opened 6 months ago by HiHiAllen
  0

• stuck on torch.distributed.barrier()

  #195 opened 7 months ago by bibibabibo26
  0

• Clarification of SEED score

  #194 opened 7 months ago by Isaachhh
  3

• mixtral multi node

  #193 opened 7 months ago by kumagai6
  0

• About the learnable skip token for SPHINX_X

  #190 opened 7 months ago by jungle-gym-ac
  0

• The weights after the pre-training stage.

  #191 opened 7 months ago by fyting
  0

• accessory.data.conversation.dataset.LabelAllZeroError: LabelAllZeroError: None

  #188 opened 7 months ago by stjuliet
  2

• Running SPHINX without SLRUM

  #161 opened 7 months ago by babla9
  1

• Does SPHINX-MLLM support batch inference for image caption

  #154 opened 7 months ago by trouble-maker007
  1

• The QR code of wechat communication group has expired

  #189 opened 7 months ago by fyting
  1

• whether full finetune mixtral-8x7B model can be converted to huggingface format for inference with transformer api ??

  #139 opened 10 months ago by hegang1-tal
  3

• any work of Large-Dit with diffusers?

  #187 opened 7 months ago by SummerTrains
  1

• More advanced support for MoE finetune

  #186 opened 7 months ago by Jack47
  0

• Finetune of SPHINX Mixtral MoE - Questions

  #156 opened 10 months ago by babla9
  6

• LargeDiT T2i - how much GPU memory is needed for sampling?

  #185 opened 7 months ago by SapirW
  1

• Can I use the --dialog argument for QPEFT in multimodal LLaMA2?

  #182 opened 8 months ago by scy-v
  1

• Can I use the --dialog argument for QPEFT in multimodal LLaMA2?

  #181 opened 8 months ago by scy-v
  0

• SPHINX-X开源了吗？

  #180 opened 8 months ago by MayCloud052
  0

• Why does LargeDiT 3B actually print 4.2B parameters ?

  #177 opened 8 months ago by nemonameless
  6

• LargeDiT T2i, which text encoder should be used?

  #178 opened 8 months ago by Miracle2333
  4

• The performance of SPHINX Mixtral MoE

  #168 opened 9 months ago by zhongshsh
  2

• Inference with Sphiny Tiny 1k consistently generates Gibberish when using Tesla T4 Google Colab Single GPU

  #155 opened 10 months ago by shelbywhite
  2

• Question : Is there a part in the SPHINX repository that switches images to Subimage?

  #169 opened 9 months ago by SuwonPabby
  5

• Demo is not working, always loading

  #167 opened 9 months ago by WilTay1
  1

• Can you please integrate large-dit into diffusers?

  #165 opened 9 months ago by nemonameless
  0

• RuntimeError: Distributed package doesn't have NCCL built in

  #166 opened 9 months ago by dibu28
  0

• About train SPHINX-X

  #162 opened 9 months ago by becauseofAI
  0

• 无法导入模块

  #153 opened 10 months ago by bao-xiaoyi
  3

• Is it true that sparse training of Mixtral-8x7B is supported?

  #143 opened 10 months ago by young-chao
  2

• VRAM requirements for SPHINX-MoE

  #151 opened 10 months ago by sisgrad
  1

• Support for Quantized MIXTRAL 8x7B models in sparse / base mode

  #142 opened 10 months ago by PlanetMacro
  2

• mixtral8x7b nccl timeout

  #147 opened 10 months ago by bao-xiaoyi
  10

• TypeError: 'type' object is not subscriptable，请问改问题怎么解决？

  #150 opened 10 months ago by zhanghaobucunzai
  2

• slow inference

  #149 opened 10 months ago by wj210
  1

• why use srun command for mixtral-8x7B full finetune??

  #138 opened 10 months ago by hegang1-tal
  3

• SPHINX-V2: how do you make improvements of SHINX-V2 ? where can I find the tech detail report of SHINX-V2?

  #141 opened 10 months ago by sxu1997
  1

• How to evaluate SPHINX on refCOCO?

  #130 opened 10 months ago by CrazyLenmon
  5

• SPHINX-MoE-1k weights seems to be corrupted

  #136 opened 10 months ago by FangGet
  4

• SEEDBench Evaluation

  #124 opened a year ago by chancharikmitra
  2