DachengLi1/LongChat

Issues

• lmsys/longchat-7b-v1.5-32k is a base model or a aligned model?

  #43 opened 6 months ago by hsiehjackson
  0

• How to prepare the training data

  #42 opened a year ago by ycsun1972
  2

• Update Anthropic Client

  #34 opened a year ago by krrishdholakia
  2

• Inference is very slow on long text input

  #39 opened a year ago by Colafei0406
  1

• license

  #37 opened a year ago by yzhang123
  1

• How was the 18k dataset prepared?

  #5 opened a year ago by musabgultekin
  3

• dummy conversations seem to be short

  #40 opened a year ago by Arist12
  2

• OOM issue

  #28 opened a year ago by WeixuanXiong
  4

• Add support for flash attention with use_cache

  #3 opened a year ago by DachengLi1
  1

• Hi, using xformers monkey patch training llama2 got loss explosion

  #35 opened a year ago by lucasjinreal
  0

• Why the use of flash attention in the inference stage will lead to slower？

  #27 opened a year ago by xyfZzz
  2

• flash attention rename

  #33 opened a year ago by Arist12
  1

• Do you support Llama-2-13b model data？

  #32 opened a year ago by brewswang
  0

• train ValueError

  #31 opened a year ago by brewswang
  0

• flash_attn installed, but got ImportErrorImportError

  #30 opened a year ago by brewswang
  4

• Output token limit

  #29 opened a year ago by MoppyDu97
  0

• Maybe a bug in the preprocess?

  #26 opened a year ago by Richar-Du
  3

• About the print message

  #25 opened a year ago by lucasjinreal
  2

• About the learning rate

  #19 opened a year ago by lucasjinreal
  1

• Xformers Monkey Patch Compatibility

  #21 opened a year ago by fahadh4ilyas
  1

• Longchat inference configuration

  #23 opened a year ago by SeekWrldTea
  1

• longchat-13b-16k chat not work

  #14 opened a year ago by ahkimkoo
  9

• Can inference be run on consumer hardware?

  #8 opened a year ago by GrahamboJangles
  8

• OutOfMemoryError: CUDA out of memory.

  #9 opened a year ago by brewswang
  5

• torch.distributed.elastic.multiprocessing.errors.ChildFailedError:

  #24 opened a year ago by ChaoyuHuang
  0

• Web GUI for longchat

  #12 opened a year ago by VVNMA
  3

• The purpose of pretrain script?

  #17 opened a year ago by fahadh4ilyas
  2

• Monkey Patch Xformers use `past_key_value` but `use_cache` can't be `True`?

  #15 opened a year ago by fahadh4ilyas
  9

• Support for other model like Baichuan

  #20 opened a year ago by lucasjinreal
  0

• why not reuse fschat code?

  #16 opened a year ago by lucasjinreal
  8

• Will it support qlora?

  #18 opened a year ago by lw3259111
  1

• Add scripts on querying closed sourced models

  #7 opened a year ago by DachengLi1
  1

• Add scripts to generate more testcases

  #6 opened a year ago by DachengLi1
  1

• How to use 3090 to train 16k model?

  #4 opened a year ago by aresa7796
  7

• Multi-node training?

  #11 opened a year ago by XueFuzhao
  1

• Load the model for inference?

  #10 opened a year ago by fahadh4ilyas
  4

• unsupervised pre-training on the model

  #2 opened a year ago by wqn1
  5