Issues

• About Experiences in Section 4.2 (Performance on Synthetic Long Context Tasks) in the paper

  datamllab/LongLM#49 · Filter by author zhlgg opened on Dec 28, 2024
  1

• Memory usage

  datamllab/LongLM#48 · Filter by author scar-on opened on Dec 13, 2024

• How to select params for self extend? (llama3.1 and gemma2)

  datamllab/LongLM#47 · Filter by author hahmad2008 opened on Nov 20, 2024

• About GPU memory usage

  datamllab/LongLM#45 · Filter by author zhongsanqiang opened on Jul 26, 2024

• LongLM really has great potential.

  datamllab/LongLM#44 · Filter by author lh0x00 opened on Jul 15, 2024
  1

• TypeError: 'NoneType' object is not subscriptable

  datamllab/LongLM#43 · Filter by author Lilly-25 opened on Jul 3, 2024
  1

• Cohere command r

  datamllab/LongLM#42 · Filter by author flaviusburca opened on Jun 4, 2024
  1

• Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! when resuming training

  datamllab/LongLM#37 · Filter by author humza-sami opened on May 14, 2024

• llama3 is not working.

  datamllab/LongLM#36 · Filter by author rayjang opened on Apr 26, 2024
  1

• Question | Has anyone tried this with GGUF models?

  datamllab/LongLM#32 · Filter by author Cognitus-Stuti opened on Apr 11, 2024
  2

• What effect on qwen1.5 will be if i use self-extend trick?

  datamllab/LongLM#30 · Filter by author WeixuanXiong opened on Mar 28, 2024
  4

• FlashAttention does not work for Batch size > 1

  bug
  datamllab/LongLM#29 · Filter by author Mooler0410 opened on Mar 28, 2024

• Support with vLLM

  datamllab/LongLM#17 · Filter by author Aniketto16 opened on Feb 1, 2024
  5