Skip to content

DS-Chat BLOOM: Fix Attention mask#4338

Merged
lekurile merged 1 commit intomasterfrom
lekurile/fix_ds_chat_bloom
Sep 14, 2023
Merged

DS-Chat BLOOM: Fix Attention mask#4338
lekurile merged 1 commit intomasterfrom
lekurile/fix_ds_chat_bloom

Conversation

@lekurile
Copy link
Contributor

This PR fixes the input mask type in DS attention when the BLOOM model is used with the Hybrid Engine in DeepSpeed-Chat.

Thanks to @shenzhuo for the debug efforts in #3518.

@lekurile lekurile added this pull request to the merge queue Sep 14, 2023
Merged via the queue into master with commit 0c83436 Sep 14, 2023
CurryRice233 added a commit to CurryRice233/DeepSpeed that referenced this pull request Sep 15, 2023
* origin/master: (48 commits)
  Fix autotune to support Triton 2.1  (deepspeedai#4340)
  Fix skipped inference tests (deepspeedai#4336)
  Suppress noise (deepspeedai#4310)
  Fix a bug in the implementation of dequantization for inference (deepspeedai#3433)
  DS-Chat BLOOM: Fix Attention mask (deepspeedai#4338)
  clear redundant timers (deepspeedai#4308)
  Add release version checking (deepspeedai#4328)
  Fix Zero3 contiguous grads, reduce scatter false  accuracy issue (deepspeedai#4321)
  Clean up modeling code (deepspeedai#4320)
  Handle empty parameter groups (deepspeedai#4277)
  Update README.md (deepspeedai#4316)
  README update (deepspeedai#4303)
  Update release and bump patch versioning flow (deepspeedai#4286)
  added a bert-model check for triton (deepspeedai#4266)
  ZeRO-Inference v2 release
  bump to 0.10.4
  Update index.md (deepspeedai#4297)
  fix user args parsing of string with spaces on runner (deepspeedai#4265)
  ZeRO-Inference refresh (deepspeedai#4197)
  AMD Kernel Compatibility Fixes (deepspeedai#3180)
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants