Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Version] v1.1.0. #67

Merged
merged 1 commit into from
Nov 30, 2023
Merged

[Version] v1.1.0. #67

merged 1 commit into from
Nov 30, 2023

Conversation

Duyi-Wang
Copy link
Contributor

v1.1.0 - Baichuan models supported.

Models

  • Introduced Baichuan models support and added the convert tool for Baichuan models.

Performance Optimizations

  • Update xDNN to version 1.2.1 to improve performance of BF16 data type with AMX instruction on 4th generation Intel Xeon Scalable processors.
  • Improved performance of BF16 data type inference by adding matMul bf16bf16bf16 primitives and optimizing kernel selection strategy.
  • Improved performance of the model with unbalanced split allocation.

Functionality

  • Auto-detect oneCCL environment and enter single-rank model if oneCCL does not exist.
  • Add sample strategy for token search, support temperature, top k, and top P parameter.
  • Introduced grouped-query attention support for Llama2.
  • Remove mklml 3rd party dependency.
  • Add C++ exit function for multi-rank model.
  • Export normalization and position embedding C++ API, including alibi embedding and rotary embedding.
  • Introduced XFT_DEBUG_DIR environment value to specify the debug file directory.

BUG fix

  • Fix runtime issue of oneCCL shared memory model.
  • Fix path concat issue in convert tools.

@changqi1
Copy link
Contributor

@Duyi-Wang Please create VERSION file to track release info.

@Duyi-Wang
Copy link
Contributor Author

@Duyi-Wang Please create VERSION file to track release info.

you mean a version file just containing a version number? or containing the changelog?

@changqi1
Copy link
Contributor

changqi1 commented Nov 21, 2023

@Duyi-Wang Please create VERSION file to track release info.

you mean a version file just containing a version number? or containing the changelog?

Yes. includes version number and the changelog.

@Duyi-Wang
Copy link
Contributor Author

@Duyi-Wang Please create VERSION file to track release info.

you mean a version file just containing a version number? or containing the changelog?

Yes. includes version number and the changelog.

I will create a version file including the version number. Is changelog necessary? It is included on release page. The PyTorch project use a version file with num https://github.com/pytorch/pytorch/blob/main/version.txt

@changqi1
Copy link
Contributor

Need to rebase the base commit.

@Duyi-Wang
Copy link
Contributor Author

Need to rebase the base commit.

Wait for PR of convert tool

@Duyi-Wang Duyi-Wang marked this pull request as draft November 29, 2023 05:39
@Duyi-Wang
Copy link
Contributor Author

Add new features:

  • Introduced prefix sharing feature.
  • Introduce convert module to xfastertransformer python API.
  • Auto-detect oneCCL environment in compilation. If not detected, oneCCL will be built from source.

@Duyi-Wang Duyi-Wang marked this pull request as ready for review November 30, 2023 07:56
@changqi1 changqi1 merged commit d5e576d into intel:main Nov 30, 2023
1 check passed
@Duyi-Wang Duyi-Wang deleted the version branch December 1, 2023 01:50
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.

3 participants