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

[BOUNTY] Add Phi-2 #117

Merged
merged 16 commits into from
Sep 3, 2024
Merged

Conversation

JushBJJ
Copy link
Contributor

@JushBJJ JushBJJ commented Jul 26, 2024

Issues that need to be addressed: tenstorrent/tt-buda#42
Video of it working: https://www.youtube.com/watch?v=a3gCi8B87zA
Card: Grayskull e75

Requirements

  • Transformers >=4.42.0
  • ~40GB+ depending on how long the generation is

Bounty: #21

@milank94
Copy link
Collaborator

Awesome @JushBJJ -- same comments as #37 on the formatting of the demo.

@milank94 milank94 self-requested a review July 30, 2024 14:10
@milank94 milank94 self-assigned this Jul 31, 2024
@JushBJJ
Copy link
Contributor Author

JushBJJ commented Jul 31, 2024

I'll add it to Phi after I rewrite Qwen's file
tenstorrent/tt-buda#42 (comment)

@JushBJJ
Copy link
Contributor Author

JushBJJ commented Aug 1, 2024

This should be good now @milank94 for final review and testing. Although note that Phi2 is incredibly slow on an e75 and RAM usage is massive, when only 9GB is used initially the usage can go up to 50GB+ depending on how long the generation takes.

@milank94
Copy link
Collaborator

Hey @JushBJJ can you switch the target branch to be: mkordic/rc_20240830?

@JushBJJ JushBJJ changed the base branch from main to mkordic/rc_20240830 August 31, 2024 00:50
@JushBJJ
Copy link
Contributor Author

JushBJJ commented Aug 31, 2024

Done

@milank94 milank94 merged commit 3c71b9f into tenstorrent:mkordic/rc_20240830 Sep 3, 2024
@milank94 milank94 mentioned this pull request Sep 3, 2024
anirudTT pushed a commit that referenced this pull request Sep 24, 2024
* Qwen1.5 0.5B pybuda implementation

* remove unneeded requirement

* rename "acceleration" to "accelerate"

* Update env vars and compiler configs

* remove undefined device_map

* Remove misleading and unnecessary environment variables

* remove qwen from phi branch

* Add Phi 2

* Update requirements.txt

* Standardize Phi2 demo and added tests

* Remove old phi2 demo

* fix missing quote in pyproject.toml

* fix

* Fix test saying qwen1_5 instead of phi2
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants