Skip to content

[SYCL] add concat through dim 1/2 #8483

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

Merged
merged 2 commits into from
Jul 15, 2024
Merged

[SYCL] add concat through dim 1/2 #8483

merged 2 commits into from
Jul 15, 2024

Conversation

airMeng
Copy link
Collaborator

@airMeng airMeng commented Jul 15, 2024

@github-actions github-actions bot added ggml changes relating to the ggml tensor library for machine learning SYCL https://en.wikipedia.org/wiki/SYCL - GPU programming language labels Jul 15, 2024
@airMeng airMeng requested a review from NeoZhangJianyu July 15, 2024 02:47
@airMeng
Copy link
Collaborator Author

airMeng commented Jul 15, 2024

@characharm can you try deepseek on this branch? Thank you in advance, your support is our motivation

@airMeng airMeng requested a review from OuadiElfarouki July 15, 2024 03:37
@airMeng
Copy link
Collaborator Author

airMeng commented Jul 15, 2024

@OuadiElfarouki, I'm wondering who I should tag for PR reviews at Codeplay. Could you or someone else represent Codeplay for the following PRs? For any Intel-related issues, feel free to tag me.

Copy link
Collaborator

@NeoZhangJianyu NeoZhangJianyu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I test UT successfully.
And verify DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf on Arc770 is passed.

@OuadiElfarouki
Copy link
Contributor

@airMeng Thanks for the tag. You can either reach out to @joeatodd (Team PO at Codeplay), or @Alcpz or myself.

@airMeng airMeng requested a review from joeatodd July 15, 2024 09:31
Copy link
Contributor

@OuadiElfarouki OuadiElfarouki left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM Thanks!

@airMeng airMeng merged commit 16bdfa4 into master Jul 15, 2024
54 checks passed
@airMeng airMeng deleted the sycl-concat branch July 15, 2024 11:32
@characharm
Copy link
Contributor

characharm commented Jul 15, 2024

I test UT successfully. And verify DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf on Arc770 is passed.

Could you please check this model: DeepSeek-Coder-V2-Lite-Instruct-Q6_K? link I have tried several times, but the computer completely froze every time after loading the model into memory. tested on this build b3398

@airMeng
Copy link
Collaborator Author

airMeng commented Jul 16, 2024

I test UT successfully. And verify DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf on Arc770 is passed.

Could you please check this model: DeepSeek-Coder-V2-Lite-Instruct-Q6_K? link I have tried several times, but the computer completely froze every time after loading the model into memory. tested on this build b3398

seems the model file itself is 14.1GB, how much vram your GPU has?

@characharm
Copy link
Contributor

I test UT successfully. And verify DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf on Arc770 is passed.

Could you please check this model: DeepSeek-Coder-V2-Lite-Instruct-Q6_K? link I have tried several times, but the computer completely froze every time after loading the model into memory. tested on this build b3398

seems the model file itself is 14.1GB, how much vram your GPU has?

16gb, but other models with the same or even slightly larger file size work fine

@characharm
Copy link
Contributor

characharm commented Jul 18, 2024

b3408 , now it shuts down at this,tried with smaller quants, with the same result

image

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ggml changes relating to the ggml tensor library for machine learning SYCL https://en.wikipedia.org/wiki/SYCL - GPU programming language
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants