regisss

Results 220 comments of regisss

llava inference works incorrectly if adapt_transformers_to_gaudi called after transformers import

Yes, that's expected. A workaround would be to fully re-import all the imports that are done in `adapt_transformers_to_gaudi`, but it's not that straightforward to do. Let's see if some other...

Quantization failed

@endomorphosis Are we still having these issues?

Quantization failed

I can reproduce. @HolyFalafel Any idea why this happens?

Improve greedy search memory usage

CI failed but it doesn't seem to be related to this PR

sasarkar/qwen finetuning bucketing

@ssarkar2 This will only work for Qwen? Or is it also appliable with other models?

fea(): Diffuser upgrade to 0.32.0

@imangohari1 We will merge this one to main directly, but this will lead to many merge conflicts. I think it would just be easier if you start a new branch...

CodeGen inference error "synNodeCreateWithId failed for node: batch_gemm with synStatus 26"

I cannot reproduce it. Can you try again upgrading the lib with ``` pip install -U optimum-habana ``` ?

CodeGen inference error "synNodeCreateWithId failed for node: batch_gemm with synStatus 26"

> This issue can only be reproduced on PRC sku which FP32 GEMM is disabled. What's PRC sku?

Upgrade to Transformers 4.43

@avbodas It should work with the latest commit I pushed

Upgrade to Transformers 4.43

@imangohari1 It's solved for PEFT and BridgeTower. I'll have to check that the other examples with PEFT work with v0.12. @vidyasiv I added the changes for all these tests.