Murali Andoorveedu

icon company CentML icon location Previously also: @aws-murandoo

Results 50 comments of Murali Andoorveedu

[Models] Add remaining model PP support

Need to ungate Pixtral for testing: https://huggingface.co/mistralai/Pixtral-12B-2409

[Models] Add remaining model PP support

@youkaichao @DarkLight1337 Possible to add this model to the HF account of the testing token?

[Models] Add remaining model PP support

Hmm how is that possible? Pixtral is only 12B and we have 4 L4

[Models] Add remaining model PP support

Also tested DBRX and Arctic locally and they passed. Didn't add to test configs cause I had to modify them to work.

[Models] Add remaining model PP support

Decoder only LM Models test failure: ``` [2024-10-03T21:16:08Z] FAILED models/decoder_only/language/test_big_models.py::test_models[32-half-EleutherAI/gpt-j-6b] - KeyError: 'transformer.h.9.ln_1.weight' --   | [2024-10-03T21:16:08Z] FAILED models/decoder_only/language/test_big_models.py::test_model_print[half-EleutherAI/gpt-j-6b] - KeyError: 'transformer.h.9.ln_1.weight'   | [2024-10-03T21:16:08Z] FAILED models/decoder_only/language/test_models.py::test_models[96-float-facebook/opt-125m] - RuntimeError: value cannot be...

[Models] Add remaining model PP support

Fixed error 2&3, error 1 passed when testing locally

[Models] Add remaining model PP support

Looks like fix to pixtral worked.

[Models] Add remaining model PP support

@youkaichao @DarkLight1337 tests pass fully

[Docs] Docs update for Pipeline Parallel

@youkaichao as soon as #6130 goes in

[Docs] Docs update for Pipeline Parallel

@simon-mo Ready for merge!