Clean log probs
What does this PR do ?
- Fixes skip_prompt_log_probs to make sure it works
- Made calculate log probs a little more efficient if materialize last token logits is set to true or is decode only
- Added test case to check log probs properly
This pull request requires additional validation before any workflows can run on NVIDIA's runners.
Pull request vetters can view their responsibilities here.
Contributors can view more details about this message here.
@santhnm2 @tdene can you please take a look at this MR?
/ok to test 0e5e279
/ok to test 97dcd9a
/ok to test a07978d
/ok to test a07978d
/ok to test a07978d
@shanmugamr1992, there was an error processing your request: E2
See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/2/
/ok to test da4c0e8
/ok to test f94ae26
/ok to test 6fcbb84
/ok to test 62fdc5a
/ok to test 2ce454e
/ok to test 2ce454e27834fbcb947102e2a41241e03a28d52b
/ok to test 2ce454e27834fbcb947102e2a41241e03a28d52b
/ok to test 2ce454e27834fbcb947102e2a41241e03a28d52b
/ok to test ff035d6