DeepSpeed
a3926bbb - infV2 fix for OPT size variants (#4694)

Commit

1 year ago

infV2 fix for OPT size variants (#4694) Co-authored-by: Jeff Rasley <jerasley@microsoft.com>

References

#4694 - infV2 fix for OPT size variants

Author

mrwyattii

mrwyattii

Parents

Files5

.github/workflows
- nv-a6000.yml
deepspeed/inference/v2
- engine_factory.py
- model_implementations
  - layer_container_base.py
  - opt
    - container.py
    - policy.py