Fix generate_text.sh Megatron text-generation example working w/ DS inference #99
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR gets text-generation working in the
examples/generate_text.sh
example working w/ DS inference enabled. For the main fix, thesample_sequence_batch
function has been updated to perform the softmax when calculatinglog_probs
, instead of setting it to 1's usingtorch.ones_like(...)
, although a few minor fixes were applied as well. Extra whitespace is also removed.Currently, the
examples/generate_text.sh
example produces the same nonsensical output irrespective of input context or whether DS inference is enabled or not:This PR gets the example working w/o DS inference, e.g.:
In conjunction with PR #2579 in the DeepSpeed repo, the example also works w/ DS inference enabled e.g.: