Add position_ids to BloomForCausalLM forward pass#44706
Closed
saivedant169 wants to merge 2 commits into
Closed
Conversation
Threads position_ids through BloomForCausalLM -> BloomModel for API consistency with other CausalLM models (Llama, Falcon, Gemma, etc.). Bloom uses ALiBi for positional encoding, so position_ids is accepted but not consumed by the attention layer — this matches the pattern where GenerationMixin passes position_ids and the model should accept it explicitly rather than silently absorbing it through **kwargs. Part of huggingface#32937 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: bloom |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes part of #32937
What does this PR do?
Adds
position_idsas an explicit parameter toBloomForCausalLM.forward()andBloomModel.forward(), bringing Bloom in line with other CausalLM models like Llama, Falcon, Gemma, and Mistral.Bloom uses ALiBi for positional encoding, so
position_idsisn't consumed by the attention computation. But it should still be accepted explicitly rather than silently absorbed through**kwargs:GenerationMixinconstructs and passesposition_idsduring autoregressive decoding. Without it in the signature, the parameter goes into**kwargsand is silently ignored — which works but is fragile and inconsistent.This is a follow-up to #44705 (RoFormer), continuing the effort to standardize the CausalLM forward signature across architectures.
How was it tested?
make stylecleanCoordination
Commented on #32937 here. This is part of a series — RoFormer (#44705) was the first.