Skip to content

[Auto] Refactor output tracing decorators (cluster-43979-11): merged 6 of 10 PRs#33

Open
evalstate wants to merge 21 commits intomainfrom
merge-cluster-cluster-43979-11-20260424132122
Open

[Auto] Refactor output tracing decorators (cluster-43979-11): merged 6 of 10 PRs#33
evalstate wants to merge 21 commits intomainfrom
merge-cluster-cluster-43979-11-20260424132122

Conversation

@evalstate
Copy link
Copy Markdown
Owner

Cluster: cluster-43979-11
Base: origin/main

Merged PRs:

Skipped PRs:

Failed PRs:

Notes:

Next steps:

  • Review merged output-tracing behavior semantically before upstreaming.
  • Run targeted model tests for cvt, fnet, regnet, resnet, rt_detr, mobilenet_v2, deberta_v2, codegen, gptj, and efficientnet.
  • Rework SpeechT5 and VITS manually against current main if desired.

beelapranay and others added 21 commits February 14, 2026 13:04
Migrate GPT-J from manual boilerplate output collection to the new
decorator-based output tracing system:

- Add _can_record_outputs to GPTJPreTrainedModel
- Add @capture_outputs and @merge_with_config_defaults to GPTJModel.forward
- Add @can_return_tuple to GPTJForCausalLM, GPTJForSequenceClassification,
  and GPTJForQuestionAnswering forwards
- Simplify GPTJBlock.forward to return hidden_states directly
- Remove output_attentions, output_hidden_states, return_dict params
  from signatures (now handled by decorators)
- Propagate changes to CodeGen via Copied from annotation

Net reduction of ~70 lines of boilerplate code.
@evalstate
Copy link
Copy Markdown
Owner Author

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants