Repository navigation
feat: torch.compile for OCR det/rec inference - #12
Open
abimaelmartell wants to merge 2 commits into
Open
abimaelmartell wants to merge 2 commits into
abimaelmartell wants to merge 2 commits into
Conversation
Patch OCR detection and recognition models to use torch.compile with dynamic shapes. Controlled via TORCH_COMPILE env var (default: enabled). Compilation happens lazily on first inference, then subsequent calls use optimized Triton kernels.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
torch.compile(dynamic=True)for optimized GPU kernels via TritonTORCH_COMPILEenv var (set to0to disable)TORCH_COMPILE_MODE(default:default)Targets the two models with the most forward passes per request (OCR-det has ~255 resolution groups). YOLO models (layout/MFD) are skipped since they go through ultralytics' predict pipeline.
Pairs well with the warmup PR (#11) — warmup inference triggers compilation so real requests hit pre-compiled kernels.