Reranker & Embedding: Qwen3-VL single-shot inference with single-specialization compile#1031
Open
quic-amitraj wants to merge 11 commits into
Open
Reranker & Embedding: Qwen3-VL single-shot inference with single-specialization compile#1031quic-amitraj wants to merge 11 commits into
quic-amitraj wants to merge 11 commits into
Conversation
quic-amitraj
changed the title
vbaddi
requested changes
Jun 4, 2026
quic-amitraj
added
embedding
quic-amitraj
changed the title
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit <amitraj@qti.qualcomm.com> Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Mirror of the reranker fix: Qwen3-VL embedding is single-shot prefill (reads last-token hidden state as embedding vector, no decode loop). `get_compile_specs` now returns ctx_len == prefill_seq_len, triggering Solution A in modeling_auto.py to compile only the Prefill kernel. Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
… simplify config path Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
Contributor
Author
|
@quic-rishinr @vbaddi Please review it, added few more changes.
|
…out kv input outpur Signed-off-by: Amit Raj <amitraj@qti.qualcomm.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
This PR adds end-to-end AI100 inference support for Qwen3-VL multimodal reranker and embedding models, and fixes the compile pipeline so both model types always produce exactly one QPC specialization (Prefill only — no wasted Decode kernel).
Supported models:
Qwen/Qwen3-VL-Reranker-2BandQwen/Qwen3-VL-Reranker-8BQwen/Qwen3-VL-Embedding-8BTest Results