Qwen3-VL-2B export#68
Open
carinapeng wants to merge 9 commits into
Open
Conversation
Add Qwen3-VL vision-language model export support. - gpu/qwen3_vl.py: Qwen3VLForCausalLM (text decoder, input_ids) and Qwen3VLForCausalLMEmbeddings (inputs_embeds variant for VLM fusion) - primitives/macos/cache_scatter.py: slice_scatter-based explicit KV cache (avoids stateful mutable_slice_update Metal kernel issues) - export_qwen3vl_explicit_kv.py: text decoder export (inputs_embeds, explicit KV) - export_vision_encoder_224.py: vision encoder export (448x448 -> 196 visual tokens) - registry.py: register qwen3_vl model entry
stikves
reviewed
Jun 30, 2026
stikves
reviewed
Jun 30, 2026
| Usage: | ||
| cd <repo-root> | ||
| uv run python python/export_qwen3vl.py [--max-ctx 4096] [--num-layers N] |
Contributor
There was a problem hiding this comment.
shall we use the existing pattern?
uv run coreai.vlm.export qwen3-vl
or similar
And also update
- Model registry
- A new model specific subfolder (or
models/vlm, ormodels/qwen) - README?
Contributor
Author
There was a problem hiding this comment.
We don't have coreai.vlm.* namespace yet, at least in pyproject, this is a bigger refactor, I think we can have a follow up for it since the scope of this is a first model support for VLM infra?
Contributor
There was a problem hiding this comment.
Should be a simple refactor, and will help with discoverability.
stikves
reviewed
Jun 30, 2026
stikves
reviewed
Jun 30, 2026
stikves
left a comment
stikves
left a comment
Contributor
There was a problem hiding this comment.
Overall looks good, but we might want to move the model to a better home
models/qwen3-vl or models/vlm?
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.
CoreAI export for Qwen3-VL-2B