model: Add tomoro-colqwen3-embed embedding models#3627
Merged
KennethEnevoldsen merged 14 commits intoembeddings-benchmark:mainfrom Dec 3, 2025
Merged
model: Add tomoro-colqwen3-embed embedding models#3627KennethEnevoldsen merged 14 commits intoembeddings-benchmark:mainfrom
KennethEnevoldsen merged 14 commits intoembeddings-benchmark:mainfrom
Conversation
…milarity scoring - Changed default dtype from float16 to bfloat16 for improved performance. - Added max_num_visual_tokens parameter to AutoProcessor initialization. - Refined embedding extraction logic to avoid boolean casting issues. - Introduced support for score_multi_vector in similarity computation. - Added new model metadata for colqwen3_4b with relevant attributes.
Samoed
reviewed
Nov 26, 2025
Samoed
reviewed
Nov 26, 2025
5 tasks
Samoed
added
the
new model
Samoed
reviewed
Nov 26, 2025
Contributor
Author
|
Updated the huggingface revision with updated processor supporting fused embeddings. It was tested locally. |
Contributor
There was a problem hiding this comment.
Great addition!
A few minor things on the metadata, otherwise do see @Samoed's comment on get_fused_embeddings
Samoed
reviewed
Nov 29, 2025
Co-authored-by: Roman Solomatin <samoed.roman@gmail.com>
Contributor
Author
|
Any remaining items from my side? |
Member
|
I don't think so |
KennethEnevoldsen
approved these changes
Dec 3, 2025
KennethEnevoldsen
changed the title
KennethEnevoldsen
merged commit 71ac96c
into
embeddings-benchmark:main
9 of 10 checks passed
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add inference code and requirements for tomoro-colqwen3-embed 8B and 4B models.
They are based on merged Qwen3-VL-Instruct and Qwen3-Embedding checkpoints finetuned using the ColQwen method. They produce 320-dimension embeddings per text/image token. Finetuning data is a subset of nvidia/llama-nemoretriever-colembed-3b-v1 training data.
Model checkpoints available at:
https://huggingface.co/TomoroAI/tomoro-colqwen3-embed-4b
https://huggingface.co/TomoroAI/tomoro-colqwen3-embed-8b
Checklist:
mteb.get_model(model_name, revision)andmteb.get_model_meta(model_name, revision)