Skip to content

[examples] Simplify Qwen3-VL configs#832

Merged
mhs4670go merged 1 commit into
Samsung:mainfrom
mhs4670go:qw
Jul 22, 2026
Merged

[examples] Simplify Qwen3-VL configs#832
mhs4670go merged 1 commit into
Samsung:mainfrom
mhs4670go:qw

Conversation

@mhs4670go

Copy link
Copy Markdown
Contributor

This commit simplifies Qwen3-VL configs and consolidate evaluation presets.

@Torrero Please check if it helps your evaluation flows.

TICO-DCO-1.0-Signed-off-by: seongwoo mhs4670go@naver.com

This commit simplifies Qwen3-VL configs and consolidate evaluation presets.

TICO-DCO-1.0-Signed-off-by: seongwoo <mhs4670go@naver.com>
@mhs4670go
mhs4670go requested a review from Torrero July 21, 2026 01:24
@Torrero

Torrero commented Jul 21, 2026

Copy link
Copy Markdown
Contributor

Thank you! Now the Qwen3-VL configs are clearer.

summary: ./out/qwen3_vl_eval_suite/llava_bench/qwen3_vl_4b.llama3_2_3b.summary.json
judge:
enabled: true
model_id: meta-llama/Llama-3.2-3B-Instruct

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We use unsloth/Llama-3.2-3B-Instruct as judge model, maybe we can set it up as default model.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, actually I thought it should be official one. But, it would be better to use what we often use. I'll modify it in a follow-up PR.

@Torrero Torrero left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall LGTM

@mhs4670go
mhs4670go merged commit 71c0f83 into Samsung:main Jul 22, 2026
7 checks passed
@mhs4670go
mhs4670go deleted the qw branch July 22, 2026 03:24
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.

2 participants