Downloads · 30 days
6
13% of all-time downloads
internvl2-5-2b-rm-v2-no-oversample
fengyikang/internvl2-5-2b-rm-v2-no-oversample
internvl2-5-2b-rm-v2-no-oversample is a machine learning model from fengyikang. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as other.
v2 recipe: MLP head, length debiasing, margin loss, resumed from v1 final. Unique training pairs: 38,496.
All-time downloads
45
Public
Parameters
2.2B
4.4 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors4.4 GB · 100%
At a glance
- License
- other
- Model type
- internvl_chat
- Access
- Public
- Created
- Jun 8, 2026
- Updated
- Jun 8, 2026
- SHA
- a4c5a889
Base models
Model card
From the Hugging Face model README
InternVL2.5-2B Reward Model (v2, MLP, no oversampling)
v2 recipe: MLP head, length debiasing, margin loss, resumed from v1 final. Unique training pairs: 38,496.
Metrics (pointwise + K=5)
| Overall | Macro | agree | reject | tie |
|---|---|---|---|---|
| 67.69% | 63.48% | 790 | 377 | 80 |
Usage
python eval_reward_vlrewardbench.py \
--checkpoint-dir ./ \
--base-model-path OpenGVLab/InternVL2_5-2B \
--use-mlp-head \
--scoring pointwise --k 5
- Type
- internvl_chat
- License
- other
- Created
- Jun 8, 2026
- Updated
- Jun 8, 2026