Downloads · 30 days
0
ours-probe-layer-sweeps-coco-llava15-7b
ToiTenBao/ours-probe-layer-sweeps-coco-llava15-7b
ours-probe-layer-sweeps-coco-llava15-7b is a machine learning model from ToiTenBao. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Artifacts are separated by layer configuration.
Access
Public
Updated Oct 17, 2026
Repo size
15.6 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt10.7 GB · 68%
At a glance
- Access
- Public
- Created
- Sep 14, 2026
- Updated
- Oct 17, 2026
- SHA
- 4758323a
Model card
From the Hugging Face model README
Ours-Probe LLaVA-1.5-7B layer studies
Artifacts are separated by layer configuration.
single-layer/raw/: single-layerh_imagecheckpoints, caches, and results.single-layer/contrastive/: single-layerh_image - h_no-imagecheckpoints, caches, and results.multi-layer/: reserved for future multi-layer checkpoints and caches.
The single-layer study covers decoder layers 0, 4, 8, 12, 16, 20, 24, 28, and 31 with seeds 0, 42, and 1337. Training uses 6,400 COCO images, validation uses 1,600 disjoint images, and the untouched test set contains 2,000 images.
- Created
- Sep 14, 2026
- Updated
- Oct 17, 2026