Downloads · 30 days
0
portal-vlm-gemma3-ported
manihani4/portal-vlm-gemma3-ported
portal-vlm-gemma3-ported is a machine learning model from manihani4. 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 mit.
Cross-family converter-only port: qwen25vl-hypernet with a fresh Gemma-3 converter refit on the full calibration split; the converter carries coordinate format as well as placement (Gemma-3 has no native dialect).
Access
Public
Updated Aug 2, 2026
Repo size
675 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt675 MB · 100%
At a glance
- License
- mit
- Access
- Public
- Created
- Aug 2, 2026
- Updated
- Aug 2, 2026
- SHA
- 0d22b6fb
Base models
Model card
From the Hugging Face model README
portal-vlm: gemma3-ported
Cross-family converter-only port: qwen25vl-hypernet with a fresh Gemma-3 converter refit on the full calibration split; the converter carries coordinate format as well as placement (Gemma-3 has no native dialect).
Part of the portal-vlm release - an independent replication of Ramp Labs' PorTAL (portable task adapters via hypernet-generated LoRA) extended to vision-language models on GUI grounding.
| ScreenSpot-v2 overall | web split | |
|---|---|---|
| this artifact | 29.6% | 19.7% |
- Base model:
google/gemma-3-4b-it@093f9f388b31de276ce2de164bdc2081324b9767 - Training config:
configs/gemma3_ported.yaml - Eval record:
results/gemma3_ported.json
Reproduce this row without training
git clone https://github.com/robbym-dev/portal-vlm && cd portal-vlm && uv sync
uv run python scripts/eval.py --config configs/gemma3_ported.yaml --adapter hf:manihani4/portal-vlm-gemma3-ported
hypernet.pt is a PorTAL hypernet checkpoint (task latent + decoder core + per-base converters), not a PEFT adapter: at eval time the hypernet generates r8 LoRA factors for the target base, which scripts/eval.py injects on the fly.
- License
- mit
- Created
- Aug 2, 2026
- Updated
- Aug 2, 2026