RSI-Jev v7.0-VL family decision model
RSI-Jev offers 4B and cut-down 3B decision packages for text and images. Compare their recorded accuracy, early exits and serving requirements.
By Shanghua Gao · First Hugging Face package release: 5 Nov 2026
Qwen3.5-derived decision models with text exits at different depths. The 4B package has 32 decoder layers; the 3.30B sibling keeps its first 20 layers and existing heads without retraining. Both read text and up to four images, returning option probabilities without generated text. Images use the full depth of the chosen package. Their runtime supplies option echo and prefix-bidirectional attention.
Quick answers
What is RSI-Jev v7.0-VL family by Shanghua Gao?
RSI-Jev v7.0-VL family: Qwen3.5-derived decision models with text exits at different depths.
Who makes RSI-Jev v7.0-VL family?
Are RSI-Jev v7.0-VL family's weights open?
RSI-Jev v7.0-VL family: Public weights with the RSI-Jev runtime; no causal prefix cache Licence: Apache-2.0
Facts
- Size
- 4B package: 4.69B parameters and about 9.7 GB; 3B package: 3.30B and 6.8 GB
- Inputs
- Text and up to four images; choice, noul and score
- Licence
- Apache-2.0
- Limits
- 32,768 tokens; up to 5,120 options; overlength requests get HTTP 422
- Access
- Public weights with the RSI-Jev runtime; no causal prefix cache
- Newer package
- 3B (cut at layer 20) uploaded 7 Nov 2026
Claims and sources
- The maker reports 60.61 on the public Decision Index 0.3, with 29.4 ms median and 236 ms p95 on one RTX PRO 6000 Blackwell at default effort. This is a public-suite run, not the full private index. Training used one seed and unreleased data. Ambiguous KoBBQ questions and some medicine and customer-support tasks regress against v6.1-VL. vendor reportedsource
- For the released 3B package, the maker reports public Decision Index 0.3 score 60.01, 19.6 ms median and 151 ms p95 on one RTX PRO 6000 Blackwell. All 4,001 checked rows match the 4B at medium effort bitwise. Against the full-depth 4B, held-out accuracy falls from 0.811 to 0.797 and MMLU-Pro from 0.502 to 0.469. The package reuses the 4B temperatures and thresholds. vendor reportedsource
Links
Source: Model card and files → Source: 3B package and correction to the 4B release score →
Related projects
No related projects are listed in the registry.