Staged VLM post-training methodology - decomposing VLM capabilities into visual perception, visual reasoning, and textual reasoning stages with specialized data and RL-based perception learning
no risky patterns detected
Type: Agent skill (SKILL.md) · Category: Science, Health & Education · Source: https://github.com/hiyenwong/ai_collection/tree/HEAD/collection/skills/ai-ml/sta
| Capabilities (fewer powerful permissions = higher) | 50/50 |
| Maintenance | 21/35 |
| Documentation | 15/15 |
Always review files before installing. How scores work. Request a correction