Systematic RSA comparison showing untrained CNNs match backpropagation-trained networks at V1 visual cortex, revealing architecture's dominant role over learning rules in neural alignment. Activation triggers: untrained cnn, backpropagation, v1, rsa, representational similarity, learning rules, architecture-driven.
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/unt
| Capabilities (fewer powerful permissions = higher) | 50/50 |
| Maintenance | 21/35 |
| Documentation | 15/15 |
Always review files before installing. How scores work. Request a correction