---
base_model:
- CohereLabs/North-Micro-Vision-Instruct
---

Doesnt work with llama.cpp convert_hf_to_gguf.py yet, but should when support comes

Refusals:  3/100, KL divergence: 0.0407  
* direction_index = 16.33
  * attn.o_proj.max_weight = 1.39
  * attn.o_proj.max_weight_position = 16.52
  * attn.o_proj.min_weight = 1.25
  * attn.o_proj.min_weight_distance = 16.04
  * mlp.down_proj.max_weight = 1.14
  * mlp.down_proj.max_weight_position = 16.53
  * mlp.down_proj.min_weight = 0.68
  * mlp.down_proj.min_weight_distance = 10.29