base_model, library_name, license, haffner, quantized_by, tags, language

base_model	library_name	license	haffner	quantized_by	tags	language
vectionlabs/VL-1-Coder	transformers	apache-2.0	readme_rev 1	haffner	heretic reasoning coding agent qwen code local vectionlabs text-generation custom-code fast coder gguf	en it

About

Heretic'ed safetensors and GGUF quants of https://huggingface.co/vectionlabs/VL-1-Coder

Refusals: 2/100, KL divergence: 0.0350

• Parameters:
  • direction_index = 23.30
  • attn.o_proj.max_weight = 1.49
  • attn.o_proj.max_weight_position = 17.90
  • attn.o_proj.min_weight = 1.47
  • attn.o_proj.min_weight_distance = 13.71
  • mlp.down_proj.max_weight = 1.44
  • mlp.down_proj.max_weight_position = 23.74
  • mlp.down_proj.min_weight = 1.09
  • mlp.down_proj.min_weight_distance = 13.05