Oct 27 15 days ago

1. 04:12

   Rename to Arbiter-8B; state the Qwen3-8B base clearly

   jsemrau
   Verified signature

   cb5578c
2. 03:10

   Arbiter v0.4 full model: LoRA merged into the dequantized 4-bit base (fp16), tokenizer, decision head, calibration; adapter kept in adapter/

   jsemrau
   Verified signature

   aa20e73
3. 02:34

   Arbiter v0.4: LoRA adapter, decision head, calibration, inference code and model card

   jsemrau
   Verified signature

   53ab3d4