browser-use/bu-30b-a3b-preview
Full fine-tune vs Qwen/Qwen3-VL-30B-A3B-Instruct. Fine-tune: 98% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/4, Norms 193/309, MLP 144/268, Attention 192/300, lm_head 1/1
4 derivatives checked
- bartowski/browser-use_bu-30b-a3b-preview-GGUF: Same weights, converted
- dennny123/BU-30B-A3B-Heretic-GGUF: Full fine-tune
- cyankiwi/bu-30b-a3b-preview-AWQ-4bit: Quantization
- sky-mighty/bu-30b-a3b-preview-quantized: Same weights, converted
1 other repos hold byte-for-byte identical weight files.