vllm-sr/Decision-2.0-Nox-4B

Different architecture vs Qwen/Qwen3.5-4B-Base. Different architecture: only 0 of 738/10 tensors match by name and shape.

2 derivatives checked

• vllm-sr/Decision-2.0-Nox-4B-GGUF: Same weights, repacked (fewer layers)
• vllm-sr/Vela-2.0-4B: Full fine-tune (fewer layers)

On the Hub · Downloads on Model Pulse