unsloth/Qwen3.5-4B-Base

Same weights, converted vs Qwen/Qwen3.5-4B-Base. Same weights stored in another dtype (BF16/F32 → BF16).

Tensors changed: Embeddings 0/4, MLP 0/196, Attention 0/320, Norms 0/203

2 derivatives checked

• yaoandy107/greyscope-v2-qwen3.5-4b: LoRA-style (fewer layers)
• Kadabra/Qwen3.5-4B-Gascon-CPT: Full fine-tune

On the Hub · Downloads on Model Pulse