mradermacher/Qwen3-1.6B-A0.9B-GGUF

Same weights, converted vs inference-optimization/Qwen3-1.6B-A0.9B. Same weights stored in another dtype (BF16 → F16/F32), plus 30 new tensors.

Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/41, MLP 0/10, Attention 0/40

On the Hub · Downloads on Model Pulse