DavidAU/Qwen3-30B-A1.5B-High-Speed
Full fine-tune vs Qwen/Qwen3-30B-A3B-Base. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 193/193, MLP 2613/2613, Attention 192/192, lm_head 1/1
2 derivatives checked
- DavidAU/Qwen3-30B-A1.5B-64K-High-Speed-NEO-Imatrix-MAX-gguf: Quantization
- mradermacher/Qwen3-30B-A1.5B-High-Speed-GGUF: Quantization
11 other repos hold byte-for-byte identical weight files.