Aratako/Qwen3-30B-A3B-NSFW-JP
Full fine-tune vs Qwen/Qwen3-30B-A3B-Base. Fine-tune: 94% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 0/1, Norms 191/193, MLP 2613/2613, Attention 192/192, lm_head 1/1
4 derivatives checked
- mmnga/Aratako-Qwen3-30B-A3B-NSFW-JP-gguf: Quantization
- mradermacher/Qwen3-30B-A3B-NSFW-JP-i1-GGUF: Quantization
- Aratako/Qwen3-30B-A3B-ERP-v0.1: Full fine-tune
- mradermacher/Qwen3-30B-A3B-NSFW-JP-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.