NandemoGHS/Anime-Speech-Japanese-Captioner
LoRA-style vs Qwen/Qwen3-Omni-30B-A3B-Captioner. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/439, MLP 1689/1989, Attention 192/556, other 0/11, lm_head 0/1
2 derivatives checked
- mradermacher/Anime-Speech-Japanese-Captioner-GGUF: Quantization
- mradermacher/Anime-Speech-Japanese-Captioner-i1-GGUF: Quantization