Marimonald/Ojisan-translator-v1-Qwen-3.5-9B
LoRA-style (fewer layers) vs unsloth/Qwen3.5-9B. Marimonald/Ojisan-translator-v1-Qwen-3.5-9B matches part of unsloth/Qwen3.5-9B (427 of its 775 tensors; the rest of unsloth/Qwen3.5-9B has no counterpart). On the shared part: 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 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, MLP 96/96, Attention 32/224, Norms 0/105
1 derivatives checked
- mradermacher/Ojisan-translator-v1-Qwen-3.5-9B-GGUF: Same weights, converted