alibiserikbay/JevK5
LoRA-style (fewer layers) vs Qwen/Qwen3.5-4B. alibiserikbay/JevK5 matches part of Qwen/Qwen3.5-4B (426 of its 738 tensors; the rest of Qwen/Qwen3.5-4B has no counterpart). On the shared part: Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of in_proj_a, in_proj_b, in_proj_qkv, in_proj_z, k_proj, o_proj, out_proj, q_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, MLP 0/96, Attention 152/224, Norms 0/105
5 derivatives checked
- alibiserikbay/JevK5-GGUF: Not comparable yet
- crh225/plumb-4b: LoRA-style
- Mr-Shmoo/sonda-1.0-4B: LoRA-style
- SirSahOl/JevK5-chat-mlx-16bit: Partial fine-tune
- SirSahOl/JevK5-chat-mlx-4bit: Quantization