hotdogs/Agents-A1-4B-kimi-Preview
LoRA-style (fewer layers) vs InternScience/Agents-A1-4B. hotdogs/Agents-A1-4B-kimi-Preview matches part of InternScience/Agents-A1-4B (426 of its 723 tensors; the rest of InternScience/Agents-A1-4B has no counterpart). On the shared part: Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of 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: Attention 32/224, MLP 96/96, Embeddings 0/1, Norms 0/105
2 derivatives checked
- hotdogs/Agents-A1-4B-kimi-Preview-GGUF: Quantization
- hotdogs/Agents-A1-4B-kimi-Preview-heretic: Abliterated