flammenai/flammen15-gutenberg-DPO-v1-7B
LoRA-style vs flammenai/flammen15-mistral-7B. 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, Norms 0/65, MLP 96/96, Attention 128/128, Embeddings 0/1
5 derivatives checked
- mradermacher/flammen15-gutenberg-DPO-v1-7B-GGUF: Quantization
- nbeerbower/HeroBophades-3x7B: MoE built from dense models
- flammenai/flammen15X-mistral-7B: LoRA-style
- nbeerbower/SuperFlammen-4x7B: MoE built from dense models
- nbeerbower/Flammen-Bophades-7B: Full fine-tune