Ateron/Gemma-4-Novelist-Eclipse-31B
LoRA-style vs BirdToast/Gemma-4-31B-glimmer-rp-v0.1. 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 60 of 60 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/523, other 0/62, MLP 180/261, Attention 230/338
6 derivatives checked
- bartowski/Ateron_Gemma-4-Novelist-Eclipse-31B-GGUF: Same weights, converted (fewer layers)
- mradermacher/Gemma-4-Novelist-Eclipse-31B-i1-GGUF: Quantization (fewer layers)
- mradermacher/Gemma-4-Novelist-Eclipse-31B-GGUF: Quantization (fewer layers)
- ailexleon/Gemma-4-Novelist-Eclipse-31B-mlx-4Bit: Quantization
- ailexleon/Gemma-4-Novelist-Eclipse-31B-mlx-8Bit: Quantization
- Cyclone-Labs/Twisted-Cyclone-31B: LoRA-style