SL-AI/GRaPE-2.5-Quasar
LoRA-style vs Qwen/Qwen3.8-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, in_proj_qkv, in_proj_z, k_proj, o_proj, out_proj, q_proj, up_proj, v_proj changed, in 41 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/325, Attention 131/560, MLP 120/307, Embeddings 0/5, lm_head 0/1, other 0/1
4 derivatives checked
- mradermacher/GRaPE-2.5-Quasar-i1-GGUF: Quantization
- mradermacher/GRaPE-2.5-Quasar-GGUF: Quantization
- SL-AI/GRaPE-2.5-Quasar-GGUF: Same weights, converted
- DedeProGames/GRaPE-2.5-Quasar-Q3_K_M-GGUF: Quantization