DuoNeural/LFM2.5-8B-A1B-Abliterated-GGUF

Add lfm25-8b-abliterated-Q8_0.gguf

Add lfm25-8b-abliterated-Q5_K_M.gguf

Add lfm25-8b-abliterated-Q4_K_M.gguf

Add lfm25-8b-abliterated-Q3_K_L.gguf

Fix: Heretic v2.0 → v1.2.0

Fix GTX 1070 age: 10 years old (2016), not 6

Add KL divergence (Heretic v2.0): 1.06e-7 nats benign prompts

Add speed benchmarks: GTX 1070 63.88 t/s Q4 (full 128k), RTX 3090 388 t/s

Add lfm25-8b-abliterated-q4_k_m.gguf

Add lfm25-8b-abliterated-q5_k_m.gguf

Add model card

initial commit