DuoNeural/LFM2.5-8B-A1B-Hermes-Agentic-Coder-Abliterated-v2

docs: add Stock LFM delta and 8B industry comparison tables (Llama-3, Gemma, Mistral, Granite, DeepSeek)

docs: update v2 model cards with official EvalPlus (HumanEval+ & MBPP+) benchmark results

feat: release DuoNeural LFM2.5-8B Hermes Agentic Coder Abliterated v2 BF16 weights

initial commit