This repository contains a surgically de-censored version of HuggingFaceTB’s SmolLM3 3B Instruct model, optimized via weight abliteration techniques. By applying the heretic framework across an extensive 1000-trial search space, we successfully isolated and neutralized the primary refusal vectors embedded within the attn.oproj and mlp.downproj layers. — Optimal Balance: Selected Trial 317 out of 1000 iterations for the perfect trade-off between freedom and reasoning capabilities. — Refusal Rate: Dropped down to 4/100 (from the original near-total refusal on safety benchmarks). — KL Divergence: 0.0491 — Demonstrates that general language capabilities are preserved relative to the original model. However, safety-aligned weights in attn.oproj and mlp.downproj have been surgically removed; this is intentional modification, not unintended degradation. Benchmark evaluations will be added soon once evaluation pipeline is completed. Unlike aggressive fine-tuning which often degrades the model’s core logic or shifts its grammar distributions, this weight manipulation directly zeros out the activation steering directions that cause alignment blocks. Retained Capabilities: Excellent at…
Модальности:
Генерация текста
Области применения:
Диалог / чат Следование инструкциям
Задача: Генерация текста
Автор: K0D3IN
Теги: smollm3, smollm, heretic, abliteration, apache2, conversational, en, endpoints_compatible
Лайков: 3 | Загрузок: 481
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
