Base Model (LFM3-1.7B) · GitHub Repository · Step Traces · Inference LFM3-1.7B-RD is an adaptive recurrent-depth language model built by refactoring the 24-layer tri-hybrid architecture of LiquidAI/LFM3-1.7B into a Prelude -> Shared Recurrent Core -> Coda system. Instead of executing a fixed forward pass on every token, LFM3-1.7B-RD executes anywhere from 1 to 32 recurrent core steps per token—either via manual effort presets (low, fast, medium, high, max) or dynamically per token in auto mode via a learned continuous depth controller. > [!NOTE] > This checkpoint is an experimental architectural conversion and continual alignment of LiquidAI/LFM3-1.7B. It focuses on dynamic per-token recurrent routing, early-exit FFN completion, and deep iterative refinement without catastrophic drift. The primary objective of LFM3-1.7B-RD is dynamic per-token compute allocation during autoregressive decoding (reasoningmode=»auto»`). Rather than using a uniform number of layers for every token, the router exits early on predictable syntax and allocates deep recurrent passes to high-entropy branching points and mathematical deductions. Below are actual step traces recorded from the checkpoint…
Модальности:
Генерация текста
Области применения:
Логика и рассуждение Диалог / чат
Задача: Генерация текста
Автор: cybertruck32489
Теги: lfm3_rd, liquid, lfm3, recurrent-depth, adaptive-computation, reasoning, custom_code, conversational
Лайков: 3 | Загрузок: 3,456
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.