MuXodious/Nemotron-Cascade-14B-Thinking-impotent-heresy
TEXT GENERATIONConcurrency Cost:1Model Size:14BQuant:FP8Ctx Length:32kPublished:Jan 18, 2026License:nvidia-open-model-licenseArchitecture:Transformer Open Weights Warm

MuXodious/Nemotron-Cascade-14B-Thinking-impotent-heresy is a 14 billion parameter Nemotron-Cascade-14B-Thinking fine-tune, based on Qwen3-14B Base, developed by MuXodious using P-E-W's Heretic engine. This model is exclusively designed for the 'thinking' mode, leveraging sequential and domain-wise reinforcement learning to achieve strong performance in reasoning tasks. It demonstrates competitive results across various benchmarks, particularly excelling in code-related challenges like LiveCodeBench, and supports a 32768 token context length with YaRN scaling for extended contexts.

Loading preview...