arXiv cs.CL· Mario Prignano, Gabriele Cirillo, Alessio Morosini, Luca Cerovaz, Alessandro Bartolocci, Emanuele Rodol\`a, Giulio Starace, Francesco Pappone·· 9 小时前AI 评分58
Limite 1B - Violetto 技术报告:10 亿参数数学推理模型开源发布
Limite 1B - Violetto: Technical Report
AI 导读
Limite 1B - Violetto 是一个面向高吞吐数学推理的 10 亿参数模型,从零在 2370 亿 token 的数学语料上训练,支持最长 131k token 上下文。
正文 · 原文
Abstract:We present Limite 1B - Violetto, a one-billion-parameter model designed for high-throughput mathematical reasoning. Limite is trained from scratch on 237 billion tokens from a carefully curated, mathematics-centric corpus and supports context lengths of up to 131k tokens. Its architecture and training recipe were developed together to maximize reasoning capability while keeping training and inference costs low. Post-training combines knowledge distillation, targeted supervised fine-tuning, and RLVR on a curated curriculum. Limite achieves 94.01% on AIME 2026 and 74.25% on BeyondAIME, placing it among models many times its size on our evaluation suite. The model is deliberately only lightly instruction-tuned: it is intended primarily as a single-turn mathematical solver and as a fast, parallel component of larger systems, rather than as a general-purpose assistant. We release the model, its base and value checkpoints, and an optimized serving implementation under the Apache 2.0 license.
| Comments: | 36 pages. Model, checkpoints and serving code: this https URL |
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2611.21646 [cs.CL] |
| (or arXiv:2611.21646v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.21646 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mario Prignano [view email]
[v1]
Wed, 14 Oct 2026 13:49:22 UTC (1,430 KB)
来源:arXiv cs.CL · arxiv.org