# Aquila2 Technical Report

arXiv: https://arxiv.org/abs/2408.07410 · HTML: https://mlsanity.com/paper/2408.07410/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Aug 2024.

## Built on by (most cited first, 5 of 5)

- [Emu3: Next-Token Prediction is All You Need](https://mlsanity.com/paper/2409.18869/) (2024)
- [Infinity Instruct: Scaling Instruction Selection and Synthesis to Enhance Language Models](https://mlsanity.com/paper/2506.11116/) (2025)
- [Multimodal Large Language Models for Medicine: A Comprehensive Survey](https://mlsanity.com/paper/2504.21051/) (2025)
- [HalluEntity: Benchmarking and Understanding Entity-Level Hallucination Detection](https://mlsanity.com/paper/2502.11948/) (2025)
- [Optimization-Inspired Few-Shot Adaptation for Large Language Models](https://mlsanity.com/paper/2505.19107/) (2025)

## Abstract (arXiv)

This paper introduces the Aquila2 series, which comprises a wide range of bilingual models with parameter sizes of 7, 34, and 70 billion. These models are trained based on an innovative framework named HeuriMentor (HM), which offers real-time insights into model convergence and enhances the training process and data management. The HM System, comprising the Adaptive Training Engine (ATE), Training State Monitor (TSM), and Data Management Unit (DMU), allows for precise monitoring of the model's training progress and enables efficient optimization of data distribution, thereby enhancing training effectiveness. Extensive evaluations show that the Aquila2 model series performs comparably well on both English and Chinese benchmarks. Specifically, Aquila2-34B demonstrates only a slight decrease in performance when quantized to Int4. Furthermore, we have made our training code (https://github.com/FlagOpen/FlagScale) and model weights (https://github.com/FlagAI-Open/Aquila2) publicly available to support ongoing research and the development of applications.