2025년 5월 14일
Qwen3 Technical Report
(Qwen Team)
In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense and Mixture-of-Expert (MoE) architectures, with parameter scales ranging from 0.6 to 235 billion. A key innovation in Qwen3 is the integration of thinking mode (for complex, multi-step reasoning) and non-thinking mode (for rapid, context-driven responses) into a unified framework. This eliminates the need to switch between different models—–such as chat-optimized models (e.g., GPT-4o) and dedicated reasoning models (e.g., QwQ- 32B)—–and enables dynamic mode switching based on user queries or chat templates. Meanwhile, Qwen3 introduces a thinking budget mechanism, allowing users to allocate computational resources adaptively during inference, thereby balancing latency and performance based on task complexity. Moreover, by leveraging the knowledge from the flagship models, we significantly reduce the computational resources required to build smaller-scale models, while ensuring their highly competitive performance. Empirical evaluations demonstrate that Qwen3 achieves state-of-the-art results across diverse benchmarks, including tasks in code generation, mathematical reasoning, agent tasks, etc., competitive against larger MoE models and proprietary models. Compared to its predecessor Qwen2.5, Qwen3 expands multilingual support from 29 to 119 languages and dialects, enhancing global accessibility through improved cross-lingual understanding and generation capabilities. To facilitate reproducibility and community-driven research and development, all Qwen3 models are publicly accessible under Apache 2.0.
Qwen3 리포트. PDF에서 Trillion 단위로 토큰을 뽑아냈군요. 생성 데이터도 Trillion 규모네요. MoE에는 역시나 Global Load Balancing을 썼군요 (https://arxiv.org/abs/2501.11873).
Qwen3 technical report. They extracted tokens at the trillion level from PDFs. The synthesized data is also at the trillion scale. For MoE, they used global load balancing, as expected (https://arxiv.org/abs/2501.11873).
#llm
AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale
(Yunjie Ji, Xiaoyu Tian, Sitong Zhao, Haotian Wang, Shuaiting Chen, Yiping Peng, Han Zhao, Xiangang Li)
We present AM-Thinking-v1, a 32B dense language model that advances the frontier of reasoning, embodying the collaborative spirit of open-source innovation. Outperforming DeepSeek-R1 and rivaling leading Mixture-of-Experts (MoE) models like Qwen3-235B-A22B and Seed1.5-Thinking, AM-Thinking-v1 achieves impressive scores of 85.3 on AIME 2024, 74.4 on AIME 2025, and 70.3 on LiveCodeBench, showcasing state-of-the-art mathematical and coding capabilities among open-source models of similar scale. Built entirely from the open-source Qwen2.5-32B base model and publicly available queries, AM-Thinking-v1 leverages a meticulously crafted post-training pipeline - combining supervised fine-tuning and reinforcement learning - to deliver exceptional reasoning capabilities. This work demonstrates that the open-source community can achieve high performance at the 32B scale, a practical sweet spot for deployment and fine-tuning. By striking a balance between top-tier performance and real-world usability, we hope AM-Thinking-v1 inspires further collaborative efforts to harness mid-scale models, pushing reasoning boundaries while keeping accessibility at the core of innovation. We have open-sourced our model on Hugging Face.
Qwen 2.5 32B 기반으로 만든 추론 모델이군요. a-m-team이 어떤 회사 소속인가 했는데 중국의 부동산 회사라고 하네요 (Beike).
A reasoning model based on Qwen 2.5 32B. I was curious about the affiliation of a-m-team, and it turns out they're associated with a Chinese real estate company (Beike).
#rl #reasoning
Aya Vision: Advancing the Frontier of Multilingual Multimodality
(Saurabh Dash, Yiyang Nan, John Dang, Arash Ahmadian, Shivalika Singh, Madeline Smith, Bharat Venkitesh, Vlad Shmyhlo, Viraat Aryabumi, Walter Beller-Morales, Jeremy Pekmez, Jason Ozuzu, Pierre Richemond, Acyr Locatelli, Nick Frosst, Phil Blunsom, Aidan Gomez, Ivan Zhang, Marzieh Fadaee, Manoj Govindassamy, Sudip Roy, Matthias Gallé, Beyza Ermis, Ahmet Üstün, Sara Hooker)
Building multimodal language models is fundamentally challenging: it requires aligning vision and language modalities, curating high-quality instruction data, and avoiding the degradation of existing text-only capabilities once vision is introduced. These difficulties are further magnified in the multilingual setting, where the need for multimodal data in different languages exacerbates existing data scarcity, machine translation often distorts meaning, and catastrophic forgetting is more pronounced. To address the aforementioned challenges, we introduce novel techniques spanning both data and modeling. First, we develop a synthetic annotation framework that curates high-quality, diverse multilingual multimodal instruction data, enabling Aya Vision models to produce natural, human-preferred responses to multimodal inputs across many languages. Complementing this, we propose a cross-modal model merging technique that mitigates catastrophic forgetting, effectively preserving text-only capabilities while simultaneously enhancing multimodal generative performance. Aya-Vision-8B achieves best-in-class performance compared to strong multimodal models such as Qwen-2.5-VL-7B, Pixtral-12B, and even much larger Llama-3.2-90B-Vision. We further scale this approach with Aya-Vision-32B, which outperforms models more than twice its size, such as Molmo-72B and LLaMA-3.2-90B-Vision. Our work advances multilingual progress on the multi-modal frontier, and provides insights into techniques that effectively bend the need for compute while delivering extremely high performance.
Cohere의 공개 VLM. 다국어에 대한 대응이나 멀티모달 데이터의 학습 비율을 조정하는 대신 모델 병합으로 처리하는 부분 등 재미있는 지점들이 있네요.
Cohere's open-source VLM. There are interesting aspects, such as their approach to multilingual support and their use of model merging instead of adjusting the training ratio of multimodal data.
#multimodal #pretraining
DeepMath-Creative: A Benchmark for Evaluating Mathematical Creativity of Large Language Models
(Xiaoyang Chen, Xinan Dai, Yu Du, Qian Feng, Naixu Guo, Tingshuo Gu, Yuting Gao, Yingyi Gao, Xudong Han, Xiang Jiang, Yilin Jin, Hongyi Lin, Shisheng Lin, Xiangnan Li, Yuante Li, Yixing Li, Zhentao Lai, Zilu Ma, Yingrong Peng, Jiacheng Qian, Hao-Yu Sun, Jianbo Sun, Zirui Wang, Siwei Wu, Zian Wang, Bin Xu, Jianghao Xu, Yiyang Yu, Zichuan Yang, Hongji Zha, Ruichong Zhang)
To advance the mathematical proficiency of large language models (LLMs), the DeepMath team has launched an open-source initiative aimed at developing an open mathematical LLM and systematically evaluating its mathematical creativity. This paper represents the initial contribution of this initiative. While recent developments in mathematical LLMs have predominantly emphasized reasoning skills, as evidenced by benchmarks on elementary to undergraduate-level mathematical tasks, the creative capabilities of these models have received comparatively little attention, and evaluation datasets remain scarce. To address this gap, we propose an evaluation criteria for mathematical creativity and introduce DeepMath-Creative, a novel, high-quality benchmark comprising constructive problems across algebra, geometry, analysis, and other domains. We conduct a systematic evaluation of mainstream LLMs' creative problem-solving abilities using this dataset. Experimental results show that even under lenient scoring criteria -- emphasizing core solution components and disregarding minor inaccuracies, such as small logical gaps, incomplete justifications, or redundant explanations -- the best-performing model, O3 Mini, achieves merely 70% accuracy, primarily on basic undergraduate-level constructive tasks. Performance declines sharply on more complex problems, with models failing to provide substantive strategies for open problems. These findings suggest that, although current LLMs display a degree of constructive proficiency on familiar and lower-difficulty problems, such performance is likely attributable to the recombination of memorized patterns rather than authentic creative insight or novel synthesis.
수학적 대상을 만들어내는 능력을 통해 수학적 창조성을 검증하려고 하는 벤치마크. 구체적으로는 명제의 사례나 반례를 만들어내라고 요구하는 형태네요.
This benchmark aims to evaluate mathematical creativity through the ability to create mathematical objects. Specifically, it requires models to generate examples or counterexamples for given propositions.
#benchmark #math