Tracked topic
Qwen3 is an AI model family developed by Alibaba, released as a set of large language models for natural-language tasks.
…Experiments on IA-Bench, Mindbench and WISE-Verified show that Qwen-Image-Agent outperforms strong baselines and achieves state-of-the-art performance. View arXiv page View PDF Add to collection Community…
…models through specialized reinforcement learning teachers and on-policy distillation, achieving superior performance over existing methods. Generated by Qwen/Qwen2.5-Coder-32B-Instruct Modern large language models (LLMs) rely on reinforcement…
…Under FP8-quantized rollout, MIPU achieves both improved reasoning performance and significantly enhanced training stability across Qwen3-1.7B and Qwen3-4B. The figure summarizes the core idea: instead of optimizing only…
…Generated by Qwen/Qwen2.5-Coder-32B-Instruct We present BioMatrix, the first multimodal foundation model that natively integrates sequences, structures, and natural language for both molecules and proteins within a single…
…A 0.6B Qwen3 interpreter executing PAW programs matches the performance of direct prompting of Qwen3-32B, while using roughly one fiftieth of the inference memory and running at 30 tokens/s…
…transfer knowledge through aligned KV-cache communication, achieving better performance than text-based methods with reduced computational costs. Generated by Qwen/Qwen2.5-Coder-32B-Instruct Multi-agent systems communicate mostly through…
…representations, offering a faster alternative to sparse autoencoder training while maintaining competitive performance in probing tasks. Generated by Qwen/Qwen2.5-Coder-32B-Instruct Finding interpretable directions in language-model representations is…
…Across FastV, PDrop, and Nüwa variants on LLaVA-1.5 and Qwen backbones, reroute improves grounding under aggressive token reduction while maintaining general VQA performance. These results suggest that VLM token reduction…
…Generated by Qwen/Qwen2.5-Coder-32B-Instruct In long-context use, large language models frequently synthesize answers from the meaning of a relevant context span rather than literally copy-pasting them…
…through expert scoring, selection, grouping, and knowledge distillation, achieving superior performance and efficiency compared to traditional pruning methods. Generated by Qwen/Qwen2.5-Coder-32B-Instruct Mixture-of-Experts (MoE) is now…