Qwen3-DNA-Lite Collection Patched versions of Qwen3 with identical weights, featuring a separated chat_template.jinja and {% generation %} tags for seamless trl assistant_only. • 8 items • Updated 11 days ago
Qwen3-DNA-Lite Collection Patched versions of Qwen3 with identical weights, featuring a separated chat_template.jinja and {% generation %} tags for seamless trl assistant_only. • 8 items • Updated 11 days ago
DNA 2.1 Collection Making Qwen3 Think in Korean with Reinforcement Learning https://arxiv.org/abs/2508.10355 • 2 items • Updated 11 days ago
Smoothie-Qwen: Post-Hoc Smoothing to Reduce Language Bias in Multilingual LLMs Paper • 2507.05686 • Published Jul 8, 2025 • 1
DNA 1.0 Collection 8B Korean SoTA model, which is instruction-tuned by Dnotitia Inc. • 3 items • Updated 11 days ago • 1