#reasoning-benchmarks

[ follow ]
#deepseek #ai-models #large-language-models #machine-learning #open-source-ai #sparse-attention-dsa
Artificial intelligence
fromTechCrunch
2 weeks ago

DeepSeek previews new AI model that 'closes the gap' with frontier models | TechCrunch

DeepSeek launched V4 models, featuring 1 million token context windows and significant parameter counts, outperforming many peers in reasoning benchmarks.
fromInfoQ
4 months ago

DeepSeek-V3.2 Outperforms GPT-5 on Reasoning Tasks

DeepSeek applied three new techniques in the development of DeepSeek-V3.2. First, they used a more efficient attention mechanism called DeepSeek Sparse Attention (DSA) that reduces the computational complexity of the model. They also scaled the reinforcement learning phase, which consumed more compute budget than did pre-training. Finally, they developed an agentic task synthesis pipeline to improve the models' tool use.
Artificial intelligence
[ Load more ]