Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
DeepSeek-Coder-V2, an open-source Mixture-of-Experts (MoE) code language model, has achieved significant success in coding and math tasks, surpassing GPT4-Turbo and other models like Claude3-Opus and Gemini-1.5 Pro. With 236B parameters, 338 programming languages support, and a 128k context length, this model is available in two sizes, 16B and 236B. It has been continuously pretrained and is now performing exceptionally well in coding, attracting attention for its cost-effectiveness and superior performance compared to GPT-4o.
Beginning to hear reports that the new deepseek code model is better than gpt4o because of gpt4o's inability to follow instructions. Anyone else feel this way?
Beginning to hear reports that deepseek is better than gpt4o because of gpt4o's inability to follow instructions. Anyone else feel this way?
Ppl are curious about the performance of DeepSeek-Coder-V2-Lite on BigCodeBench. We've added its results, along with a few other models, to the leaderboard! https://t.co/EcaiPk7FcZ DeepSeek-Coder-V2-Lite-Instruct is a beast indeed, similar to Magicoder-S-DS-6.7B, but with only… https://t.co/fC2NCmDire https://t.co/byYL02mEp4