French AI startup Mistral AI has also joined the competition.
Just now, Mistral AI announced the next generation of its flagship open source model: Mistral Large 2, which has 123 billion parameters and is on par with the latest cutting-edge models of OpenAI and Meta in terms of code generation, mathematics, reasoning, etc.
Following the Llama 3.1 405B, the release of Mistral Large 2 suddenly made the track of the open source model lively, and this model is characterized by being “big enough.”
Specifically, although the Mistral Large 2 parameter quantity is lower than Llama 3.1 ‘s 405 billion, the performance of the two is similar. It is also comparable to GPT-4o and Anthropic’s Claude 3.5 Sonnet in multiple benchmarks.
In February, Mistral AI launched its original Large model, with a context window containing 32,000 tokens. The new model builds on this and has a larger 128,000 context window (roughly the equivalent of a 300-page book)-matching OpenAI’s GPT-4o and GPT-4o mini and Meta’s Llama 3.1.
Currently, Mistral Large 2 supports dozens of languages, including French, German, Spanish, Italian, Portugal, Arabic, Hindi, Russian, Chinese, Japanese and Korean, as well as more than 80 programming languages, including Python, Java, C, C++, JavaScript and Bash.
Mistral AI pointed out that the new model will continue to push the boundaries of cost efficiency, speed and performance, while providing users with new capabilities, including advanced function calls and retrieval, to build high-performance AI applications.
Evaluation results:
Performance/Cost: Mistral Large 2 sets new standards in terms of tradeoffs between performance and cost. In MMLU (Multitask Language Understanding) evaluations, the pre-trained version achieved an accuracy rate of 84.0%, occupying the performance/cost forefront among open source models.
Code and reasoning capabilities
Code generation benchmarks: Mistral Large 2 performs well in code generation benchmarks, comparable to leading models such as GPT-4o, Claude 3 Opus, and Llama 3 405B.
Reasoning ability: The model has significantly improved its reasoning ability. Through careful fine-tuning, the “illusion” phenomenon of generating inaccurate or irrelevant information is reduced. The model’s performance on mathematical benchmarks demonstrates its enhanced reasoning and problem solving capabilities.
If you want to learn more, you can click on the link below the video.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Details:https://mistral.ai/news/mistral-large-2407/
Model download:https://huggingface.co/mistralai/Mistral-Large-Instruct-2407
Oil tubing: