Mistral’s latest Codestral Mamba for enhancing extensive code generation

The organization conducted trials on Codestral Mamba regarding in-situation retrieval capabilities up to 256k tokens — double the quantity observed in OpenAI’s GPT4o — and discovered that its 7B variant outperformed open source models in various benchmarking assessments, including HumanEval, MBPP, Spider, and CruxE.

The enhanced 22B parameter version of the recent model also exhibited notably superior performance compared to CodeLlama-34B, with the only exception being the CruxE benchmark.

Although the 7B variant is accessible under the Apache 2.0 license, the expanded 22B variant is offered under a commercial license for self-deployment or a community license for testing objectives.

About Author

AndyC

Andy Curtis is an award-winning security consultant, researcher and public speaker. He has been working in the computer security industry since the early 1990s, having been employed by state and federal government, leading healthcare and banking providers across three continents. He has given talks about computer security for some of the world’s largest companies, worked with law enforcement agencies on investigations into hacking groups, and is a regular voice on TV and radio explaining IT security threats.

See author's posts