Mistral 7B

Mistral 7B was released 7.3B parameter model released by Mistral AI under Apache 2.0.

Has 7.3B parameters model size reported as 7.3 billion parameters.

Outperforms Llama 2 13B on all evaluated benchmarks, mistral AI report comparing benchmarks.

Outperforms Llama 1 34B on many benchmarks benchmark comparisons reported by Mistral AI.

Approaches CodeLlama 7B performance on code tasks code benchmark comparisons reported by Mistral AI.