description Codestral Mamba Overview
Codestral Mamba is an open-weight code generation model released by Mistral AI in 2024. It uses the Mamba architecture, a state space model approach that differs from standard transformer designs. The model is designed for code completion and generation tasks across multiple programming languages. It can be integrated with development environments through the Continue extension.
help Codestral Mamba FAQ
When was Codestral Mamba released?
Mistral AI released Codestral Mamba on July 16, 2024. It was introduced as an open-weight code model under the Apache 2.0 license.
What architecture does Codestral Mamba use?
Codestral Mamba uses the Mamba2 state-space architecture rather than a standard Transformer design. Mistral describes it as a model designed for efficient code generation and long-sequence processing.
How large is Codestral Mamba?
The released model has about 7.285 billion parameters. Mistral also reported testing its in-context retrieval capability with sequences up to 256,000 tokens.
Can Codestral Mamba run locally?
Mistral lists local deployment through the mistral-inference SDK and TensorRT-LLM, with raw weights available through Hugging Face. Hardware requirements depend on quantization, context length, and the inference backend.
explore Explore More
Similar to Codestral Mamba
See all arrow_forward
Reviews & Comments
Write a Review
Be the first to review
Share your thoughts with the community and help others make better decisions.
