Introducing Palmyra-mini family: Reasoning models (1.5B-1.7B)

AI Impact Summary

WRITER has released three new open models, the Palmyra-mini family, each designed for efficient inference with sizes ranging from 1.5B to 1.7B parameters. These models incorporate a Chain of Thought (CoT) training approach, significantly improving their reasoning capabilities, particularly in complex logical challenges and demanding tasks. The models are available in GGUF and MLX quantizations, offering flexibility for various deployment scenarios.