fblgit/UNA-POLAR-10.7B-InstructMath-v2

The fblgit/UNA-POLAR-10.7B-InstructMath-v2 is a 10.7 billion parameter instruction-tuned language model, built upon the UNA-SOLAR-10.7B-Instruct-1.0 architecture. This model has undergone DPO (Direct Preference Optimization) specifically using the MathPILE Books dataset, enhancing its mathematical reasoning capabilities. It is optimized for tasks requiring strong mathematical understanding and problem-solving. With a context length of 4096 tokens, it is suitable for processing moderately long mathematical queries and instructions.