# High-precision Reasoning

ReasonFlux-F1-32B is a hierarchical large language model based on template-enhanced reasoning trajectory fine-tuning, excelling in reasoning tasks.

Yi-1.5 is an upgraded version of the Yi model, demonstrating superior performance in programming, mathematics, reasoning, and instruction-following capabilities

Zephyr 141B-A39B is a large language model fine-tuned from Mixtral-8x22B-v0.1, trained using the ORPO alignment algorithm, designed to be a helpful assistant.

BoreanGale-70B is a 70B-parameter large language model merged using a custom algorithm (NearSwap), combining the miqu-1-70b-sf and WinterGoddess-1.4x-70B-L2 models.

StrangeMerges_17-7B-dare_ties is the result of merging two models, Gille/StrangeMerges_16-7B-slerp and Gille/StrangeMerges_12-7B-slerp, using the dare_ties merging method via LazyMergekit.

TheBeagle is a 7-billion-parameter model trained on The Bagel dataset, optimized with DPO (Direct Preference Optimization) and UNA (Unified Neural Architecture) techniques, demonstrating excellent performance in multi-task scenarios.