Warning: This model is ranked first on the Open LLM Leaderboard among the 7B models as of January 28th, 2024, but note that it was produced from merges without fine-tuning and it was not confirmed whether any of the models were trained on evaluation benchmarks.
Supported Languages
en (proficient)
Training Details
Methodology:
Created with mergekit with the specified configuration: models: mncai/mistral-7b-dpo-v5, FelixChao/WestSeverus-7B-DPO-v2, BarryFutureman/NeuralTurdusVariant1-7B with parameters: density: 0.5, weight: 0.3 and 0.5, merge_method: ties, base_model: mncai/mistral-7b-dpo-v5, parameters: normalize: true, dtype: float16