maiush/OpenCharacterTraining: Open Character Training ·
The main requirements for installation are Python >= 3.10 and a CUDA-enabled GPU. Please install torch on your system and proceed: We use this implementation to character train the following models: Each model is fine-tuned using 11 constitutions (constitutions/few-shot/) See our paper for further details. All LoRA adapters are available at our HuggingFace collection, with corresponding training data. Constitutions (constitutions/hand-written/) DPO (character/distillation/): SFT (character/introspection/): This project is licensed under the MIT License - see the LICENSE file for details. This work was supported by the ML Alignment & Theory Scholars (MATS) program and the UKRI Centre for Doctoral Training in Application of Artificial Intelligence to the study of Environmental Risks (AI4ER) [EP/S022961/1]. For any queries or information, contact Sharan Maiya. Open Character Training