├── .gitignore ├── README.md ├── assets ├── Math_Results.png ├── SciInstruct.png ├── Sci_Results.png └── models_sci.png ├── evaluation └── SciEval │ ├── datasets │ ├── bai-scieval-test.json │ └── bai-scieval-valid.json │ ├── evaluate_SciEval.py │ └── models.py ├── inference └── cli_demo.py ├── requirements.txt ├── self_reflective_annotation ├── data_classifier │ ├── README.md │ ├── data │ │ ├── rm_test.json │ │ ├── rm_train.json │ │ └── rm_valid.json │ ├── get_rm_rank.py │ ├── rm_eval.py │ └── train_rm.py ├── gpt_labeling │ └── gpt_judge.py └── reflective_generation │ ├── README.md │ ├── api_gen_gpt.py │ ├── api_gen_gpt_revise.py │ └── prompt.py └── training ├── README.md ├── arguments.py ├── deepspeed.json ├── deepspeed3.json ├── finetune.sh ├── finetune_ds3.sh ├── main.py ├── trainer.py └── trainer_seq2seq.py /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/.gitignore -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/README.md -------------------------------------------------------------------------------- /assets/Math_Results.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/assets/Math_Results.png -------------------------------------------------------------------------------- /assets/SciInstruct.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/assets/SciInstruct.png -------------------------------------------------------------------------------- /assets/Sci_Results.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/assets/Sci_Results.png -------------------------------------------------------------------------------- /assets/models_sci.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/assets/models_sci.png -------------------------------------------------------------------------------- /evaluation/SciEval/datasets/bai-scieval-test.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/evaluation/SciEval/datasets/bai-scieval-test.json -------------------------------------------------------------------------------- /evaluation/SciEval/datasets/bai-scieval-valid.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/evaluation/SciEval/datasets/bai-scieval-valid.json -------------------------------------------------------------------------------- /evaluation/SciEval/evaluate_SciEval.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/evaluation/SciEval/evaluate_SciEval.py -------------------------------------------------------------------------------- /evaluation/SciEval/models.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/evaluation/SciEval/models.py -------------------------------------------------------------------------------- /inference/cli_demo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/inference/cli_demo.py -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/requirements.txt -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/README.md -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/data/rm_test.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/data/rm_test.json -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/data/rm_train.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/data/rm_train.json -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/data/rm_valid.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/data/rm_valid.json -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/get_rm_rank.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/get_rm_rank.py -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/rm_eval.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/rm_eval.py -------------------------------------------------------------------------------- /self_reflective_annotation/data_classifier/train_rm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/data_classifier/train_rm.py -------------------------------------------------------------------------------- /self_reflective_annotation/gpt_labeling/gpt_judge.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/gpt_labeling/gpt_judge.py -------------------------------------------------------------------------------- /self_reflective_annotation/reflective_generation/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/reflective_generation/README.md -------------------------------------------------------------------------------- /self_reflective_annotation/reflective_generation/api_gen_gpt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/reflective_generation/api_gen_gpt.py -------------------------------------------------------------------------------- /self_reflective_annotation/reflective_generation/api_gen_gpt_revise.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/reflective_generation/api_gen_gpt_revise.py -------------------------------------------------------------------------------- /self_reflective_annotation/reflective_generation/prompt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/self_reflective_annotation/reflective_generation/prompt.py -------------------------------------------------------------------------------- /training/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/README.md -------------------------------------------------------------------------------- /training/arguments.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/arguments.py -------------------------------------------------------------------------------- /training/deepspeed.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/deepspeed.json -------------------------------------------------------------------------------- /training/deepspeed3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/deepspeed3.json -------------------------------------------------------------------------------- /training/finetune.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/finetune.sh -------------------------------------------------------------------------------- /training/finetune_ds3.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/finetune_ds3.sh -------------------------------------------------------------------------------- /training/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/main.py -------------------------------------------------------------------------------- /training/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/trainer.py -------------------------------------------------------------------------------- /training/trainer_seq2seq.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/THUDM/SciGLM/HEAD/training/trainer_seq2seq.py --------------------------------------------------------------------------------