├── .gitignore ├── README.md ├── aligner ├── scripts │ └── sft-correction.sh ├── template-dataset.json └── training │ ├── __init__.py │ ├── algorithms │ ├── __init__.py │ ├── dpo │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ ├── ppo │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ ├── ppo_lag │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ └── ppo_reward_shaping │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ ├── configs │ ├── __init__.py │ ├── constants.py │ ├── deepspeed_config.py │ ├── ds_eval_config_template.json │ ├── ds_train_config_template.json │ └── fsdp_config.json │ ├── datasets │ ├── __init__.py │ ├── base.py │ ├── preference.py │ ├── prompt_only.py │ ├── raw │ │ ├── __init__.py │ │ ├── alpaca.py │ │ ├── correction.py │ │ ├── empathy.py │ │ ├── firefly.py │ │ ├── hh_rlhf.py │ │ ├── moss.py │ │ └── safe_rlhf.py │ ├── safety_preference.py │ ├── supervised.py │ └── utils.py │ ├── finetune │ ├── __init__.py │ ├── __main__.py │ ├── deepspeed.py │ ├── huggingface.py │ ├── main.py │ └── trainer.py │ ├── logger.py │ ├── models │ ├── __init__.py │ ├── normalizer.py │ ├── pretrained.py │ └── score_model │ │ ├── __init__.py │ │ ├── bloom │ │ ├── __init__.py │ │ └── modeling_bloom.py │ │ ├── gpt2 │ │ ├── __init__.py │ │ └── modeling_gpt2.py │ │ ├── gpt_neo │ │ ├── __init__.py │ │ └── modeling_gpt_neo.py │ │ ├── gpt_neox │ │ ├── __init__.py │ │ └── modeling_gpt_neox.py │ │ ├── gptj │ │ ├── __init__.py │ │ └── modeling_gptj.py │ │ ├── llama │ │ ├── __init__.py │ │ └── modeling_llama.py │ │ └── opt │ │ ├── __init__.py │ │ └── modeling_opt.py │ ├── trainers │ ├── __init__.py │ ├── base.py │ ├── rl_trainer.py │ └── supervised_trainer.py │ ├── utils.py │ ├── values │ ├── __init__.py │ ├── cost │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ └── reward │ │ ├── __init__.py │ │ ├── __main__.py │ │ ├── main.py │ │ └── trainer.py │ └── version.py ├── conda-recipe.yaml ├── images ├── arch-0521.pdf ├── intep_copy.png ├── intep_correction.png ├── intep_representation_control.png ├── main-paradigm.jpg ├── multi_round_pipeline.png ├── performance.png ├── semantic_space.png ├── w2s_illustration.png └── w2s_paradigm.png ├── logo └── pku-ai.png ├── requirements.txt └── results ├── ablation_critique.png ├── ablation_preference_dataset.png ├── distribution_shift.png ├── identity_mapping.png ├── interp.png ├── iterative_performance.png ├── multi_round.png └── performance_all.png /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/.gitignore -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/README.md -------------------------------------------------------------------------------- /aligner/scripts/sft-correction.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/scripts/sft-correction.sh -------------------------------------------------------------------------------- /aligner/template-dataset.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/template-dataset.json -------------------------------------------------------------------------------- /aligner/training/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/dpo/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/dpo/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/dpo/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/dpo/__main__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/dpo/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/dpo/main.py -------------------------------------------------------------------------------- /aligner/training/algorithms/dpo/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/dpo/trainer.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo/__main__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo/main.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo/trainer.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_lag/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_lag/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_lag/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_lag/__main__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_lag/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_lag/main.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_lag/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_lag/trainer.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_reward_shaping/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_reward_shaping/__init__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_reward_shaping/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_reward_shaping/__main__.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_reward_shaping/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_reward_shaping/main.py -------------------------------------------------------------------------------- /aligner/training/algorithms/ppo_reward_shaping/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/algorithms/ppo_reward_shaping/trainer.py -------------------------------------------------------------------------------- /aligner/training/configs/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/__init__.py -------------------------------------------------------------------------------- /aligner/training/configs/constants.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/constants.py -------------------------------------------------------------------------------- /aligner/training/configs/deepspeed_config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/deepspeed_config.py -------------------------------------------------------------------------------- /aligner/training/configs/ds_eval_config_template.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/ds_eval_config_template.json -------------------------------------------------------------------------------- /aligner/training/configs/ds_train_config_template.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/ds_train_config_template.json -------------------------------------------------------------------------------- /aligner/training/configs/fsdp_config.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/configs/fsdp_config.json -------------------------------------------------------------------------------- /aligner/training/datasets/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/__init__.py -------------------------------------------------------------------------------- /aligner/training/datasets/base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/base.py -------------------------------------------------------------------------------- /aligner/training/datasets/preference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/preference.py -------------------------------------------------------------------------------- /aligner/training/datasets/prompt_only.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/prompt_only.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/__init__.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/alpaca.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/alpaca.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/correction.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/correction.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/empathy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/empathy.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/firefly.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/firefly.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/hh_rlhf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/hh_rlhf.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/moss.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/moss.py -------------------------------------------------------------------------------- /aligner/training/datasets/raw/safe_rlhf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/raw/safe_rlhf.py -------------------------------------------------------------------------------- /aligner/training/datasets/safety_preference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/safety_preference.py -------------------------------------------------------------------------------- /aligner/training/datasets/supervised.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/supervised.py -------------------------------------------------------------------------------- /aligner/training/datasets/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/datasets/utils.py -------------------------------------------------------------------------------- /aligner/training/finetune/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/__init__.py -------------------------------------------------------------------------------- /aligner/training/finetune/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/__main__.py -------------------------------------------------------------------------------- /aligner/training/finetune/deepspeed.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/deepspeed.py -------------------------------------------------------------------------------- /aligner/training/finetune/huggingface.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/huggingface.py -------------------------------------------------------------------------------- /aligner/training/finetune/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/main.py -------------------------------------------------------------------------------- /aligner/training/finetune/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/finetune/trainer.py -------------------------------------------------------------------------------- /aligner/training/logger.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/logger.py -------------------------------------------------------------------------------- /aligner/training/models/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/normalizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/normalizer.py -------------------------------------------------------------------------------- /aligner/training/models/pretrained.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/pretrained.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/bloom/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/bloom/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/bloom/modeling_bloom.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/bloom/modeling_bloom.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt2/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt2/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt2/modeling_gpt2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt2/modeling_gpt2.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt_neo/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt_neo/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt_neo/modeling_gpt_neo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt_neo/modeling_gpt_neo.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt_neox/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt_neox/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gpt_neox/modeling_gpt_neox.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gpt_neox/modeling_gpt_neox.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gptj/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gptj/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/gptj/modeling_gptj.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/gptj/modeling_gptj.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/llama/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/llama/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/llama/modeling_llama.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/llama/modeling_llama.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/opt/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/opt/__init__.py -------------------------------------------------------------------------------- /aligner/training/models/score_model/opt/modeling_opt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/models/score_model/opt/modeling_opt.py -------------------------------------------------------------------------------- /aligner/training/trainers/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/trainers/__init__.py -------------------------------------------------------------------------------- /aligner/training/trainers/base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/trainers/base.py -------------------------------------------------------------------------------- /aligner/training/trainers/rl_trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/trainers/rl_trainer.py -------------------------------------------------------------------------------- /aligner/training/trainers/supervised_trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/trainers/supervised_trainer.py -------------------------------------------------------------------------------- /aligner/training/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/utils.py -------------------------------------------------------------------------------- /aligner/training/values/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/__init__.py -------------------------------------------------------------------------------- /aligner/training/values/cost/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/cost/__init__.py -------------------------------------------------------------------------------- /aligner/training/values/cost/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/cost/__main__.py -------------------------------------------------------------------------------- /aligner/training/values/cost/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/cost/main.py -------------------------------------------------------------------------------- /aligner/training/values/cost/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/cost/trainer.py -------------------------------------------------------------------------------- /aligner/training/values/reward/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/reward/__init__.py -------------------------------------------------------------------------------- /aligner/training/values/reward/__main__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/reward/__main__.py -------------------------------------------------------------------------------- /aligner/training/values/reward/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/reward/main.py -------------------------------------------------------------------------------- /aligner/training/values/reward/trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/values/reward/trainer.py -------------------------------------------------------------------------------- /aligner/training/version.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/aligner/training/version.py -------------------------------------------------------------------------------- /conda-recipe.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/conda-recipe.yaml -------------------------------------------------------------------------------- /images/arch-0521.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/arch-0521.pdf -------------------------------------------------------------------------------- /images/intep_copy.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/intep_copy.png -------------------------------------------------------------------------------- /images/intep_correction.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/intep_correction.png -------------------------------------------------------------------------------- /images/intep_representation_control.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/intep_representation_control.png -------------------------------------------------------------------------------- /images/main-paradigm.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/main-paradigm.jpg -------------------------------------------------------------------------------- /images/multi_round_pipeline.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/multi_round_pipeline.png -------------------------------------------------------------------------------- /images/performance.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/performance.png -------------------------------------------------------------------------------- /images/semantic_space.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/semantic_space.png -------------------------------------------------------------------------------- /images/w2s_illustration.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/w2s_illustration.png -------------------------------------------------------------------------------- /images/w2s_paradigm.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/images/w2s_paradigm.png -------------------------------------------------------------------------------- /logo/pku-ai.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/logo/pku-ai.png -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/requirements.txt -------------------------------------------------------------------------------- /results/ablation_critique.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/ablation_critique.png -------------------------------------------------------------------------------- /results/ablation_preference_dataset.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/ablation_preference_dataset.png -------------------------------------------------------------------------------- /results/distribution_shift.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/distribution_shift.png -------------------------------------------------------------------------------- /results/identity_mapping.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/identity_mapping.png -------------------------------------------------------------------------------- /results/interp.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/interp.png -------------------------------------------------------------------------------- /results/iterative_performance.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/iterative_performance.png -------------------------------------------------------------------------------- /results/multi_round.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/multi_round.png -------------------------------------------------------------------------------- /results/performance_all.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/PKU-Alignment/aligner/HEAD/results/performance_all.png --------------------------------------------------------------------------------