├── .gitignore ├── LICENSE ├── README.md ├── assets ├── data_statistics.png ├── method.png ├── question_examples.png └── teaser.png ├── configs ├── zero1_no_optimizer.json ├── zero2.json ├── zero3.json └── zero3_offload.json ├── scripts ├── eval_general_video_bench.sh ├── eval_seed_bench_r1.sh ├── run_grpo_care_margin0.01_seed_bench_r1.sh ├── run_grpo_care_margin0.01_video_r1.sh └── run_grpo_care_margin0_seed_bench_r1.sh ├── setup.sh └── src ├── eval_bench.py ├── qwen-vl-utils ├── .python-version ├── README.md ├── pyproject.toml ├── requirements-dev.lock ├── requirements.lock └── src │ └── qwen_vl_utils │ ├── __init__.py │ └── vision_process.py ├── r1-v ├── .gitignore ├── Evaluation │ └── check_file_mp4.py ├── LICENSE ├── Makefile ├── configs │ ├── ddp.yaml │ ├── qwen2vl_sft_config.yaml │ ├── zero2.yaml │ └── zero3.yaml ├── setup.cfg ├── setup.py └── src │ └── open_r1 │ ├── __init__.py │ ├── grpo.py │ └── trainer │ ├── __init__.py │ ├── ema_trainer.py │ └── grpo_trainer_ref_ema.py └── unzip.py /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/.gitignore -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/README.md -------------------------------------------------------------------------------- /assets/data_statistics.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/assets/data_statistics.png -------------------------------------------------------------------------------- /assets/method.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/assets/method.png -------------------------------------------------------------------------------- /assets/question_examples.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/assets/question_examples.png -------------------------------------------------------------------------------- /assets/teaser.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/assets/teaser.png -------------------------------------------------------------------------------- /configs/zero1_no_optimizer.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/configs/zero1_no_optimizer.json -------------------------------------------------------------------------------- /configs/zero2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/configs/zero2.json -------------------------------------------------------------------------------- /configs/zero3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/configs/zero3.json -------------------------------------------------------------------------------- /configs/zero3_offload.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/configs/zero3_offload.json -------------------------------------------------------------------------------- /scripts/eval_general_video_bench.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/scripts/eval_general_video_bench.sh -------------------------------------------------------------------------------- /scripts/eval_seed_bench_r1.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/scripts/eval_seed_bench_r1.sh -------------------------------------------------------------------------------- /scripts/run_grpo_care_margin0.01_seed_bench_r1.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/scripts/run_grpo_care_margin0.01_seed_bench_r1.sh -------------------------------------------------------------------------------- /scripts/run_grpo_care_margin0.01_video_r1.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/scripts/run_grpo_care_margin0.01_video_r1.sh -------------------------------------------------------------------------------- /scripts/run_grpo_care_margin0_seed_bench_r1.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/scripts/run_grpo_care_margin0_seed_bench_r1.sh -------------------------------------------------------------------------------- /setup.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/setup.sh -------------------------------------------------------------------------------- /src/eval_bench.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/eval_bench.py -------------------------------------------------------------------------------- /src/qwen-vl-utils/.python-version: -------------------------------------------------------------------------------- 1 | 3.8.19 2 | -------------------------------------------------------------------------------- /src/qwen-vl-utils/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/README.md -------------------------------------------------------------------------------- /src/qwen-vl-utils/pyproject.toml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/pyproject.toml -------------------------------------------------------------------------------- /src/qwen-vl-utils/requirements-dev.lock: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/requirements-dev.lock -------------------------------------------------------------------------------- /src/qwen-vl-utils/requirements.lock: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/requirements.lock -------------------------------------------------------------------------------- /src/qwen-vl-utils/src/qwen_vl_utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/src/qwen_vl_utils/__init__.py -------------------------------------------------------------------------------- /src/qwen-vl-utils/src/qwen_vl_utils/vision_process.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/qwen-vl-utils/src/qwen_vl_utils/vision_process.py -------------------------------------------------------------------------------- /src/r1-v/.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/.gitignore -------------------------------------------------------------------------------- /src/r1-v/Evaluation/check_file_mp4.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/Evaluation/check_file_mp4.py -------------------------------------------------------------------------------- /src/r1-v/LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/LICENSE -------------------------------------------------------------------------------- /src/r1-v/Makefile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/Makefile -------------------------------------------------------------------------------- /src/r1-v/configs/ddp.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/configs/ddp.yaml -------------------------------------------------------------------------------- /src/r1-v/configs/qwen2vl_sft_config.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/configs/qwen2vl_sft_config.yaml -------------------------------------------------------------------------------- /src/r1-v/configs/zero2.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/configs/zero2.yaml -------------------------------------------------------------------------------- /src/r1-v/configs/zero3.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/configs/zero3.yaml -------------------------------------------------------------------------------- /src/r1-v/setup.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/setup.cfg -------------------------------------------------------------------------------- /src/r1-v/setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/setup.py -------------------------------------------------------------------------------- /src/r1-v/src/open_r1/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /src/r1-v/src/open_r1/grpo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/src/open_r1/grpo.py -------------------------------------------------------------------------------- /src/r1-v/src/open_r1/trainer/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/src/open_r1/trainer/__init__.py -------------------------------------------------------------------------------- /src/r1-v/src/open_r1/trainer/ema_trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/src/open_r1/trainer/ema_trainer.py -------------------------------------------------------------------------------- /src/r1-v/src/open_r1/trainer/grpo_trainer_ref_ema.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/r1-v/src/open_r1/trainer/grpo_trainer_ref_ema.py -------------------------------------------------------------------------------- /src/unzip.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TencentARC/GRPO-CARE/HEAD/src/unzip.py --------------------------------------------------------------------------------