Loading README.md +6 −0 Changes for README.md: 6 added lines, 0 removed lines. Original line number Diff line number Diff line Loading @@ -17,3 +17,9 @@ git submodule update --init --recursive ``` ## TODO 1. Implement the validation in ray_trainer 2. Transfer the old ray_trainer file from RAGEN to VAGEN - Implement the metric for wandb 3. Transfer the train.py from RAGEN to VAGEN 4. Change recorder to a logging class No newline at end of file vagen/mllm_agent/rollout.py +4 −5 Changes for vagen/mllm_agent/rollout.py: 4 added lines, 5 removed lines. Original line number Diff line number Diff line Loading @@ -20,11 +20,10 @@ from vagen.env.base import EnvFeedback, EnvConfig # TODO an extra class for recorder @dataclass class QwenVLRolloutConifg: window_size=5 max_prompt_length=1024 max_turns=5 n_gpu_per_node=1 # used for multigpu batch balancing # use factory to initialize the list window_size: int = 5 max_prompt_length: int = 1024 max_turns: int = 5 n_gpu_per_node: int = 1 # used for multigpu batch balancing sptk_for_loss_mask: List[str] = field(default_factory=lambda: ['<|box_start|>', '<|box_end|>']) class QwenVLRolloutManger(): Loading Loading
README.md +6 −0 Changes for README.md: 6 added lines, 0 removed lines. Original line number Diff line number Diff line Loading @@ -17,3 +17,9 @@ git submodule update --init --recursive ``` ## TODO 1. Implement the validation in ray_trainer 2. Transfer the old ray_trainer file from RAGEN to VAGEN - Implement the metric for wandb 3. Transfer the train.py from RAGEN to VAGEN 4. Change recorder to a logging class No newline at end of file
vagen/mllm_agent/rollout.py +4 −5 Changes for vagen/mllm_agent/rollout.py: 4 added lines, 5 removed lines. Original line number Diff line number Diff line Loading @@ -20,11 +20,10 @@ from vagen.env.base import EnvFeedback, EnvConfig # TODO an extra class for recorder @dataclass class QwenVLRolloutConifg: window_size=5 max_prompt_length=1024 max_turns=5 n_gpu_per_node=1 # used for multigpu batch balancing # use factory to initialize the list window_size: int = 5 max_prompt_length: int = 1024 max_turns: int = 5 n_gpu_per_node: int = 1 # used for multigpu batch balancing sptk_for_loss_mask: List[str] = field(default_factory=lambda: ['<|box_start|>', '<|box_end|>']) class QwenVLRolloutManger(): Loading
