Commit ba3b8410 authored by jameskrw's avatar jameskrw
Browse files

minor

parent 053cd33c

scripts/exps/format/frozenlake/free_think/run.sh

+3 −3
Original line number Diff line number Diff line
@@ -26,8 +26,8 @@ python3 -m vagen.trainer.main_ppo \ 
    data.val_files=data/$EXPERIMENT_NAME/test.parquet \

    data.train_batch_size=64 \

    data.max_prompt_length=1024 \

    data.max_response_length=128 \

    data.max_trajectory_length=2400 \

    data.max_response_length=150 \

    data.max_trajectory_length=2200 \

    data.image_key=images \

    data.truncation=error \

    actor_rollout_ref.model.path=Qwen/Qwen2.5-VL-3B-Instruct \
@@ -44,7 +44,7 @@ python3 -m vagen.trainer.main_ppo \ 
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=1 \

    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \

    actor_rollout_ref.rollout.name=vllm \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.2 \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.1 \

    actor_rollout_ref.rollout.enable_chunked_prefill=False \

    actor_rollout_ref.rollout.enforce_eager=False \

    actor_rollout_ref.rollout.free_cache_engine=False \

scripts/exps/format/frozenlake/grounding/run.sh

+3 −3
Original line number Diff line number Diff line
@@ -26,8 +26,8 @@ python3 -m vagen.trainer.main_ppo \ 
    data.val_files=data/$EXPERIMENT_NAME/test.parquet \

    data.train_batch_size=64 \

    data.max_prompt_length=1024 \

    data.max_response_length=128 \

    data.max_trajectory_length=2400 \

    data.max_response_length=150 \

    data.max_trajectory_length=2200 \

    data.image_key=images \

    data.truncation=error \

    actor_rollout_ref.model.path=Qwen/Qwen2.5-VL-3B-Instruct \
@@ -44,7 +44,7 @@ python3 -m vagen.trainer.main_ppo \ 
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=1 \

    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \

    actor_rollout_ref.rollout.name=vllm \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.2 \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.1 \

    actor_rollout_ref.rollout.enable_chunked_prefill=False \

    actor_rollout_ref.rollout.enforce_eager=False \

    actor_rollout_ref.rollout.free_cache_engine=False \

scripts/exps/format/frozenlake/no_think/run.sh

+3 −3
Original line number Diff line number Diff line
@@ -26,8 +26,8 @@ python3 -m vagen.trainer.main_ppo \ 
    data.val_files=data/$EXPERIMENT_NAME/test.parquet \

    data.train_batch_size=64 \

    data.max_prompt_length=1024 \

    data.max_response_length=128 \

    data.max_trajectory_length=2400 \

    data.max_response_length=150 \

    data.max_trajectory_length=2200 \

    data.image_key=images \

    data.truncation=error \

    actor_rollout_ref.model.path=Qwen/Qwen2.5-VL-3B-Instruct \
@@ -44,7 +44,7 @@ python3 -m vagen.trainer.main_ppo \ 
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=1 \

    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \

    actor_rollout_ref.rollout.name=vllm \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.2 \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.1 \

    actor_rollout_ref.rollout.enable_chunked_prefill=False \

    actor_rollout_ref.rollout.enforce_eager=False \

    actor_rollout_ref.rollout.free_cache_engine=False \

scripts/exps/format/frozenlake/worldmodeling/run.sh

+3 −3
Original line number Diff line number Diff line
@@ -26,8 +26,8 @@ python3 -m vagen.trainer.main_ppo \ 
    data.val_files=data/$EXPERIMENT_NAME/test.parquet \

    data.train_batch_size=64 \

    data.max_prompt_length=1024 \

    data.max_response_length=128 \

    data.max_trajectory_length=2400 \

    data.max_response_length=150 \

    data.max_trajectory_length=2200 \

    data.image_key=images \

    data.truncation=error \

    actor_rollout_ref.model.path=Qwen/Qwen2.5-VL-3B-Instruct \
@@ -44,7 +44,7 @@ python3 -m vagen.trainer.main_ppo \ 
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=1 \

    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \

    actor_rollout_ref.rollout.name=vllm \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.2 \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.1 \

    actor_rollout_ref.rollout.enable_chunked_prefill=False \

    actor_rollout_ref.rollout.enforce_eager=False \

    actor_rollout_ref.rollout.free_cache_engine=False \

scripts/exps/format/sokoban/free_think/run.sh

+3 −3
Original line number Diff line number Diff line
@@ -26,8 +26,8 @@ python3 -m vagen.trainer.main_ppo \ 
    data.val_files=data/$EXPERIMENT_NAME/test.parquet \

    data.train_batch_size=128 \

    data.max_prompt_length=1024 \

    data.max_response_length=128 \

    data.max_trajectory_length=2400 \

    data.max_response_length=150 \

    data.max_trajectory_length=2200 \

    data.image_key=images \

    data.truncation=error \

    actor_rollout_ref.model.path=Qwen/Qwen2.5-VL-3B-Instruct \
@@ -44,7 +44,7 @@ python3 -m vagen.trainer.main_ppo \ 
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=1 \

    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \

    actor_rollout_ref.rollout.name=vllm \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.2 \

    actor_rollout_ref.rollout.gpu_memory_utilization=0.1 \

    actor_rollout_ref.rollout.enable_chunked_prefill=False \

    actor_rollout_ref.rollout.enforce_eager=False \

    actor_rollout_ref.rollout.free_cache_engine=False \

CRA Git | Maintained and supported by SUSTech CRA and CCSE