follow the official code,got the error:because name 'global_poolverl_group_2:0' already exists

[ArlanCooper]

i ran the code as readme,

# 设置参数
export CUDA_VISIBLE_DEVICES=2,3
export N_GPUS=2
export BASE_MODEL=/home/octopus/data/llm_list/Qwen2.5-3B
export DATA_DIR=/home/powerop/work/rwq/myoperate/TinyZero/data_sets
export ROLLOUT_TP_SIZE=2
export EXPERIMENT_NAME=countdown-qwen2.5-3b
export VLLM_ATTENTION_BACKEND=XFORMERS

bash ./scripts/train_tiny_zero.sh

i got this error:

Traceback (most recent call last):
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/main_ppo.py", line 103, in main
    ray.get(main_task.remote(config))
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/auto_init_hook.py", line 21, in auto_init_wrapper
    return fn(*args, **kwargs)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/client_mode_hook.py", line 103, in wrapper
    return func(*args, **kwargs)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/worker.py", line 2755, in get
    values, debugger_breakpoint = worker.get_objects(object_refs, timeout=timeout)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/worker.py", line 906, in get_objects
    raise value.as_instanceof_cause()
ray.exceptions.RayTaskError(RaySystemError): ray::main_task() (pid=88844, ip=10.59.144.61)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/main_ppo.py", line 188, in main_task
    trainer.init_workers()
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/ppo/ray_trainer.py", line 494, in init_workers
    wg_dict = self.ray_worker_group_cls(resource_pool=resource_pool, ray_cls_with_init=worker_dict_cls)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 197, in __init__
    self._init_with_resource_pool(resource_pool=resource_pool,
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 220, in _init_with_resource_pool
    pgs = resource_pool.get_placement_groups(strategy=strategy)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 80, in get_placement_groups
    pgs = [
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 81, in <listcomp>
    placement_group(bundles=bundles, strategy=strategy, name=pg_name_prefix + str(idx), lifetime=lifetime)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/util/placement_group.py", line 211, in placement_group
    placement_group_id = worker.core_worker.create_placement_group(
  File "python/ray/includes/common.pxi", line 104, in ray._raylet.check_status
ray.exceptions.RaySystemError: System error: Failed to create placement group '34872bc423461785f8db1e66eac101000000' because name 'global_poolverl_group_2:0' already exists.

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.

what is the reason?

[ArlanCooper]

my env is

cuda 12.2
torch 2.4.0
vllm 0.6.3

[Hanpx20]

same error

[yangzhch6]

Multi GPU works for me, but single gpu script encountered the same error

[ITSXU0]

i ran the code as readme,

# 设置参数
export CUDA_VISIBLE_DEVICES=2,3
export N_GPUS=2
export BASE_MODEL=/home/octopus/data/llm_list/Qwen2.5-3B
export DATA_DIR=/home/powerop/work/rwq/myoperate/TinyZero/data_sets
export ROLLOUT_TP_SIZE=2
export EXPERIMENT_NAME=countdown-qwen2.5-3b
export VLLM_ATTENTION_BACKEND=XFORMERS

bash ./scripts/train_tiny_zero.sh

i got this error:

Traceback (most recent call last):
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/main_ppo.py", line 103, in main
    ray.get(main_task.remote(config))
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/auto_init_hook.py", line 21, in auto_init_wrapper
    return fn(*args, **kwargs)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/client_mode_hook.py", line 103, in wrapper
    return func(*args, **kwargs)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/worker.py", line 2755, in get
    values, debugger_breakpoint = worker.get_objects(object_refs, timeout=timeout)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/_private/worker.py", line 906, in get_objects
    raise value.as_instanceof_cause()
ray.exceptions.RayTaskError(RaySystemError): ray::main_task() (pid=88844, ip=10.59.144.61)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/main_ppo.py", line 188, in main_task
    trainer.init_workers()
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/trainer/ppo/ray_trainer.py", line 494, in init_workers
    wg_dict = self.ray_worker_group_cls(resource_pool=resource_pool, ray_cls_with_init=worker_dict_cls)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 197, in __init__
    self._init_with_resource_pool(resource_pool=resource_pool,
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 220, in _init_with_resource_pool
    pgs = resource_pool.get_placement_groups(strategy=strategy)
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 80, in get_placement_groups
    pgs = [
  File "/home/powerop/work/rwq/myoperate/TinyZero/verl/single_controller/ray/base.py", line 81, in <listcomp>
    placement_group(bundles=bundles, strategy=strategy, name=pg_name_prefix + str(idx), lifetime=lifetime)
  File "/home/octopus/work/conda/envs/ds_zero/lib/python3.9/site-packages/ray/util/placement_group.py", line 211, in placement_group
    placement_group_id = worker.core_worker.create_placement_group(
  File "python/ray/includes/common.pxi", line 104, in ray._raylet.check_status
ray.exceptions.RaySystemError: System error: Failed to create placement group '34872bc423461785f8db1e66eac101000000' because name 'global_poolverl_group_2:0' already exists.

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.

what is the reason?

Have you solved this problem？ I have the same problem, thank you.

[anapple-hub]

same error

[PinzhengWang322]

same error