Refactor configs to have env in seperate yaml + Fix training

这个提交包含在:
Cadene
2024-02-25 17:42:47 +00:00
父节点 eec134d72b
当前提交 b16c334825
共有 13 个文件被更改,包括 146 次插入 和 54 次删除

查看文件

@ -10,28 +10,29 @@ def make_offline_buffer(cfg, sampler=None):
overwrite_sampler = sampler is not None
if not overwrite_sampler:
num_traj_per_batch = cfg.batch_size # // cfg.horizon
# TODO(rcadene): move batch_size outside
num_traj_per_batch = cfg.policy.batch_size # // cfg.horizon
# TODO(rcadene): Sampler outputs a batch_size <= cfg.batch_size.
# We would need to add a transform to pad the tensordict to ensure batch_size == cfg.batch_size.
sampler = PrioritizedSliceSampler(
max_capacity=100_000,
alpha=cfg.per_alpha,
beta=cfg.per_beta,
alpha=cfg.policy.per_alpha,
beta=cfg.policy.per_beta,
num_slices=num_traj_per_batch,
strict_length=False,
)
if cfg.env == "simxarm":
if cfg.env.name == "simxarm":
# TODO(rcadene): add PrioritizedSliceSampler inside Simxarm to not have to `sampler.extend(index)` here
offline_buffer = SimxarmExperienceReplay(
f"xarm_{cfg.task}_medium",
f"xarm_{cfg.env.task}_medium",
# download="force",
download=True,
streaming=False,
root="data",
sampler=sampler,
)
elif cfg.env == "pusht":
elif cfg.env.name == "pusht":
offline_buffer = PushtExperienceReplay(
"pusht",
# download="force",
@ -41,7 +42,7 @@ def make_offline_buffer(cfg, sampler=None):
sampler=sampler,
)
else:
raise ValueError(cfg.env)
raise ValueError(cfg.env.name)
if not overwrite_sampler:
num_steps = len(offline_buffer)