github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-07-23 19:49:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix loop issue

Browse Source
This commit is contained in:
Tong Li 2025-06-26 15:08:27 +08:00
parent 8abf186ce2
commit 71ef6b32c6
1 changed files with 33 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
applications/ColossalChat/coati/distributed/consumer.py
View File

@ -151,8 +151,9 @@ class BaseConsumer:
for step in pbar:
torch.cuda.reset_peak_memory_stats()
i = 0
for _ in range(self.num_recv_per_update):
self.profiler.enter(f"rollout_episode_{episode}_step_{step}")
for _ in range(self.num_recv_per_update):
if self.n_behind > 0:
# after sync model, do not wait for more data to arrive as rollout takes time, use buffered data
effective_group_to_raw_group_mapping = self.calculate_effective_group_to_raw_group_mapping(
@ -244,36 +245,36 @@ class BaseConsumer:
f"[T{dist.get_rank()}] Collect Effective Prompt: {len(effective_group_to_raw_group_mapping)}/{self.dp_size * self.minibatch_size}"
)
if self.n_behind == 0:
# If n_behind is 0, we start training after receiving data from producers.
effective_group_to_raw_group_mapping = self.calculate_effective_group_to_raw_group_mapping(
step=step
)
while len(effective_group_to_raw_group_mapping) >= self.dp_size * self.minibatch_size:
self.profiler.log(
f"Still have {len(effective_group_to_raw_group_mapping)} effective groups, greater than {self.dp_size * self.minibatch_size}, start training"
)
batch, raw_mini_batches_metric_dict = self.prepare_mini_batch(
effective_group_to_raw_group_mapping
)
self.profiler.enter("step")
loss = self.step(i, pbar, **batch, **raw_mini_batches_metric_dict)
self.profiler.exit("step")
self.buffer = self.buffer[
effective_group_to_raw_group_mapping[self.dp_size * self.minibatch_size - 1] + 1 :
]
# recalculate the effective group to raw group mapping
effective_group_to_raw_group_mapping_size_before = len(effective_group_to_raw_group_mapping)
effective_group_to_raw_group_mapping = self.calculate_effective_group_to_raw_group_mapping(
step=step
)
assert (
len(effective_group_to_raw_group_mapping)
== effective_group_to_raw_group_mapping_size_before - self.dp_size * self.minibatch_size
)
if loss is not None:
pbar.set_postfix({"loss": loss})
i += 1
if self.n_behind == 0:
# If n_behind is 0, we start training after receiving data from producers.
while len(effective_group_to_raw_group_mapping) >= self.dp_size * self.minibatch_size:
self.profiler.log(
f"Collect {len(effective_group_to_raw_group_mapping)} effective groups, greater than {self.dp_size * self.minibatch_size}, start training"
)
batch, raw_mini_batches_metric_dict = self.prepare_mini_batch(
effective_group_to_raw_group_mapping
)
self.profiler.enter("step")
loss = self.step(i, pbar, **batch, **raw_mini_batches_metric_dict)
self.profiler.exit("step")
self.buffer = self.buffer[
effective_group_to_raw_group_mapping[self.dp_size * self.minibatch_size - 1] + 1 :
]
# recalculate the effective group to raw group mapping
effective_group_to_raw_group_mapping_size_before = len(
effective_group_to_raw_group_mapping
)
effective_group_to_raw_group_mapping = (
self.calculate_effective_group_to_raw_group_mapping(step=step)
)
assert (
len(effective_group_to_raw_group_mapping)
== effective_group_to_raw_group_mapping_size_before
- self.dp_size * self.minibatch_size
)
if loss is not None:
pbar.set_postfix({"loss": loss})
i += 1
if self.lr_scheduler is not None:
self.lr_scheduler.step()
@ -314,6 +315,7 @@ class BaseConsumer:
torch.cuda.empty_cache()
self.profiler.exit("sync_model")
self.profiler.log(f"Peak memory usage: {torch.cuda.max_memory_allocated() / 1024 ** 2:.2f} MB")
self.profiler.exit(f"rollout_episode_{episode}_step_{step}")
def __del__(self):
if hasattr(self, "profiler"):