zenlm/enso
1
0
Fork 0
mirror of https://github.com/zenlm/enso.git synced 2026-07-26 22:30:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

debug

Browse Source
This commit is contained in:
费政聪
2024-08-15 16:40:37 +08:00
committed by GitHub
parent 1818c20b33
commit 547a0961c8
1 changed files with 7 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
train_deepspeed.py
+5 -5
View File
@@ -216,20 +216,20 @@ def main(args):
if (data_iter_step + 1) % args.accum_iter == 0:
model_engine.step()
log_steps += 1
train_steps += 1
log_steps += 1
train_steps += 1
data_iter_step += 1
# Log loss values:
running_loss += loss.item()
if train_steps % args.log_every == 0 and train_steps > 0:
if train_steps % args.log_every == 0:
# Measure training speed:
torch.cuda.synchronize()
end_time = time()
steps_per_sec = log_steps / (end_time - start_time)
# Reduce loss history over all processes:
avg_loss = torch.tensor(running_loss / data_iter_step, device=device)
avg_loss = torch.tensor(running_loss / log_steps, device=device)
dist.all_reduce(avg_loss, op=dist.ReduceOp.SUM)
avg_loss = avg_loss.item() / dist.get_world_size()
logger.info(f"(step={train_steps:07d}) Train Loss: {avg_loss:.4f}, Train Steps/Sec: {steps_per_sec:.2f}")
@@ -267,7 +267,7 @@ if __name__ == "__main__":
parser.add_argument("--global-seed", type=int, default=2023)
parser.add_argument("--num-workers", type=int, default=4)
parser.add_argument("--log-every", type=int, default=100)
parser.add_argument('--accum_iter', default=8, type=int,)
parser.add_argument('--accum_iter', default=4, type=int,)
parser.add_argument('--num_experts', default=8, type=int,)
parser.add_argument('--num_experts_per_tok', default=2, type=int,)
parser.add_argument("--ckpt-every", type=int, default=10_000)