[Public release 26/04] Introducing Mega MoE, FP4 Indexer and other features/fixes (#304)

* Merge with private repo

* Update README

* Update README

* Update README

* Add PyTorch requirements

* Fix sync scopes for MQA logits (#256)

* Update README
This commit is contained in:
Chenggang Zhao
2026-04-17 09:45:14 +08:00
committed by GitHub
parent d30fc36c8f
commit 7f2a703ed5
109 changed files with 12101 additions and 3219 deletions

View File

@@ -1,3 +1,4 @@
import argparse
import torch
import torch.multiprocessing as mp
import deep_gemm
@@ -8,7 +9,11 @@ def main(local_rank: int):
if __name__ == '__main__':
procs = [mp.Process(target=main, args=(i, ), ) for i in range(8)]
parser = argparse.ArgumentParser(description='Test lazy initialization')
parser.add_argument('--num-processes', type=int, default=8, help='Number of processes to spawn (default: 8)')
args = parser.parse_args()
procs = [mp.Process(target=main, args=(i, ), ) for i in range(args.num_processes)]
for p in procs:
p.start()
for p in procs: