Commit 2f3a2d68 authored by Jared Casper's avatar Jared Casper
Browse files

Merge branch 'extra_assertion' into 'main'

Make sure pipeline-model-parallel size is greater than 2 for interleaved schedule

See merge request ADLR/megatron-lm!253
parents a6e00d97 182841f7
...@@ -116,6 +116,9 @@ def parse_args(extra_args_provider=None, defaults={}, ...@@ -116,6 +116,9 @@ def parse_args(extra_args_provider=None, defaults={},
args.global_batch_size), flush=True) args.global_batch_size), flush=True)
assert args.global_batch_size > 0 assert args.global_batch_size > 0
if args.num_layers_per_virtual_pipeline_stage is not None: if args.num_layers_per_virtual_pipeline_stage is not None:
assert args.pipeline_model_parallel_size > 2, \
'pipeline-model-parallel size should be greater than 2 with ' \
'interleaved schedule'
assert args.num_layers % args.num_layers_per_virtual_pipeline_stage == 0, \ assert args.num_layers % args.num_layers_per_virtual_pipeline_stage == 0, \
'number of layers is not divisible by number of layers per virtual ' \ 'number of layers is not divisible by number of layers per virtual ' \
'pipeline stage' 'pipeline stage'
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment