[AMD][CI][BugFix] Fix Distributed Compile Unit Tests (2xH100-2xMI300) group (#43120)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>
This commit is contained in:
@@ -75,6 +75,7 @@ llama3_8b = ModelFusionInfo(
|
||||
model_name="meta-llama/Llama-3.1-8B-Instruct",
|
||||
matches=lambda n_layers: Matches(
|
||||
ar_rms_fusion=n_layers * 2 + 1,
|
||||
aiter_ar_rms_fusion=n_layers * 2,
|
||||
sequence_parallel=n_layers * 2 + 1,
|
||||
async_tp=n_layers * 4,
|
||||
),
|
||||
@@ -136,6 +137,7 @@ qwen3_a3b = ModelFusionInfo(
|
||||
matches=lambda n_layers: Matches(
|
||||
norm_rope_fusion=n_layers,
|
||||
ar_rms_fusion=n_layers * 2 + 1,
|
||||
aiter_ar_rms_fusion=n_layers * 2,
|
||||
sequence_parallel=n_layers * 2 + 1,
|
||||
async_tp=n_layers * 2,
|
||||
),
|
||||
@@ -211,6 +213,7 @@ gpt_oss_20b = ModelFusionInfo(
|
||||
model_name="openai/gpt-oss-20b",
|
||||
matches=lambda n_layers: Matches(
|
||||
ar_rms_fusion=n_layers * 2 + 1,
|
||||
aiter_ar_rms_fusion=n_layers + 1,
|
||||
sequence_parallel=n_layers * 2 + 1,
|
||||
async_tp=n_layers * 2,
|
||||
),
|
||||
|
||||
Reference in New Issue
Block a user