[Refactor] Clean up log once scope="local" (#40540)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
This commit is contained in:
Wentao Ye
2026-04-22 16:42:43 -04:00
committed by GitHub
co-authored by gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
parent cfa49213d7
commit 8f87eb4622
44 changed files with 56 additions and 148 deletions
+1 -6
View File
@@ -4855,7 +4855,6 @@ class GPUModelRunner(
"Model loading took %s GiB memory and %.6f seconds",
format_gib(self.model_memory_usage),
time_after_load - time_before_load,
scope="local",
)
if not load_dummy_weights:
prepare_communication_buffer_for_model(self.model)
@@ -4989,7 +4988,7 @@ class GPUModelRunner(
)
# begin loading weights
logger.info_once("Reloading weights inplace...", scope="local")
logger.info_once("Reloading weights inplace...")
if is_checkpoint_format:
# load weights from checkpoint/ original model format
initialize_layerwise_reload(model)
@@ -5001,7 +5000,6 @@ class GPUModelRunner(
logger.warning_once(
"Reloading with `is_checkpoint_format=True` requires that "
"weights be in kernel format and already sharded",
scope="local",
)
loaded_weights = set()
for name, loaded_weight in weights_iterator:
@@ -5015,7 +5013,6 @@ class GPUModelRunner(
logger.info_once(
"Reloading and processing weights took %.2f seconds",
diff_seconds,
scope="local",
)
if self.model_config.quantization is None and loaded_weights is not None:
weights_not_loaded = weights_to_load - loaded_weights
@@ -5802,7 +5799,6 @@ class GPUModelRunner(
encoder_budget,
max_mm_items_per_batch,
dummy_modality,
scope="local",
)
# Create dummy batch of multimodal inputs.
@@ -6099,7 +6095,6 @@ class GPUModelRunner(
"Graph capturing finished in %.0f secs, took %.2f GiB",
elapsed_time,
cuda_graph_size / (1 << 30),
scope="local",
)
return cuda_graph_size