[CI] Disable non-lazy string operation on logging (#4326)
Co-authored-by: Danny Guinther <dguinther@neuralmagic.com>
This commit is contained in:
co-authored by
Danny Guinther
parent
2f30e7c72f
commit
a88081bf76
@@ -334,10 +334,10 @@ class TensorizerAgent:
|
||||
per_second = convert_bytes(deserializer.total_tensor_bytes / duration)
|
||||
after_mem = get_mem_usage()
|
||||
deserializer.close()
|
||||
logger.info(f"Deserialized {total_bytes_str} in "
|
||||
f"{end - start:0.2f}s, {per_second}/s")
|
||||
logger.info(f"Memory usage before: {before_mem}")
|
||||
logger.info(f"Memory usage after: {after_mem}")
|
||||
logger.info("Deserialized %s in %0.2fs, %f/s", total_bytes_str,
|
||||
end - start, per_second)
|
||||
logger.info("Memory usage before: %s", before_mem)
|
||||
logger.info("Memory usage after: %s", after_mem)
|
||||
|
||||
self._check_tensors_on_meta_device()
|
||||
self._resize_lora_embeddings()
|
||||
|
||||
@@ -190,7 +190,7 @@ def download_weights_from_hf(model_name_or_path: str,
|
||||
allow_patterns = [pattern]
|
||||
break
|
||||
|
||||
logger.info(f"Using model weights format {allow_patterns}")
|
||||
logger.info("Using model weights format %s", allow_patterns)
|
||||
# Use file lock to prevent multiple processes from
|
||||
# downloading the same model weights at the same time.
|
||||
with get_lock(model_name_or_path, cache_dir):
|
||||
@@ -310,17 +310,17 @@ def kv_cache_scales_loader(
|
||||
return layer_scales_map.items()
|
||||
|
||||
except FileNotFoundError:
|
||||
logger.error(f"File or directory '{filename}' not found.")
|
||||
logger.error("File or directory '%s' not found.", filename)
|
||||
except json.JSONDecodeError:
|
||||
logger.error(f"Error decoding JSON in file '{filename}'.")
|
||||
logger.error("Error decoding JSON in file '%s'.", filename)
|
||||
except Exception as e:
|
||||
logger.error(f"An error occurred while reading '{filename}': {e}")
|
||||
logger.error("An error occurred while reading '%s': %s", filename, e)
|
||||
# This section is reached if and only if any of the excepts are hit
|
||||
# Return an empty iterable (list) => no KV cache scales are loaded
|
||||
# which ultimately defaults to 1.0 scales
|
||||
logger.warning("Defaulting to KV cache scaling factors = 1.0 "
|
||||
f"for all layers in TP rank {tp_rank} "
|
||||
"as an error occurred during loading.")
|
||||
logger.warning(
|
||||
"Defaulting to KV cache scaling factors = 1.0 for all "
|
||||
"layers in TP rank %d as an error occurred during loading.", tp_rank)
|
||||
return []
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user