om4r932 commited on
Commit
c1729b7
·
1 Parent(s): 4010d0e

Increased max workers for concurrency

Browse files
Files changed (2) hide show
  1. spec_indexer.py +1 -1
  2. tdoc_indexer.py +2 -2
spec_indexer.py CHANGED
@@ -269,7 +269,7 @@ def main():
269
  if os.path.exists("indexed_docs_content.zip"):
270
  +print(f"Chargement de {len(documents_by_spec_num)} documents depuis le cache.")
271
  try:
272
- with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:
273
  futures = [executor.submit(process_specification, spec) for spec in specifications]
274
  while True:
275
  if all(f.done() for f in futures):
 
269
  if os.path.exists("indexed_docs_content.zip"):
270
  +print(f"Chargement de {len(documents_by_spec_num)} documents depuis le cache.")
271
  try:
272
+ with concurrent.futures.ThreadPoolExecutor(max_workers=16) as executor:
273
  futures = [executor.submit(process_specification, spec) for spec in specifications]
274
  while True:
275
  if all(f.done() for f in futures):
tdoc_indexer.py CHANGED
@@ -14,7 +14,7 @@ from typing import List, Dict, Any
14
  warnings.filterwarnings("ignore")
15
 
16
  class TsgDocIndexer:
17
- def __init__(self, max_workers=10):
18
  self.indexer = self.load_indexer()
19
  self.main_ftp_url = "https://3gpp.org/ftp"
20
  self.dataset = load_dataset("OrganizedProgrammers/3GPPTDocLocation")
@@ -266,7 +266,7 @@ class TsgDocIndexer:
266
 
267
  def main():
268
  # Nombre de workers pour le multithreading (ajustable selon les ressources disponibles)
269
- max_workers = 64
270
 
271
  indexer = TsgDocIndexer(max_workers=max_workers)
272
 
 
14
  warnings.filterwarnings("ignore")
15
 
16
  class TsgDocIndexer:
17
+ def __init__(self, max_workers=16):
18
  self.indexer = self.load_indexer()
19
  self.main_ftp_url = "https://3gpp.org/ftp"
20
  self.dataset = load_dataset("OrganizedProgrammers/3GPPTDocLocation")
 
266
 
267
  def main():
268
  # Nombre de workers pour le multithreading (ajustable selon les ressources disponibles)
269
+ max_workers = 16
270
 
271
  indexer = TsgDocIndexer(max_workers=max_workers)
272