Use latest dataset (#204)

mila-iqia · Feb 27, 2024 · 77fa0bc · 77fa0bc
1 parent e681406
commit 77fa0bc
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/benchmarks/accelerate_opt/main.py b/benchmarks/accelerate_opt/main.py
@@ -145,19 +145,20 @@ def mblog(data):
  validation_split_percentage = config["validation_split_percentage"]
  dataset_name = config["dataset_name"]
  dataset_config_name = config["dataset_config_name"]
- raw_datasets = load_dataset(dataset_name, dataset_config_name, revision="f5562967961a45407fa15044c5535a607200983f")
+ revision = "b08601e04326c79dfdd32d625aee71d232d685c3"
+ raw_datasets = load_dataset(dataset_name, dataset_config_name, revision=revision)
  if "validation" not in raw_datasets.keys():
  raw_datasets["validation"] = load_dataset(
  dataset_name,
  dataset_config_name,
  split=f"train[:{validation_split_percentage}%]",
- revision="f5562967961a45407fa15044c5535a607200983f",
+ revision=revision,
  )
  raw_datasets["train"] = load_dataset(
  dataset_name,
  dataset_config_name,
  split=f"train[{validation_split_percentage}%:]",
- revision="f5562967961a45407fa15044c5535a607200983f",
+ revision=revision,
  )
 
  model_name = config["model_name"]