This commit is contained in:
Falko Victor Habel 2025-01-27 08:53:01 +01:00
parent d205346741
commit 32526c3c30
1 changed files with 3 additions and 3 deletions

View File

@ -9,13 +9,13 @@ import copy
def pretrain_model(data_path1, data_path2, num_epochs=3): def pretrain_model(data_path1, data_path2, num_epochs=3):
# Read and merge datasets # Read and merge datasets
df1 = pd.read_parquet(data_path1).head(5000) df1 = pd.read_parquet(data_path1).head(2500)
df2 = pd.read_parquet(data_path2).head(5000) df2 = pd.read_parquet(data_path2).head(2500)
merged_df = pd.concat([df1, df2], ignore_index=True) merged_df = pd.concat([df1, df2], ignore_index=True)
# Model configuration # Model configuration
config = AIIAConfig( config = AIIAConfig(
model_name="AIIA-Base-512x10k", model_name="AIIA-Base-512x5k",
) )
# Initialize model and data loader # Initialize model and data loader