diff --git a/README.md b/README.md index 97bb0fa..9a263c0 100644 --- a/README.md +++ b/README.md @@ -72,6 +72,7 @@ We have a series of technical articles on the functionalities of EasyDistill. "per_device_train_batch_size": 1, "gradient_accumulation_steps": 8, "save_steps": 1000, + "max_length": 512, "logging_steps": 1, "learning_rate": 2e-5, "weight_decay": 0.05, diff --git a/requirements.txt b/requirements.txt index 3331459..56f1a73 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -transformers==4.51.0 +transformers==4.51.1 transformers-stream-generator==0.0.5 trl==0.17.0 tokenizers==0.21.1