bump transformer

Author: Vincent Chen

setup.py

@@ -55,7 +55,7 @@
     'mosaicml[libcloud,wandb,oci,gcs,mlflow]>=0.28.0,<0.29',
     'mlflow>=2.14.1,<2.19',
     'accelerate>=0.25,<1.4',  # for HF inference `device_map`
-    'transformers>=4.43.2,<4.47',
+    'transformers>=v4.49.0,<4.50',
     'mosaicml-streaming>=0.11.0,<0.12',
     'torch>=2.5.1,<2.5.2',
     'datasets>=3.3.2,<3.4',