declare safetensors, average_checkpoints reads it

Author: Kaushik-Kumar-CEG

etc/requirements-ml.txt

@@ -13,9 +13,8 @@ accelerate>=0.33
 # CRF layer on top of the token classifier
 pytorch-crf>=0.7.2
 
-# reading the trained weights and pulling the model from the hub
+# reading and averaging the saved checkpoints
 safetensors>=0.4
-huggingface-hub>=0.24
 
 # 8 bit optimizer to fit deberta-large on a 16gb gpu, optional at runtime
 bitsandbytes>=0.43