Minor updates

lm_eval/models/gpt2.py

@@ -18,7 +18,7 @@ class GPT2LM(LM):
         self.gpt2.eval()
 
         # pretrained tokenizer for neo is broken for now so just hardcoding this to gpt2
-        self.tokenizer = transformers.AutoTokenizer.from_pretrained('gpt2')
+        self.tokenizer = transformers.GPT2TokenizerFast.from_pretrained('gpt2')
         self.tokenizer.pad_token = "<|endoftext|>"
         self.max_length = self.gpt2.config.n_ctx
 

lm_eval/tasks/hendrycks_test.py

@@ -44,7 +44,7 @@ class GeneralHendrycksTest(MultipleChoiceTask):
         if not self.DATASET_PATH.exists():
             sh("""
                 mkdir -p data
-                wget https://people.eecs.berkeley.edu/~hendrycks/data.tar -P data/
+                wget -c https://people.eecs.berkeley.edu/~hendrycks/data.tar -P data/
                 tar -xf data/data.tar -C data/
                 rm data/data.tar
                 mv data/data data/hendrycksTest