make gpt2 decode output just once at the end (#2869)

also updated function name from greedy_until to generate, as it's not greedy nor until

examples/gpt2.py

@@ -132,7 +132,7 @@ class GPT2:
     self.model = model
     self.tokenizer = tokenizer
 
-  def greedy_until(self, prompt:str, max_length:int, temperature:float, timing:bool=False, batch_size:int=1):
+  def generate(self, prompt:str, max_length:int, temperature:float, timing:bool=False, batch_size:int=1):
     prompt_tokens = self.tokenizer.encode(prompt, allowed_special={"<|endoftext|>"})
     toks = [prompt_tokens[:] for _ in range(batch_size)]
     start_pos = 0
@@ -153,8 +153,7 @@ class GPT2:
         tok = probs.multinomial().flatten().numpy().tolist()
       start_pos = len(toks[0])
       for i,t in enumerate(tok): toks[i].append(t)
-      output = [self.tokenizer.decode(x) for x in toks]
-    return output
+    return [self.tokenizer.decode(x) for x in toks]
 
 # **** main code ****
 
@@ -189,7 +188,7 @@ if __name__ == "__main__":
   if args.benchmark != -1:
     gpt2.model(Tensor.rand(args.batch_size, args.benchmark), Variable("a", 0, MAX_CONTEXT).bind(0)).realize()
   else:
-    texts = gpt2.greedy_until(args.prompt, args.count, args.temperature, timing=args.timing, batch_size=args.batch_size)
+    texts = gpt2.generate(args.prompt, args.count, args.temperature, timing=args.timing, batch_size=args.batch_size)
     if not args.noshow:
       print('Generating text...')
       if len(texts) == 1: print(texts[0])