Update README.md

README.md

@@ -12,21 +12,9 @@ news articles using a transformer model**.
 
 ## Usage
 ```python
-from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
-name = "LukasStankevicius/t5-base-lithuanian-news-summaries-175"
-tokenizer = AutoTokenizer.from_pretrained(name)
-model = AutoModelForSeq2SeqLM.from_pretrained(name)
-
-def decode(x):
-    return tokenizer.decode(x, skip_special_tokens=True)
-    
-def summarize(text_, **g_kwargs):
-  text_ = ' '.join(text_.strip().split())
-  input_dict = tokenizer(text_,  padding=True, return_tensors="pt", 
-                         return_attention_mask=True)
-  output = model.generate(**input_dict, **g_kwargs)
-  predicted = list(map(decode, output.tolist()))[0]
-  return predicted
+from transformers import pipeline
+name= "LukasStankevicius/t5-base-lithuanian-news-summaries-175"
+my_pipeline = pipeline(task="text2text-generation", model=name, framework="pt")
 ```
 Given the following article body from [15min](#https://www.15min.lt/24sek/naujiena/lietuva/tarp-penkiu-rezultatyviausiu-tsrs-rinktines-visu-laiku-zaideju-trys-lietuviai-875-1380030):
 ```
@@ -41,9 +29,10 @@ Tarp žaidėjų, kurie sužaidė bent po 50 oficialių rungtynių Lietuvos rinkt
 ```
 The summary can be obtained by:
 ```
+text = ' '.join(text.strip().split())
 g_kwargs = dict(max_length=512, num_beams=10, no_repeat_ngram_size=2, 
                 early_stopping=True)
-summarize(text, **g_kwargs)
+my_pipeline(text, truncation=True, **g_kwargs)[0]["generated_text"]
 ```
 Output from above would be: