Update README.md

README.md

@@ -37,7 +37,19 @@ tokenizer = AutoTokenizer.from_pretrained('tinkoff-ai/ruDialoGPT-small')
 model = AutoModelWithLMHead.from_pretrained('tinkoff-ai/ruDialoGPT-small')
 inputs = tokenizer('@@ПЕРВЫЙ@@ привет @@ВТОРОЙ@@ привет @@ПЕРВЫЙ@@ как дела? @@ВТОРОЙ@@', return_tensors='pt')
 with torch.inference_mode():
-    generated_token_ids = model.generate(**inputs)
-    context_with_response = tokenizer.decode(generated_token_ids[0])
+    generated_token_ids = model.generate(
+        **inputs,
+        top_k=10,
+        top_p=0.95,
+        num_beams=3,
+        num_return_sequences=3,
+        do_sample=True,
+        no_repeat_ngram_size=2,
+        temperature=1.2,
+        repetition_penalty=1.2,
+        length_penalty=1.0,
+        eos_token_id=50257
+    )
+    context_with_response = [tokenizer.decode(sample_token_ids) for sample_token_ids in generated_token_ids]
 context_with_response
 ```