Update README.md

README.md

@@ -36,20 +36,20 @@ from transformers import AutoTokenizer, AutoModelWithLMHead
 tokenizer = AutoTokenizer.from_pretrained('tinkoff-ai/ruDialoGPT-medium')
 model = AutoModelWithLMHead.from_pretrained('tinkoff-ai/ruDialoGPT-medium')
 inputs = tokenizer('@@ПЕРВЫЙ@@ привет @@ВТОРОЙ@@ привет @@ПЕРВЫЙ@@ как дела? @@ВТОРОЙ@@', return_tensors='pt')
-with torch.inference_mode():
-    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]
+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,
+    max_new_tokens=40
+)
+context_with_response = [tokenizer.decode(sample_token_ids) for sample_token_ids in generated_token_ids]
 context_with_response
 ```