Update app.py
Browse files
app.py
CHANGED
@@ -109,7 +109,13 @@ def chat_with_model(user_input, chat_history=[]):
|
|
109 |
)
|
110 |
|
111 |
outputs = model.generate(input_ids, max_new_tokens=1200, do_sample=True, top_k=50, temperature=0.7)
|
112 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
|
114 |
# End model generation span
|
115 |
generation_span.end(
|
|
|
109 |
)
|
110 |
|
111 |
outputs = model.generate(input_ids, max_new_tokens=1200, do_sample=True, top_k=50, temperature=0.7)
|
112 |
+
# response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
113 |
+
|
114 |
+
# Decode and clean the response to remove unwanted repetitions of "assistant"
|
115 |
+
response = self.tokenizer.decode(outputs[0][input_ids.shape[-1]:], skip_special_tokens=True).strip()
|
116 |
+
|
117 |
+
# Remove potential repeated assistant text
|
118 |
+
response = response.replace("<|assistant|>", "").strip()
|
119 |
|
120 |
# End model generation span
|
121 |
generation_span.end(
|