Spaces:
Sleeping
Sleeping
Commit
·
749e9d7
1
Parent(s):
1a5b783
Adding chat history
Browse files
app.py
CHANGED
@@ -26,8 +26,15 @@ tokenizer.padding_side = "right"
|
|
26 |
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
|
27 |
|
28 |
def infer(message, history):
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, num_beams=1, temperature=0.3, top_k=50, top_p=0.95, max_time= 180)
|
31 |
return outputs[0]['generated_text'][len(prompt):].strip()
|
32 |
|
33 |
-
gr.ChatInterface(infer).launch()
|
|
|
26 |
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
|
27 |
|
28 |
def infer(message, history):
|
29 |
+
chat_list = []
|
30 |
+
for chat in history:
|
31 |
+
chat_user = {"role":"user", "content":chat[0]}
|
32 |
+
chat_assistant = {"role":"assistant", "content":chat[1]}
|
33 |
+
chat_list.append(chat_user)
|
34 |
+
chat_list.append(chat_assistant)
|
35 |
+
chat_list.append({"role": "user", "content": message})
|
36 |
+
prompt = pipe.tokenizer.apply_chat_template(chat_list, tokenize=False, add_generation_prompt=True)
|
37 |
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, num_beams=1, temperature=0.3, top_k=50, top_p=0.95, max_time= 180)
|
38 |
return outputs[0]['generated_text'][len(prompt):].strip()
|
39 |
|
40 |
+
gr.ChatInterface(infer, title="Phi-3 Assistant").launch()
|