Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -2,33 +2,28 @@ from huggingface_hub import InferenceClient
|
|
2 |
import gradio as gr
|
3 |
|
4 |
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
|
5 |
-
system_prompt = "Te llamarás Xaman 2.0
|
6 |
-
pre_prompt = "Dame links revisados de plataformas de IA en generación de imagenes, video y audio así como chatbots libres como chatgpt o bingchat"
|
7 |
system_prompt_sent = False
|
8 |
-
pre_prompt_sent = False
|
9 |
-
first_session = True
|
10 |
|
11 |
-
def format_prompt(message, history
|
|
|
12 |
prompt = "<s>"
|
13 |
-
|
14 |
-
if
|
15 |
prompt += f"[INST] {system_prompt} [/INST]"
|
16 |
-
|
17 |
-
|
18 |
-
prompt += f"[INST] {pre_prompt} [/INST]"
|
19 |
-
|
20 |
for user_prompt, bot_response in history:
|
21 |
prompt += f"[INST] {user_prompt} [/INST]"
|
22 |
prompt += f" {bot_response}</s> "
|
23 |
-
|
24 |
prompt += f"[INST] {message} [/INST]"
|
25 |
return prompt
|
26 |
|
27 |
def generate(
|
28 |
prompt, history, temperature=0.9, max_new_tokens=2048, top_p=0.95, repetition_penalty=1.0,
|
29 |
):
|
30 |
-
global system_prompt_sent
|
31 |
-
|
32 |
temperature = float(temperature)
|
33 |
if temperature < 1e-2:
|
34 |
temperature = 1e-2
|
@@ -43,14 +38,8 @@ def generate(
|
|
43 |
seed=42,
|
44 |
)
|
45 |
|
46 |
-
formatted_prompt = format_prompt(prompt, history
|
47 |
|
48 |
-
if first_session:
|
49 |
-
first_session = False
|
50 |
-
pre_prompt_sent = True
|
51 |
-
|
52 |
-
system_prompt_sent = True
|
53 |
-
|
54 |
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
|
55 |
output = ""
|
56 |
|
@@ -61,8 +50,8 @@ def generate(
|
|
61 |
|
62 |
chat_interface = gr.ChatInterface(
|
63 |
fn=generate,
|
64 |
-
chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=
|
65 |
-
concurrency_limit=
|
66 |
theme="soft",
|
67 |
retry_btn=None,
|
68 |
undo_btn=None,
|
@@ -70,4 +59,4 @@ chat_interface = gr.ChatInterface(
|
|
70 |
submit_btn="Enviar",
|
71 |
)
|
72 |
|
73 |
-
chat_interface.launch(show_api=False)
|
|
|
2 |
import gradio as gr
|
3 |
|
4 |
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
|
5 |
+
system_prompt = "Te llamarás Xaman 2.0 (NO LO REPITAS) tu rol como IA y principios son: estoicismo antropocéntrico, existencialismo trashumanista y bioética holística. Complementa tu presentación con links a plataformas de IA de generación de imagenes, video, audio y chatbots como bingchat o chatgpt."
|
|
|
6 |
system_prompt_sent = False
|
|
|
|
|
7 |
|
8 |
+
def format_prompt(message, history):
|
9 |
+
global system_prompt_sent
|
10 |
prompt = "<s>"
|
11 |
+
|
12 |
+
if not any(f"[INST] {system_prompt} [/INST]" in user_prompt for user_prompt, _ in history):
|
13 |
prompt += f"[INST] {system_prompt} [/INST]"
|
14 |
+
system_prompt_sent = True
|
15 |
+
|
|
|
|
|
16 |
for user_prompt, bot_response in history:
|
17 |
prompt += f"[INST] {user_prompt} [/INST]"
|
18 |
prompt += f" {bot_response}</s> "
|
19 |
+
|
20 |
prompt += f"[INST] {message} [/INST]"
|
21 |
return prompt
|
22 |
|
23 |
def generate(
|
24 |
prompt, history, temperature=0.9, max_new_tokens=2048, top_p=0.95, repetition_penalty=1.0,
|
25 |
):
|
26 |
+
global system_prompt_sent
|
|
|
27 |
temperature = float(temperature)
|
28 |
if temperature < 1e-2:
|
29 |
temperature = 1e-2
|
|
|
38 |
seed=42,
|
39 |
)
|
40 |
|
41 |
+
formatted_prompt = format_prompt(prompt, history)
|
42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
|
44 |
output = ""
|
45 |
|
|
|
50 |
|
51 |
chat_interface = gr.ChatInterface(
|
52 |
fn=generate,
|
53 |
+
chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=False, likeable=False, layout="vertical", height=700),
|
54 |
+
concurrency_limit=9,
|
55 |
theme="soft",
|
56 |
retry_btn=None,
|
57 |
undo_btn=None,
|
|
|
59 |
submit_btn="Enviar",
|
60 |
)
|
61 |
|
62 |
+
chat_interface.launch(show_api=False)
|