Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,18 +1,15 @@
|
|
| 1 |
from huggingface_hub import InferenceClient
|
| 2 |
import gradio as gr
|
| 3 |
|
| 4 |
-
client = InferenceClient(
|
| 5 |
-
"mistralai/Mistral-7B-Instruct-v0.1"
|
| 6 |
-
)
|
| 7 |
-
|
| 8 |
|
| 9 |
def format_prompt(message, history):
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
|
| 17 |
def generate(
|
| 18 |
prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
|
|
@@ -33,7 +30,9 @@ def generate(
|
|
| 33 |
|
| 34 |
formatted_prompt = format_prompt(prompt, history)
|
| 35 |
|
| 36 |
-
stream = client.text_generation(
|
|
|
|
|
|
|
| 37 |
output = ""
|
| 38 |
|
| 39 |
for response in stream:
|
|
@@ -41,8 +40,7 @@ def generate(
|
|
| 41 |
yield output
|
| 42 |
return output
|
| 43 |
|
| 44 |
-
|
| 45 |
-
additional_inputs=[
|
| 46 |
gr.Slider(
|
| 47 |
label="Temperature",
|
| 48 |
value=0.9,
|
|
@@ -78,13 +76,21 @@ additional_inputs=[
|
|
| 78 |
step=0.05,
|
| 79 |
interactive=True,
|
| 80 |
info="Penalize repeated tokens",
|
| 81 |
-
)
|
| 82 |
]
|
| 83 |
|
|
|
|
| 84 |
|
| 85 |
gr.ChatInterface(
|
| 86 |
fn=generate,
|
| 87 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
additional_inputs=additional_inputs,
|
| 89 |
-
title="""Gen-AIInterviewUseCase featuring Mistral 7B"""
|
| 90 |
-
).launch(show_api=False)
|
|
|
|
| 1 |
from huggingface_hub import InferenceClient
|
| 2 |
import gradio as gr
|
| 3 |
|
| 4 |
+
client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.1")
|
|
|
|
|
|
|
|
|
|
| 5 |
|
| 6 |
def format_prompt(message, history):
|
| 7 |
+
prompt = "<s>"
|
| 8 |
+
for user_prompt, bot_response in history:
|
| 9 |
+
prompt += f"[INST] {user_prompt} [/INST]"
|
| 10 |
+
prompt += f" {bot_response}</s> "
|
| 11 |
+
prompt += f"[INST] {message} [/INST]"
|
| 12 |
+
return prompt
|
| 13 |
|
| 14 |
def generate(
|
| 15 |
prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
|
|
|
|
| 30 |
|
| 31 |
formatted_prompt = format_prompt(prompt, history)
|
| 32 |
|
| 33 |
+
stream = client.text_generation(
|
| 34 |
+
formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False
|
| 35 |
+
)
|
| 36 |
output = ""
|
| 37 |
|
| 38 |
for response in stream:
|
|
|
|
| 40 |
yield output
|
| 41 |
return output
|
| 42 |
|
| 43 |
+
additional_inputs = [
|
|
|
|
| 44 |
gr.Slider(
|
| 45 |
label="Temperature",
|
| 46 |
value=0.9,
|
|
|
|
| 76 |
step=0.05,
|
| 77 |
interactive=True,
|
| 78 |
info="Penalize repeated tokens",
|
| 79 |
+
),
|
| 80 |
]
|
| 81 |
|
| 82 |
+
initial_prompt = "Hello John! Welcome to our simulation interview for the Solution Architect position. I understand you may have some initial thoughts or questions about the job requirements. Feel free to share those with me."
|
| 83 |
|
| 84 |
gr.ChatInterface(
|
| 85 |
fn=generate,
|
| 86 |
+
inputs=[gr.Textbox(default=initial_prompt, label="Initial Prompt", type="text")],
|
| 87 |
+
chatbot=gr.Chatbot(
|
| 88 |
+
show_label=False,
|
| 89 |
+
show_share_button=False,
|
| 90 |
+
show_copy_button=True,
|
| 91 |
+
likeable=True,
|
| 92 |
+
layout="panel",
|
| 93 |
+
),
|
| 94 |
additional_inputs=additional_inputs,
|
| 95 |
+
title="""Gen-AIInterviewUseCase featuring Mistral 7B""",
|
| 96 |
+
).launch(show_api=False)
|