Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,10 +1,7 @@
|
|
| 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 |
prompt = "<s>"
|
|
@@ -15,7 +12,7 @@ def format_prompt(message, history):
|
|
| 15 |
return prompt
|
| 16 |
|
| 17 |
def generate(
|
| 18 |
-
prompt, history, temperature=0.
|
| 19 |
):
|
| 20 |
temperature = float(temperature)
|
| 21 |
if temperature < 1e-2:
|
|
@@ -41,50 +38,15 @@ def generate(
|
|
| 41 |
yield output
|
| 42 |
return output
|
| 43 |
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
step=0.05,
|
| 52 |
-
interactive=True,
|
| 53 |
-
info="Higher values produce more diverse outputs",
|
| 54 |
-
),
|
| 55 |
-
gr.Slider(
|
| 56 |
-
label="Max new tokens",
|
| 57 |
-
value=256,
|
| 58 |
-
minimum=0,
|
| 59 |
-
maximum=1049,
|
| 60 |
-
step=64,
|
| 61 |
-
interactive=True,
|
| 62 |
-
info="The maximum numbers of new tokens",
|
| 63 |
-
),
|
| 64 |
-
gr.Slider(
|
| 65 |
-
label="Top-p (nucleus sampling)",
|
| 66 |
-
value=0.90,
|
| 67 |
-
minimum=0.0,
|
| 68 |
-
maximum=1,
|
| 69 |
-
step=0.05,
|
| 70 |
-
interactive=True,
|
| 71 |
-
info="Higher values sample more low-probability tokens",
|
| 72 |
-
),
|
| 73 |
-
gr.Slider(
|
| 74 |
-
label="Repetition penalty",
|
| 75 |
-
value=1.2,
|
| 76 |
-
minimum=1.0,
|
| 77 |
-
maximum=2.0,
|
| 78 |
-
step=0.05,
|
| 79 |
-
interactive=True,
|
| 80 |
-
info="Penalize repeated tokens",
|
| 81 |
-
)
|
| 82 |
-
]
|
| 83 |
-
|
| 84 |
|
| 85 |
-
|
| 86 |
-
fn=generate,
|
| 87 |
-
chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
|
| 88 |
-
additional_inputs=additional_inputs,
|
| 89 |
-
title="""Mistral 7B"""
|
| 90 |
-
).launch(show_api=False)
|
|
|
|
| 1 |
from huggingface_hub import InferenceClient
|
| 2 |
import gradio as gr
|
| 3 |
|
| 4 |
+
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
|
|
|
|
|
|
|
|
|
|
| 5 |
|
| 6 |
def format_prompt(message, history):
|
| 7 |
prompt = "<s>"
|
|
|
|
| 12 |
return prompt
|
| 13 |
|
| 14 |
def generate(
|
| 15 |
+
prompt, history, temperature=0.2, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
|
| 16 |
):
|
| 17 |
temperature = float(temperature)
|
| 18 |
if temperature < 1e-2:
|
|
|
|
| 38 |
yield output
|
| 39 |
return output
|
| 40 |
|
| 41 |
+
|
| 42 |
+
mychatbot = gr.Chatbot(
|
| 43 |
+
avatar_images=["./user.png", "./botm.png"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
|
| 44 |
|
| 45 |
+
demo = gr.ChatInterface(fn=generate,
|
| 46 |
+
chatbot=mychatbot,
|
| 47 |
+
title=" Mixtral 8x7b Chat",
|
| 48 |
+
retry_btn=None,
|
| 49 |
+
undo_btn=None
|
| 50 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
+
demo.queue().launch(show_api=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|