Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1 +1,69 @@
|
|
| 1 |
-
from huggingface_hub import InferenceClient
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from huggingface_hub import InferenceClient
|
| 2 |
+
import gradio as gr
|
| 3 |
+
import random
|
| 4 |
+
import prompts
|
| 5 |
+
|
| 6 |
+
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
|
| 7 |
+
|
| 8 |
+
def format_prompt(message, history):
|
| 9 |
+
prompt = "<s>"
|
| 10 |
+
for user_prompt, bot_response in history:
|
| 11 |
+
prompt += f"[INST] {user_prompt} [/INST]"
|
| 12 |
+
prompt += f" {bot_response}</s> "
|
| 13 |
+
prompt += f"[INST] {message} [/INST]"
|
| 14 |
+
return prompt
|
| 15 |
+
|
| 16 |
+
def generate(prompt, history, max_new_tokens, seed):
|
| 17 |
+
print(seed)
|
| 18 |
+
system_prompt = prompts.ARTICLE_WRITER.format(history=history)
|
| 19 |
+
generate_kwargs = dict(
|
| 20 |
+
temperature=0.9,
|
| 21 |
+
max_new_tokens=max_new_tokens,
|
| 22 |
+
top_p=0.95,
|
| 23 |
+
repetition_penalty=1.0,
|
| 24 |
+
do_sample=True,
|
| 25 |
+
seed=seed,
|
| 26 |
+
)
|
| 27 |
+
formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
|
| 28 |
+
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
| 29 |
+
output = ""
|
| 30 |
+
for response in stream:
|
| 31 |
+
output += response.token.text
|
| 32 |
+
yield output
|
| 33 |
+
|
| 34 |
+
additional_inputs = [
|
| 35 |
+
gr.Slider(
|
| 36 |
+
label="Max new tokens",
|
| 37 |
+
value=4096,
|
| 38 |
+
minimum=10,
|
| 39 |
+
maximum=1048*10,
|
| 40 |
+
step=64,
|
| 41 |
+
interactive=True,
|
| 42 |
+
info="The maximum numbers of words the chatbot will return",
|
| 43 |
+
),
|
| 44 |
+
gr.Slider(
|
| 45 |
+
label="Seed",
|
| 46 |
+
value=random.randint(1,9999999999999999),
|
| 47 |
+
minimum=0,
|
| 48 |
+
maximum=9999999999999999,
|
| 49 |
+
step=64,
|
| 50 |
+
interactive=True,
|
| 51 |
+
info="Each seed produces a different output to a single prompt",
|
| 52 |
+
),
|
| 53 |
+
]
|
| 54 |
+
|
| 55 |
+
examples = [
|
| 56 |
+
["Write a 1000-word article about the benefits of a healthy lifestyle."],
|
| 57 |
+
["Compose a 2000-word essay on the impact of technology on modern society."],
|
| 58 |
+
["Generate a 1500-word article discussing the importance of environmental conservation."],
|
| 59 |
+
["Create a 3000-word piece exploring the role of art in shaping cultural identity."],
|
| 60 |
+
]
|
| 61 |
+
|
| 62 |
+
gr.ChatInterface(
|
| 63 |
+
fn=generate,
|
| 64 |
+
chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
|
| 65 |
+
title="Article Writer",
|
| 66 |
+
additional_inputs=additional_inputs,
|
| 67 |
+
examples=examples,
|
| 68 |
+
concurrency_limit=20,
|
| 69 |
+
).launch(show_api=False)
|