File size: 4,148 Bytes
2e74f63 3ab75c1 2e74f63 a2e12c6 4c87f5c a2e12c6 2e74f63 4c87f5c 20f4682 3ab75c1 20e3a37 82d43cd ab8e6c2 2e74f63 a2e12c6 2e74f63 3ab75c1 2e74f63 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 |
import os
import gradio as gr
from text_generation import Client
HF_TOKEN = os.getenv("HF_TOKEN")
INFERENCE_ENDPOINT = os.getenv("INFERENCE_ENDPOINT")
USER_NAME = "User"
BOT_NAME = "Falcon"
DEFAULT_INSTRUCTIONS = f"""The following is a conversation between a highly knowledgeable and intelligent AI assistant, called Falcon, and a human user, called User. In the following interactions, User and Falcon will converse in natural language, and Falcon will answer User's questions. Falcon was built to be respectful, polite and inclusive. Falcon was built by the Technology Innovation Institute in Abu Dhabi. Falcon will never decline to answer a question, and always attempts to give an answer that User would be satisfied with. It knows a lot, and always tells the truth. The conversation begins.
"""
RETRY_COMMAND = "/retry"
STOP_STR = f"\n{USER_NAME}:"
STOP_SUSPECT_LIST = [":", "\n", "User"]
client = None
if INFERENCE_ENDPOINT:
client = Client(INFERENCE_ENDPOINT, headers={"Authorization": f"Bearer {HF_TOKEN}"})
def format_chat_prompt(message: str, chat_history, instructions: str) -> str:
instructions = instructions.strip(" ").strip("\n")
prompt = instructions
for turn in chat_history:
user_message, bot_message = turn
prompt = f"{prompt}\n{USER_NAME}: {user_message}\n{BOT_NAME}: {bot_message}"
prompt = f"{prompt}\n{USER_NAME}: {message}\n{BOT_NAME}:"
return prompt
def run_chat(message: str, chat_history):
prompt = format_chat_prompt(message, chat_history, DEFAULT_INSTRUCTIONS)
chat_history = chat_history + [[message, ""]]
response = client.generate(
prompt,
do_sample=True,
max_new_tokens=1024,
stop_sequences=[STOP_STR, "<|endoftext|>"],
temperature=0.8,
top_p=0.9,
).generated_text.replace("\nUser:", "")
chat_history[-1][1] = response
return response, chat_history
with gr.Blocks() as demo:
gr.Markdown("""
# Falcon-7b-instruct Discord Bot Powered by Gradio and Hugging Face Endpoints
Make sure you read the 'Inference Endpoints' section below first! 🦅
### First install the `gradio_client`
```bash
pip install gradio_client
```
### Then deploy to discord in one line! ⚡️
```python
secrets = {"HF_TOKEN": "<your-key-here>", "INFERENCE_ENDPOINT": "<endpoint-url>"}
client = grc.Client.duplicate("gradio-discord-bots/falcon-7b-instruct", private=False, secrets=secrets, sleep_timeout=2880)
client.deploy_discord(api_names=["chat"])
""")
with gr.Accordion(label="Inference Endpoints", open=False):
gr.Markdown("""
## Setting Up Inference Endpoints 💪
To deploy this space as a discord bot, you will need to deploy your own Falcon model to Hugging Face Endpoints.
Don't worry it's super easy!
1. Go to the [model page](tiiuae/falcon-7b-instruct) 🦅
2. Click Deploy > Inference Endpoints
<img src="https://gradio-builds.s3.amazonaws.com/demo-files/discordbots/inference_endpoints/modelpage.png" alt="drawing" width="800" height=400/>
3. Select your desired cloud provider and region ☁️
<img src="https://gradio-builds.s3.amazonaws.com/demo-files/discordbots/inference_endpoints/falcon_instruct.png" alt="drawing" width="800" height=400/>
4. Optional: Set Automatic Scale to Zero. This will pause your endpoint after 15 minutes of inactivity to prevent unwanted billing. 💰
<img src="https://gradio-builds.s3.amazonaws.com/demo-files/discordbots/inference_endpoints/autoscale.png" alt="drawing" width="800" height=400/>
5. Create the endpoint! Copy the endpoint URL after it's complete.
<img src="https://gradio-builds.s3.amazonaws.com/demo-files/discordbots/inference_endpoints/running_model.png" alt="drawing" width="800" height=400/>
"""
)
button = gr.Button(visible=False)
history = gr.State([])
message = gr.Textbox(visible=False)
response = gr.Textbox(visible=False)
button.click(run_chat, [message, history], [response, history], api_name="chat")
demo.queue(concurrency_count=70).launch() |