Spaces:
Runtime error
Runtime error
legolasyiu
commited on
Commit
•
0a48bf2
1
Parent(s):
aaa5372
Update app.py
Browse files
app.py
CHANGED
@@ -1,20 +1,47 @@
|
|
|
|
|
|
1 |
import torch
|
2 |
-
from transformers import pipeline
|
3 |
|
4 |
-
# Load the model and tokenizer
|
5 |
-
model = pipeline('question-answering', model='EpistemeAI/Fireball-Meta-Llama-3.1-8B-Instruct-Agent-0.003-128K', tokenizer='EpistemeAI/Fireball-Meta-Llama-3.1-8B-Instruct-Agent-0.003-128K')
|
6 |
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
|
11 |
-
# Create a Gradio interface
|
12 |
-
demo = gr.Interface(
|
13 |
-
fn=chatbot,
|
14 |
-
inputs=gr.Textbox(label="User Input"),
|
15 |
-
outputs=gr.Textbox(label="Chatbot Response"),
|
16 |
-
title="Chatbot Demo"
|
17 |
-
)
|
18 |
|
19 |
-
|
20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
2 |
+
import gradio as gr
|
3 |
import torch
|
|
|
4 |
|
|
|
|
|
5 |
|
6 |
+
title = "AI Agemt ChatBot"
|
7 |
+
description = "A State-of-the-Art Agent Chatbot"
|
8 |
+
examples = [["How are you?"]]
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
+
tokenizer = AutoTokenizer.from_pretrained("EpistemeAI/Fireball-Meta-Llama-3.1-8B-Instruct-Agent-0.003-128K")
|
12 |
+
model = AutoModelForCausalLM.from_pretrained("EpistemeAI/Fireball-Meta-Llama-3.1-8B-Instruct-Agent-0.003-128K")
|
13 |
+
|
14 |
+
|
15 |
+
def predict(input, history=[]):
|
16 |
+
# tokenize the new input sentence
|
17 |
+
new_user_input_ids = tokenizer.encode(
|
18 |
+
input + tokenizer.eos_token, return_tensors="pt"
|
19 |
+
)
|
20 |
+
|
21 |
+
# append the new user input tokens to the chat history
|
22 |
+
bot_input_ids = torch.cat([torch.LongTensor(history), new_user_input_ids], dim=-1)
|
23 |
+
|
24 |
+
# generate a response
|
25 |
+
history = model.generate(
|
26 |
+
bot_input_ids, max_length=4000, pad_token_id=tokenizer.eos_token_id
|
27 |
+
).tolist()
|
28 |
+
|
29 |
+
# convert the tokens to text, and then split the responses into lines
|
30 |
+
response = tokenizer.decode(history[0]).split("<|endoftext|>")
|
31 |
+
# print('decoded_response-->>'+str(response))
|
32 |
+
response = [
|
33 |
+
(response[i], response[i + 1]) for i in range(0, len(response) - 1, 2)
|
34 |
+
] # convert to tuples of list
|
35 |
+
# print('response-->>'+str(response))
|
36 |
+
return response, history
|
37 |
+
|
38 |
+
|
39 |
+
gr.Interface(
|
40 |
+
fn=predict,
|
41 |
+
title=title,
|
42 |
+
description=description,
|
43 |
+
examples=examples,
|
44 |
+
inputs=["text", "state"],
|
45 |
+
outputs=["chatbot", "state"],
|
46 |
+
theme="finlaymacklon/boxy_violet",
|
47 |
+
).launch()
|