Marcus Posey
commited on
Commit
·
109f4c0
1
Parent(s):
7fa4251
Add gradio UI
Browse files- app.py +38 -0
- requirements.txt +2 -0
app.py
ADDED
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from vllm import LLM, SamplingParams
|
2 |
+
import gradio as gr
|
3 |
+
import os
|
4 |
+
from huggingface_hub import login
|
5 |
+
|
6 |
+
|
7 |
+
class TextCompletion:
|
8 |
+
def __init__(self, model, sampling_params):
|
9 |
+
self.model = model
|
10 |
+
self.sampling_params = sampling_params
|
11 |
+
|
12 |
+
def generate(self, prompt: str):
|
13 |
+
output = self.model.generate(prompt, self.sampling_params)
|
14 |
+
response = output[0].outputs[0].text
|
15 |
+
return response
|
16 |
+
|
17 |
+
|
18 |
+
if __name__ == "__main__":
|
19 |
+
HF_TOKEN = os.getenv('HF_TOKEN')
|
20 |
+
login(token=HF_TOKEN)
|
21 |
+
|
22 |
+
model = LLM(
|
23 |
+
model="mep296/llama-3-8b-entigraph-quality",
|
24 |
+
tokenizer="meta-llama/Meta-Llama-3-8B",
|
25 |
+
device="cuda"
|
26 |
+
)
|
27 |
+
tokenizer = model.get_tokenizer()
|
28 |
+
sampling_params = SamplingParams(
|
29 |
+
temperature=0.1,
|
30 |
+
max_tokens=500,
|
31 |
+
stop=[tokenizer.eos_token, "## Example 7", "##"]
|
32 |
+
)
|
33 |
+
|
34 |
+
def text_completion_fn(prompt):
|
35 |
+
text_completer = TextCompletion(model, sampling_params)
|
36 |
+
return text_completer.generate(prompt)
|
37 |
+
demo = gr.Interface(fn=text_completion_fn, inputs="textbox", outputs="textbox")
|
38 |
+
demo.launch()
|
requirements.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
gradio==4.44.0
|
2 |
+
vllm==0.5.3.post1
|