Update app.py
Browse files
app.py
CHANGED
@@ -1,29 +1,16 @@
|
|
1 |
import os
|
2 |
import spaces
|
3 |
import gradio as gr
|
|
|
|
|
|
|
4 |
from huggingface_hub import login as hf_login
|
5 |
-
|
6 |
from vllm import LLM
|
|
|
7 |
|
8 |
hf_login(token=os.getenv("HF_TOKEN"))
|
9 |
|
10 |
-
class PatientRecord(BaseModel):
|
11 |
-
life_style: str
|
12 |
-
family_history: str
|
13 |
-
social_history: str
|
14 |
-
medical_surgical_history: str
|
15 |
-
signs_symptoms: str
|
16 |
-
comorbidities: str
|
17 |
-
diagnostic_techniques_procedures: str
|
18 |
-
diagnosis: str
|
19 |
-
laboratory_values: str
|
20 |
-
pathology: str
|
21 |
-
pharmacological_therapy: str
|
22 |
-
interventional_therapy: str
|
23 |
-
patient_outcome_assessment: str
|
24 |
-
age: str
|
25 |
-
gender: str
|
26 |
-
|
27 |
model_name = "meta-llama/Llama-3.2-1B-Instruct"
|
28 |
|
29 |
model = LLM(
|
@@ -33,8 +20,77 @@ model = LLM(
|
|
33 |
enforce_eager=True,
|
34 |
)
|
35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
with gr.Blocks() as demo:
|
37 |
-
gr.Markdown("
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
38 |
|
39 |
if __name__ == "__main__":
|
40 |
demo.launch()
|
|
|
1 |
import os
|
2 |
import spaces
|
3 |
import gradio as gr
|
4 |
+
|
5 |
+
import torch
|
6 |
+
from transformers import AutoTokenizer
|
7 |
from huggingface_hub import login as hf_login
|
8 |
+
|
9 |
from vllm import LLM
|
10 |
+
from pydantic import BaseModel
|
11 |
|
12 |
hf_login(token=os.getenv("HF_TOKEN"))
|
13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
model_name = "meta-llama/Llama-3.2-1B-Instruct"
|
15 |
|
16 |
model = LLM(
|
|
|
20 |
enforce_eager=True,
|
21 |
)
|
22 |
|
23 |
+
class Info(BaseModel):
|
24 |
+
name: str
|
25 |
+
age: int
|
26 |
+
|
27 |
+
json_schema = Info.model_json_schema()
|
28 |
+
guided_decoding_params = GuidedDecodingParams(json=json_schema)
|
29 |
+
sampling_params = SamplingParams(
|
30 |
+
temperature=0.1,
|
31 |
+
max_tokens=2048,
|
32 |
+
guided_decoding=guided_decoding_params,
|
33 |
+
)
|
34 |
+
|
35 |
+
prompt = "You are a helpful assistant."
|
36 |
+
|
37 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
38 |
+
model_name,
|
39 |
+
padding_side='right',
|
40 |
+
trust_remote_code=True,
|
41 |
+
)
|
42 |
+
|
43 |
+
if tokenizer.pad_token is None:
|
44 |
+
tokenizer.add_special_tokens({'pad_token': '<pad>'})
|
45 |
+
|
46 |
+
@spaces.GPU(duration=60)
|
47 |
+
def summarize(text):
|
48 |
+
if not text.strip():
|
49 |
+
return "Please enter some text to summarize."
|
50 |
+
|
51 |
+
messages = [
|
52 |
+
{"role": "system", "content": prompt},
|
53 |
+
{"role": "user", "content": text},
|
54 |
+
]
|
55 |
+
|
56 |
+
input_text = tokenizer.apply_chat_template(
|
57 |
+
messages,
|
58 |
+
tokenize=False,
|
59 |
+
add_generation_prompt=True,
|
60 |
+
enable_thinking=False,
|
61 |
+
)
|
62 |
+
|
63 |
+
outputs = model.generate([input_text], sampling_params)
|
64 |
+
prediction = outputs[0].outputs[0].text
|
65 |
+
return prediction
|
66 |
+
|
67 |
with gr.Blocks() as demo:
|
68 |
+
gr.Markdown("## 📝 Summarization for News, SciTLDR and Dialog Texts")
|
69 |
+
|
70 |
+
with gr.Row():
|
71 |
+
input_text = gr.Textbox(
|
72 |
+
label="Input Text",
|
73 |
+
autoscroll=False,
|
74 |
+
lines=15,
|
75 |
+
max_lines=15,
|
76 |
+
placeholder="Paste your article or paragraph here...",
|
77 |
+
)
|
78 |
+
output_text = gr.Textbox(
|
79 |
+
label="Summary",
|
80 |
+
autoscroll=False,
|
81 |
+
lines=15,
|
82 |
+
max_lines=15,
|
83 |
+
show_copy_button=True,
|
84 |
+
)
|
85 |
+
|
86 |
+
with gr.Row():
|
87 |
+
summarize_btn = gr.Button("Summarize")
|
88 |
+
summarize_btn.click(
|
89 |
+
fn=summarize,
|
90 |
+
inputs=input_text,
|
91 |
+
outputs=output_text,
|
92 |
+
show_progress=True,
|
93 |
+
)
|
94 |
|
95 |
if __name__ == "__main__":
|
96 |
demo.launch()
|