Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
!pip install transformers datasets evaluate seqeval pipeline gradio typing-extensions
|
2 |
+
import datasets
|
3 |
+
from transformers import pipeline
|
4 |
+
from transformers.pipelines.pt_utils import KeyDataset
|
5 |
+
from tqdm.auto import tqdm
|
6 |
+
|
7 |
+
pipe = pipe = pipeline("token-classification", model="erdometo/xlm-roberta-base-finetuned-TQuad2")
|
8 |
+
dataset = datasets.load_dataset("superb", name="asr", split="test")
|
9 |
+
|
10 |
+
for out in tqdm(pipe(KeyDataset(dataset, "file"))):
|
11 |
+
print(out)
|
12 |
+
|
13 |
+
|
14 |
+
|
15 |
+
import gradio as gr
|
16 |
+
from transformers import pipeline, AutoModelForQuestionAnswering, AutoTokenizer, AutoModelForTokenClassification
|
17 |
+
|
18 |
+
# Load your custom model and tokenizer
|
19 |
+
qa_model_name = "erdometo/xlm-roberta-base-finetuned-TQuad2"
|
20 |
+
token_classification_model_name = "FacebookAI/xlm-roberta-large-finetuned-conll03-german"
|
21 |
+
|
22 |
+
qa_model = AutoModelForQuestionAnswering.from_pretrained(qa_model_name)
|
23 |
+
qa_tokenizer = AutoTokenizer.from_pretrained(qa_model_name)
|
24 |
+
|
25 |
+
token_classification_model = AutoModelForTokenClassification.from_pretrained(token_classification_model_name)
|
26 |
+
token_classification_tokenizer = AutoTokenizer.from_pretrained(token_classification_model_name)
|
27 |
+
# Define a function for inference based on pipeline type
|
28 |
+
def predict(pipeline_type, question, context):
|
29 |
+
if pipeline_type == "question-answering":
|
30 |
+
qa_pipeline = pipeline("question-answering", model=qa_model, tokenizer=qa_tokenizer)
|
31 |
+
result = qa_pipeline(question=question, context=context)
|
32 |
+
response = [(result['answer'], result['score'])]
|
33 |
+
return response
|
34 |
+
elif pipeline_type == "token-classification":
|
35 |
+
token_classification_pipeline = pipeline("token-classification", model=token_classification_model, tokenizer=token_classification_tokenizer)
|
36 |
+
result = token_classification_pipeline(context)
|
37 |
+
highlighted_text = {"text": context, "entities": result}
|
38 |
+
return gr.HighlightedText(highlighted_text)
|
39 |
+
|
40 |
+
# Create a Gradio Interface with dropdown and two text inputs
|
41 |
+
iface = gr.Interface(
|
42 |
+
fn=predict,
|
43 |
+
inputs=[
|
44 |
+
gr.Dropdown(choices=["question-answering", "token-classification"], label="Choose Pipeline"),
|
45 |
+
"text",
|
46 |
+
"text"
|
47 |
+
],
|
48 |
+
outputs=gr.Highlight()
|
49 |
+
)
|
50 |
+
|
51 |
+
# Launch the interface
|
52 |
+
iface.launch()
|