Commit
Β·
06e9286
1
Parent(s):
b251121
first version of the app
Browse files
app.py
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 3 |
+
|
| 4 |
+
tokenizer = AutoTokenizer.from_pretrained("impresso-project/nel-hipe-multilingual")
|
| 5 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(
|
| 6 |
+
"impresso-project/nel-hipe-multilingual"
|
| 7 |
+
).eval()
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def disambiguate_sentences(sentences):
|
| 11 |
+
results = []
|
| 12 |
+
for sentence in sentences:
|
| 13 |
+
outputs = model.generate(
|
| 14 |
+
**tokenizer([sentence], return_tensors="pt"),
|
| 15 |
+
num_beams=5,
|
| 16 |
+
num_return_sequences=5
|
| 17 |
+
)
|
| 18 |
+
decoded = tokenizer.batch_decode(outputs, skip_special_tokens=True)
|
| 19 |
+
results.append(decoded)
|
| 20 |
+
return results
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
input_sentences = gr.inputs.Textbox(
|
| 24 |
+
lines=5,
|
| 25 |
+
label="Input Sentences",
|
| 26 |
+
placeholder="Enter your sentence here in the following format: \\ `It is reported in [START] Paris [END], "
|
| 27 |
+
"that the opening of the chambers will take place on the 27th January.' \\ "
|
| 28 |
+
"This format ensures that the model knows which entities to disambiguate, more exactly the entity should "
|
| 29 |
+
"be surrounded by `[START]` and `[END]`.",
|
| 30 |
+
)
|
| 31 |
+
output_predictions = gr.outputs.Textbox(label="Predictions")
|
| 32 |
+
|
| 33 |
+
gr.Interface(
|
| 34 |
+
fn=disambiguate_sentences,
|
| 35 |
+
inputs=input_sentences,
|
| 36 |
+
outputs=output_predictions,
|
| 37 |
+
title="NEL Disambiguation",
|
| 38 |
+
).launch()
|
x
DELETED
|
File without changes
|