Update frontend.py
Browse files- frontend.py +86 -30
frontend.py
CHANGED
|
@@ -1,43 +1,37 @@
|
|
| 1 |
-
# frontend.py
|
| 2 |
-
import os
|
| 3 |
-
import requests
|
| 4 |
import streamlit as st
|
|
|
|
|
|
|
| 5 |
|
| 6 |
-
|
| 7 |
|
| 8 |
st.set_page_config(page_title="StackOverflow Tagger", layout="wide")
|
| 9 |
st.title("🔖 StackOverflow Tag Predictor")
|
| 10 |
|
| 11 |
-
st.
|
| 12 |
-
"Entrez une question (titre + éventuellement description) "
|
| 13 |
-
"et récupérez les probabilités des tags StackOverflow."
|
| 14 |
-
)
|
| 15 |
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
)
|
| 21 |
|
| 22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
|
|
|
| 29 |
with st.spinner("Prédiction en cours..."):
|
| 30 |
-
|
| 31 |
-
f"{API_URL}/predict",
|
| 32 |
-
json={"text": question, "top_k": top_k},
|
| 33 |
-
timeout=60,
|
| 34 |
-
)
|
| 35 |
-
resp.raise_for_status()
|
| 36 |
-
data = resp.json()
|
| 37 |
-
tags = data.get("tags", [])
|
| 38 |
|
| 39 |
if not tags:
|
| 40 |
-
st.warning("
|
| 41 |
else:
|
| 42 |
st.subheader("Résultats")
|
| 43 |
for t in tags:
|
|
@@ -47,5 +41,67 @@ if st.button("Prédire"):
|
|
| 47 |
scores = {t["label"]: t["score"] for t in tags}
|
| 48 |
st.bar_chart(scores)
|
| 49 |
|
| 50 |
-
|
| 51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
+
import pandas as pd
|
| 3 |
+
from io import StringIO
|
| 4 |
|
| 5 |
+
from model_utils import predict_proba
|
| 6 |
|
| 7 |
st.set_page_config(page_title="StackOverflow Tagger", layout="wide")
|
| 8 |
st.title("🔖 StackOverflow Tag Predictor")
|
| 9 |
|
| 10 |
+
tab_single, tab_csv = st.tabs(["Question unique", "CSV batch"])
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
+
with tab_single:
|
| 13 |
+
st.write(
|
| 14 |
+
"Entrez une question (titre + éventuellement description) "
|
| 15 |
+
"et récupérez les probabilités des tags StackOverflow prédits par le modèle."
|
| 16 |
+
)
|
| 17 |
|
| 18 |
+
question = st.text_area(
|
| 19 |
+
"Question StackOverflow",
|
| 20 |
+
height=200,
|
| 21 |
+
placeholder="Ex: How to fine-tune BERT for multi-label classification?",
|
| 22 |
+
)
|
| 23 |
|
| 24 |
+
top_k = st.slider("Nombre de tags à afficher (top_k)", 1, 20, 5, key="topk_single")
|
| 25 |
+
|
| 26 |
+
if st.button("Prédire", key="predict_single"):
|
| 27 |
+
if not question.strip():
|
| 28 |
+
st.warning("Merci d'entrer une question.")
|
| 29 |
+
else:
|
| 30 |
with st.spinner("Prédiction en cours..."):
|
| 31 |
+
tags = predict_proba(question, top_k=top_k)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
if not tags:
|
| 34 |
+
st.warning("Pas de tags prédits.")
|
| 35 |
else:
|
| 36 |
st.subheader("Résultats")
|
| 37 |
for t in tags:
|
|
|
|
| 41 |
scores = {t["label"]: t["score"] for t in tags}
|
| 42 |
st.bar_chart(scores)
|
| 43 |
|
| 44 |
+
with tab_csv:
|
| 45 |
+
st.write(
|
| 46 |
+
"Uploade un fichier CSV contenant des questions. "
|
| 47 |
+
"On ajoutera une colonne avec le tag principal prédit pour chaque ligne."
|
| 48 |
+
)
|
| 49 |
+
|
| 50 |
+
uploaded_file = st.file_uploader("Choisir un fichier CSV", type=["csv"])
|
| 51 |
+
|
| 52 |
+
if uploaded_file is not None:
|
| 53 |
+
df = pd.read_csv(uploaded_file)
|
| 54 |
+
|
| 55 |
+
st.write("Aperçu du CSV :")
|
| 56 |
+
st.dataframe(df.head())
|
| 57 |
+
|
| 58 |
+
text_column = st.selectbox(
|
| 59 |
+
"Colonne contenant la question",
|
| 60 |
+
options=list(df.columns),
|
| 61 |
+
)
|
| 62 |
+
|
| 63 |
+
top_k_batch = st.slider(
|
| 64 |
+
"Nombre de tags à considérer pour le batch (pour choisir le meilleur)",
|
| 65 |
+
1,
|
| 66 |
+
20,
|
| 67 |
+
5,
|
| 68 |
+
key="topk_batch",
|
| 69 |
+
)
|
| 70 |
+
|
| 71 |
+
if st.button("Lancer la prédiction sur le CSV"):
|
| 72 |
+
if df[text_column].isnull().all():
|
| 73 |
+
st.error("La colonne choisie ne contient pas de texte.")
|
| 74 |
+
else:
|
| 75 |
+
preds = []
|
| 76 |
+
with st.spinner("Prédiction en cours sur le CSV..."):
|
| 77 |
+
for text in df[text_column].fillna(""):
|
| 78 |
+
if not str(text).strip():
|
| 79 |
+
preds.append({"best_tag": None, "best_score": None})
|
| 80 |
+
continue
|
| 81 |
+
tags = predict_proba(str(text), top_k=top_k_batch)
|
| 82 |
+
if len(tags) == 0:
|
| 83 |
+
preds.append({"best_tag": None, "best_score": None})
|
| 84 |
+
else:
|
| 85 |
+
best = tags[0]
|
| 86 |
+
preds.append(
|
| 87 |
+
{"best_tag": best["label"], "best_score": best["score"]}
|
| 88 |
+
)
|
| 89 |
+
|
| 90 |
+
df["predicted_tag"] = [p["best_tag"] for p in preds]
|
| 91 |
+
df["predicted_score"] = [p["best_score"] for p in preds]
|
| 92 |
+
|
| 93 |
+
st.subheader("Résultats enrichis")
|
| 94 |
+
st.dataframe(df.head())
|
| 95 |
+
|
| 96 |
+
csv_buffer = StringIO()
|
| 97 |
+
df.to_csv(csv_buffer, index=False)
|
| 98 |
+
csv_bytes = csv_buffer.getvalue().encode("utf-8")
|
| 99 |
+
|
| 100 |
+
st.download_button(
|
| 101 |
+
label="📥 Télécharger le CSV avec tags prédits",
|
| 102 |
+
data=csv_bytes,
|
| 103 |
+
file_name="questions_with_tags.csv",
|
| 104 |
+
mime="text/csv",
|
| 105 |
+
)
|
| 106 |
+
else:
|
| 107 |
+
st.info("Uploade un fich
|