Premier commit

Requirements.txt

@@ -0,0 +1,9 @@
+transformers 
+datasets 
+huggingface_hub
+flask==2.2.5
+requests
+tensorflow
+streamlit
+tf-keras
+pymupdf

app.py

@@ -0,0 +1,26 @@
+from flask import Flask, request
+from transformers import pipeline
+app = Flask(__name__)
+question_answerer = None
+
+@app.before_first_request
+def load_pipeline():
+    global question_answerer
+    question_answerer = pipeline("question-answering", "cancerfarore/bert-base-uncased-CancerFarore-Model", framework="tf")
+
+@app.route("/answer", methods=["POST"])
+def answer():
+    global question_answerer
+    obj = request.get_json()
+    context = obj['context']
+    question = obj['prompt']
+    return {"reponse" : question_answerer(context=context, question=question)['answer'], "score" : question_answerer(context=context, question=question)['score']}
+
+@app.route("/load_model", methods=["POST"])
+def load_model():
+    global question_answerer
+    obj = request.get_json()
+    model_name = obj['model']
+    question_answerer = pipeline("question-answering", model_name, framework="tf")
+    return f"Model {model_name}", 200
+

app_streamlit.py

@@ -0,0 +1,112 @@
+import streamlit as st
+import fitz
+import requests
+import shelve
+from config import *
+
+
+def load_chat_history():
+    with shelve.open("Message_history/ARDB") as db:
+        return db.get("messages", [])
+
+def load_current_model():
+    with shelve.open("Message_history/ARDB") as db:
+        return db.get("current_model", "BERT")
+
+def save_current_model(model_name):
+    with shelve.open("Message_history/ARDB") as db:
+        db["current_model"] = model_name
+
+
+def save_chat_history(messages):
+    with shelve.open("Message_history/ARDB") as db:
+        db["messages"] = messages
+
+
+if "messages" not in st.session_state:
+    st.session_state.messages = load_chat_history()
+
+if "current_model" not in st.session_state:
+    st.session_state.current_model = load_current_model()
+if "current_context" not in st.session_state:
+    st.session_state.current_context = ""
+
+st.title(TITRE + " - " + st.session_state.current_model)
+
+with st.sidebar:
+    st.title(body="Veuillez choisir un modèle")
+    option = st.selectbox(
+        label='Veuillez choisir un modèle',
+        index=MODEL_NAMES.index(st.session_state.current_model),
+        options=MODEL_NAMES)
+
+    st.write('Vous avez selectionné :', option)
+
+    if st.button("Changer le modèle"):
+        if option == st.session_state.current_model:
+            st.warning("🤖 Ce modèle est déjà chargé", icon="⚠️")
+        else:
+            req = requests.post(API_URL + "/load_model", json={'model': AVAILABLE_MODELS[option]})
+            if req.status_code == 200:
+                save_current_model(option)
+                st.session_state.current_model = load_current_model()
+                st.session_state.messages = []
+                save_chat_history([])
+                st.rerun()        
+            else:
+                st.warning("🤖 Je ne peux pas charger le modèle veuillez ressayer", icon="⚠️")
+
+    st.title(body="Veuillez donner un contexte")
+    upload_file = st.file_uploader("Importer un fichier texte ou pdf pour charger le contexte", type=['txt','pdf'])
+    if st.button("Changer le contexte"):
+        if upload_file:
+            if upload_file.type == 'text/plain':
+                st.session_state.current_context = upload_file.getvalue().decode("utf-8")
+            elif upload_file.type == 'application/pdf':
+                document = fitz.open(stream=upload_file.read())
+                text = ''
+                for page in document:
+                    page_text = page.get_text()
+                    page_text = page_text.replace('\n', ' ')
+                    text += page_text
+                st.session_state.current_context = text  
+            st.rerun()
+        else:
+            st.warning("🤖 Il n'y a aucun contexte à charger", icon="⚠️")
+
+    context = st.sidebar.text_area(value=st.session_state.current_context, label = "Ou écrire le contexte dans ce champs", placeholder = "Veuillez écrire le contexte de la question ici", height=300)
+    if st.button("Effacer le contexte"):
+        st.session_state.current_context = ""
+    if st.button("Supprimer l'historique"):
+        st.session_state.messages = []
+        save_chat_history([])
+
+# Display chat messages
+for message in st.session_state.messages:
+    avatar = USER_AVATAR if message["role"] == "user" else BOT_AVATAR
+    with st.chat_message(message["role"], avatar=avatar):
+        st.markdown(message["prompt"])
+
+prompt = st.chat_input("How can I help?")
+if prompt:
+    if len(context) > 20:
+        st.session_state.current_context = context
+        promptBody = {"role": "user", "prompt": prompt, "context": context}
+        req = requests.post(API_URL + "/answer", json = promptBody)
+        if req.status_code == 200:
+            st.session_state.messages.append(promptBody)
+            with st.chat_message("user", avatar=USER_AVATAR):
+                st.markdown(prompt)
+            with st.chat_message("assistant", avatar=BOT_AVATAR):
+                message_placeholder = st.empty()
+                res = req.json()
+                response = res['reponse']
+                message_placeholder.markdown(response)
+            st.session_state.messages.append({"role": "assistant", "prompt": response})
+        else:
+            st.warning("🤖 Je ne peux pas répondre à la question car il y a eu un problème avec l'API veuillez reposer votre question", icon="⚠️")
+    else:
+        st.warning("🤖 Je ne peux pas répondre à la question car il n'y a pas contexte ou ce dernier est trop court. Veuillez importer un contexte ou le coller dans son champ", icon="⚠️")
+        
+
+save_chat_history(st.session_state.messages)

config.py

@@ -0,0 +1,16 @@
+AVAILABLE_MODELS = {
+    "BERT" : 'cancerfarore/bert-base-uncased-CancerFarore-Model',
+    "ALBERT" : "cancerfarore/albert-base-v2-CancerFarore-Model",
+    "ROBERTA" : "cancerfarore/roberta-base-CancerFarore-Modela",
+    "PACIFISTA" : "cancerfarore/roberta-base-CancerFarore-Model"
+}
+
+MODEL_NAMES = []
+for NAME in AVAILABLE_MODELS:
+    MODEL_NAMES.append(NAME)
+
+TITRE  = "Pacifista 🐻"
+
+API_URL = 'http://127.0.0.1:5000'
+USER_AVATAR = "👤"
+BOT_AVATAR = "🤖"