Move binary files to Git LFS properly

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.index filter=lfs diff=lfs merge=lfs -text

app.py

@@ -83,65 +83,31 @@ import faiss
 import pickle
 import numpy as np
 import torch
-import gdown
-from transformers import T5Tokenizer, T5ForConditionalGeneration
-# from sentence_transformers import SentenceTransformer
-
-# Function to download a full folder from Google Drive
-def download_folder_from_google_drive(folder_url, output_path):
-    if not os.path.exists(output_path):
-        gdown.download_folder(url=folder_url, output=output_path, quiet=False, use_cookies=False)
-
-# Download individual files
-def download_file_from_google_drive(file_id, destination):
-    if not os.path.exists(destination):
-        url = f"https://drive.google.com/uc?id={file_id}"
-        gdown.download(url, destination, quiet=False)
-
-# Setup models and files
-@st.cache_resource
-def setup_files():
-    os.makedirs("models/embedding_model", exist_ok=True)
-    os.makedirs("models/generator_model", exist_ok=True)
-    os.makedirs("models/files", exist_ok=True)
-
-    # Download embedding model (folder)
-    download_folder_from_google_drive(
-        "https://drive.google.com/drive/folders/1GzPk2ehr7rzOr65Am1Hg3A87FOTNHLAM?usp=sharing",
-        "models/embedding_model"
-    )
-
-    # Download generator model (folder)
-    download_folder_from_google_drive(
-        "https://drive.google.com/drive/folders/1338KWiBE-6sWsTO2iH7Pgu8eRI7EE7Vr?usp=sharing",
-        "models/generator_model"
-    )
+from transformers import AutoTokenizer, AutoModel, T5Tokenizer, T5ForConditionalGeneration
 
-    # Download FAISS index, texts.pkl, embeddings.npy
-    download_file_from_google_drive("11J_VI1buTgnvhoP3z2HM6X5aPzbBO2ed", "models/files/faiss_index_file.index")
-    download_file_from_google_drive("1RTEwp8xDgxLnRUiy7ClTskFuTu0GtWBT", "models/files/texts.pkl")
-    download_file_from_google_drive("1N54imsqJIJGeqM3buiRzp1ivK_BtC7rR", "models/files/embeddings.npy")
+# Paths (everything is local now)
+FAISS_INDEX_PATH = "faiss_index_file.index"
+TEXTS_PATH = "texts.pkl"
+EMBEDDINGS_PATH = "embeddings.npy"
 
-# Paths
-EMBEDDING_MODEL_PATH = "models/embedding_model"
-GENERATOR_MODEL_PATH = "models/generator_model"
-FAISS_INDEX_PATH = "models/files/faiss_index_file.index"
-TEXTS_PATH = "models/files/texts.pkl"
-EMBEDDINGS_PATH = "models/files/embeddings.npy"
+EMBEDDING_MODEL_NAME = "Ah1111/Embedding_Model"
+GENERATOR_MODEL_NAME = "Ah1111/Generator_Model"
 
-# Load LLM model (Generator model)
+# Load generator model (T5)
 @st.cache_resource
 def load_llm():
-    tokenizer = T5Tokenizer.from_pretrained(GENERATOR_MODEL_PATH)
-    model = T5ForConditionalGeneration.from_pretrained(GENERATOR_MODEL_PATH)
+    tokenizer = T5Tokenizer.from_pretrained(GENERATOR_MODEL_NAME)
+    model = T5ForConditionalGeneration.from_pretrained(GENERATOR_MODEL_NAME)
     return tokenizer, model
 
-# Load embedding model
+# Load embedding model (custom Hugging Face model)
 @st.cache_resource
 def load_embedding_model():
-    return SentenceTransformer(EMBEDDING_MODEL_PATH)
+    tokenizer = AutoTokenizer.from_pretrained(EMBEDDING_MODEL_NAME)
+    model = AutoModel.from_pretrained(EMBEDDING_MODEL_NAME)
+    return tokenizer, model
 
-# Load FAISS index and embeddings
+# Load FAISS index and texts
 @st.cache_resource
 def load_faiss():
     faiss_index = faiss.read_index(FAISS_INDEX_PATH)
@@ -150,17 +116,24 @@ def load_faiss():
     embeddings = np.load(EMBEDDINGS_PATH, allow_pickle=True)
     return faiss_index, data, embeddings
 
+# Function to encode query using the embedding model
+def encode_query(query, tokenizer, model):
+    inputs = tokenizer(query, return_tensors="pt", truncation=True, padding=True)
+    with torch.no_grad():
+        embeddings = model(**inputs).last_hidden_state.mean(dim=1)
+    return embeddings.cpu().numpy()
+
 # Search top-k contexts
-def search(query, embed_model, index, data, k=5):
-    query_embedding = embed_model.encode([query]).astype('float32')
+def search(query, tokenizer, model, index, data, k=5):
+    query_embedding = encode_query(query, tokenizer, model).astype('float32')
     _, I = index.search(query_embedding, k)
     results = [data[i] for i in I[0] if i != -1]
     return results
 
-# Generate response
+# Generate response using generator model
 def generate_response(context, query, tokenizer, model):
     input_text = f"Context: {context}\n\nQuestion: {query}\n\nAnswer:"
-    inputs = tokenizer.encode(input_text, return_tensors="pt")
+    inputs = tokenizer.encode(input_text, return_tensors="pt", truncation=True)
     outputs = model.generate(inputs, max_length=512, do_sample=True, temperature=0.7)
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return response
@@ -177,19 +150,18 @@ def main():
         """
     )
 
-    # Download + Load everything
-    setup_files()
-    tokenizer, llm_model = load_llm()
-    embed_model = load_embedding_model()
+    # Load models and files
+    embed_tokenizer, embed_model = load_embedding_model()
+    gen_tokenizer, gen_model = load_llm()
     faiss_index, data, embeddings = load_faiss()
 
     query = st.text_input("💬 Your Question:")
 
     if query:
         with st.spinner("🔍 Retrieving and Generating..."):
-            contexts = search(query, embed_model, faiss_index, data)
+            contexts = search(query, embed_tokenizer, embed_model, faiss_index, data)
             combined_context = " ".join(contexts)
-            response = generate_response(combined_context, query, tokenizer, llm_model)
+            response = generate_response(combined_context, query, gen_tokenizer, gen_model)
 
             st.success("✅ Answer Ready!")
             st.subheader("📄 Response:")

embeddings_file.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a139ec8e59563899d337ae2728894067f6ddd85c605b2ac93d6e4183d047979
+size 3038336

faiss_index_file.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62ca8fec53b892d868c44273411851a30427ea16e106efb4224fcff3e343d52b
+size 3038253

texts.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d046912f1311f941915f5f03b84bd502c15be19a6f7058ba82a5ebe9b44ff392
+size 2823783