Create app.py

app.py

@@ -0,0 +1,56 @@
+# import packages
+__import__('pysqlite3')
+import sys
+sys.modules['sqlite3'] = sys.modules.pop('pysqlite3')
+from sentence_transformers import SentenceTransformer
+import chromadb
+from datasets import load_dataset
+from gpt4all import GPT4All
+
+# Embedding vector
+class VectorStore:
+    def __init__(self, collection_name):
+       # Initialize the embedding model
+        self.embedding_model = SentenceTransformer('sentence-transformers/multi-qa-MiniLM-L6-cos-v1')
+        self.chroma_client = chromadb.Client()
+        self.collection = self.chroma_client.create_collection(name=collection_name)
+
+    # Method to populate the vector store with embeddings from a dataset
+    def populate_vectors(self, dataset):
+        # Select the text columns to concatenate
+        title = dataset['train']['title_cleaned'][:5000]  # Limiting to 100 examples for the demo
+        recipe = dataset['train']['recipe_new'][:5000]
+        meal_type = dataset['train']['meal_type'][:5000]
+        allergy = dataset['train']['allergy_type'][:5000]
+        ingredients_alternative = dataset['train']['ingredients_alternatives'][:5000]
+
+        # Concatenate the text from both columns
+        texts = [f"{tit} {rep} {meal} {alle} {ingr} " for tit, rep, meal,alle, ingr in zip(title,recipe,meal_type,allergy,ingredients_alternative)]
+        for i, item in enumerate(texts):
+            embeddings = self.embedding_model.encode(item).tolist()
+            self.collection.add(embeddings=[embeddings], documents=[item], ids=[str(i)])
+
+    # # Method to search the ChromaDB collection for relevant context based on a query
+    def search_context(self, query, n_results=1):
+        query_embeddings = self.embedding_model.encode(query).tolist()
+        return self.collection.query(query_embeddings=query_embeddings, n_results=n_results)
+
+
+# importing dataset hosted on huggingface
+# dataset details - https://huggingface.co/datasets/Thefoodprocessor/recipe_new_with_features_full
+dataset = load_dataset('Thefoodprocessor/recipe_new_with_features_full')
+
+# create a vector embedding
+vector_store = VectorStore("embedding_vector")
+vector_store.populate_vectors(dataset)
+
+
+# loading gpt4all language model
+# load model Chat based model mistral-7b-openorca.gguf2.Q4_0.gguf
+# detail about gpt4all and model information - https://gpt4all.io/index.html
+model_name = 'Meta-Llama-3-8B-Instruct.Q4_0.gguf' # .gguf represents quantized model
+model_path = "gpt4all"
+# add path to download load the model locally, download once and load for subsequent inference
+model = GPT4All(model_name=model_name, model_path=model_path,device="cuda")
+
+