Upload folder using huggingface_hub

app.py

@@ -1,9 +1,11 @@
 import os;
 import cohere;
-import numpy as np;
-from dotenv import load_dotenv;
 import chromadb;
+import numpy as np;
 import gradio as gr;
+from dotenv import load_dotenv;
+from chromadb.config import Settings
+
 
 load_dotenv()
 
@@ -13,213 +15,207 @@ co = cohere.Client( CO_API_KEY );
 EMBED_MODEL = "embed-v4.0"
 CHAT_MODEL = "command-r-08-2024"
 
+PERSIST_DIR = "/mnt/data/chromadb_storage"
+os.makedirs(PERSIST_DIR, exist_ok=True)
+
+client = chromadb.Client( Settings( chroma_db_impl="duckdb+parquet", persist_directory=PERSIST_DIR ) );
 
-client = chromadb.Client();
 collection = client.get_or_create_collection( name="brampton" );
 
-brampton_docs = [
-    # Shopping Centres
-    "Bramalea City Centre is one of Canada’s largest malls, with 1.5 million sq ft of retail space and over 300 stores.",
-    "Trinity Common Mall is a large outdoor shopping centre spanning over 850,000 sq ft with 69 retail stores, services, and restaurants at Hwy 410 & Bovaird Dr.",
-    "Shoppers World Brampton is a community shopping destination featuring over 180 shops and services at 499 Main St S, Suite 56, L6Y 1N7.",
-    "Brampton Mall houses 38 brand‑name stores at 152–160 Main St S, L6W 2C9, with services, restaurants, and customer amenities.",
-    # Entertainment & Recreation
-    "Toprock Climbing is Brampton’s first rock‑climbing gym (opened 2017), a community hub to meet new friends, challenge yourself, and stay fit.",
-    "Glow Zone 360 offers glow‑in‑the‑dark mini‑golf, laser tag, and arcade games at 9446 McLaughlin Rd N, Unit 7.",
-    "The CAA Centre is Brampton’s indoor sports and entertainment complex, hosting hockey games, trade shows, and community events at 7575 Kennedy Rd.",
-    # Parks & Outdoors
-    "Gage Park is an urban park at 45 Main St S offering strolling paths, picnic areas, flower gardens, free evening concerts, and a seasonal ice rink.",
-    "Professor’s Lake Recreation Centre features a 65‑acre lake with sandy beach, paddleboat/canoe/kayak rentals, walking trails, and fishing spots.",
-    "Eldorado Park is a scenic riverside park on the Credit River with picnic areas, walking trails, and a small lake for outdoor recreation.",
-    "Chinguacousy Park features a water park, splash pads, mini golf, ski hills, and biking trails at 9050 Bramalea Rd.",
-    "Heart Lake Conservation Park offers hiking, fishing, treetop trek canopy tours, and skiing at 11073 Heart Lake Rd.",
-    "Loafer’s Lake Recreation Centre is a community facility with ice rinks, outdoor pool, basketball courts, and scenic walking paths.",
-    "Claireville Conservation Area spans over 2,000 acres of wildlife habitat along the Humber River, with trails, picnic areas, and off‑road biking.",
-    # Culture & History
-    "Peel Art Gallery, Museum and Archives (PAMA) is a museum, art gallery, and archives for the Peel Region located at 9 Wellington St E in Brampton :contentReference[oaicite:3]{index=3}.",
-    "Rose Theatre Brampton is a municipally‑owned performing arts centre with an 868‑seat main hall and a smaller studio venue at 1 Theatre Ln :contentReference[oaicite:4]{index=4}.",
-    "Historic Bovaird House is a restored 1860s heritage farmhouse offering guided tours and seasonal events at 10 Chinguacousy Rd.",
-    "Historic Churchville Village is a preserved 19th‑century rural settlement, featuring St. James Anglican Church and original pioneer buildings.",
-    # Transit & Infrastructure
-    "Brampton GO Station connects downtown Brampton to Toronto via the GO Transit rail network at 27 Railway St.",
-    "Brampton Gateway Terminal is a major Brampton Transit bus hub offering city‑wide connections at Steeles Ave E & Main St.",
-    # Education & Health
-    "University of Guelph‑Humber is a joint campus offering undergraduate degrees in business, health, and arts at 207 Humber College Blvd.",
-    "Sheridan College Davis Campus offers programs in applied health, community studies, and media at 7899 McLaughlin Rd S.",
-    "Peel Memorial Centre for Integrated Health and Wellness provides hospital and outpatient services at 50 Queen St E.",
-    # Events & Festivals
-    "Carabram is an annual multicultural festival every July at Garden Square, showcasing pavilions, food, and performances from around the world.",
-    "The Brampton Santa Claus Parade, held every November along Queen St, features colourful floats, marching bands, and a visit from Santa.",
-    # Neighbourhoods & Demographics
-    "Springdale is a northern suburban district nicknamed “Singhdale” for its large Sikh Canadian community, bounded by Bovaird Dr, Heart Lake Rd, and Airport Rd :contentReference[oaicite:5]{index=5}.",
-
-    # General City Facts
-    "Official Name: City of Brampton.",
-    "Settlement Type: lower-tier city.",
-    "Nickname: Flower City (formerly 'Flower Town of Canada').",
-    "Part of the Greater Toronto Area.",
-    "Regional seat of the Regional Municipality of Peel.",
-    "Incorporated as a village in 1853.",
-    "Incorporated as a town in 1873.",
-    "Incorporated as a city in 1974.",
-    "Major highways: Highway 410 and Highway 407 ETR.",
-    # Geography & Climate
-    "Borders Vaughan to the east.",
-    "Borders Halton Hills to the west.",
-    "Borders Caledon to the north.",
-    "Borders Mississauga to the south.",
-    "Borders Etobicoke (Toronto) to the southeast.",
-    "Climate: Humid continental (Köppen Dfa).",
-    # Sister Cities
-    "Miami Beach, Florida.",
-    "Plano, Texas.",
-    "Ribeira Grande, Azores, Portugal.",
-    "Xuzhou, Jiangsu, China.",
-    "Brampton, Cumbria, England.",
-    "Marikina, Philippines.",
-    "Gapyeong, South Korea.",
-    # Neighbourhoods
-    "Alloa: rural former community in northwest Brampton.",
-    "Bramalea: Canada’s first planned satellite town.",
-    "Castlemore: suburban area with luxury homes in West Humber River valley.",
-    "Churchville: preserved 19th-century village on the Credit River.",
-    "Claireville: neighbourhood around Claireville Conservation Area.",
-    "Mount Pleasant: fast‑growing community since 2006 GO station.",
-    "Snelgrove: straddles Mayfield Road between Brampton and Caledon.",
-    "Springdale: 4,000‑acre northern district with large Sikh Canadian community.",
-    "Tullamore: small residential neighbourhood in Brampton.",
-    # History Highlights
-    "Originally inhabited by the Mississaugas of the Credit First Nation.",
-    "Treaty 19 (Ajetance Purchase) signed October 1818.",
-    "Village incorporation: 1853.",
-    "Town incorporation: 1873.",
-    "City status granted: 1974.",
-    "Named after Brampton, Cumberland (England).",
-    "Dale’s Nursery: once North America’s largest cut‑flower business.",
-    "Historic Bovaird House (1860s farmhouse) offers tours.",
-    "Historic Churchville Village preserves pioneer buildings.",
-    "Rose Theatre opened in September 2006.",
-    # Landmarks & Culture
-    "Peel Art Gallery, Museum and Archives (PAMA) at 9 Wellington St E.",
-    "Rose Theatre Brampton: 868‑seat main hall + studio theatre.",
-    "Heritage Theatre (former Capitol Theatre) preceded Rose Theatre.",
-    "Brampton City Hall at 2 Wellington St W; nine‑storey tower added 2014.",
-    "Garden Square: public plaza for concerts and events.",
-    "A. Grenville & W. Davis Courthouse at Hurontario & County Court St.",
-    # Shopping Centres
-    "Bramalea City Centre: 1.5 million sq ft, 300+ stores.",
-    "Trinity Common Mall: 850,000 sq ft open‑air centre, 69 retailers.",
-    "Shoppers World Brampton: 180+ shops at 499 Main St S.",
-    "Brampton Mall: 38 brand‑name stores at 152–160 Main St S.",
-    # Markets & Festivals
-    "Downtown Brampton Farmers’ Market at 30 Church St E (since 1859).",
-    "Carabram: annual multicultural festival every July.",
-    # Parks & Recreation
-    "Gage Park: oldest municipal park (est. 1903) with gardens, concerts.",
-    "Professor’s Lake Recreation Centre: 65‑acre lake, beach, rentals.",
-    "Eldorado Park: riverside park with playgrounds and pool on Credit River.",
-    "Chinguacousy Park: 100 acre green space with zoo, mini‑golf, ski hill.",
-    "Heart Lake Conservation Park: hiking, fishing, canopy tours.",
-    "Loafer’s Lake Recreation Centre: rinks, pool, courts, walking trails.",
-    "Claireville Conservation Area: 2,000 acre wildlife habitat.",
-    # Transit & Infrastructure
-    "Brampton GO Station connects via GO Transit rail (27 Railway St).",
-    "Mount Pleasant GO Station opened 2006 on Bradford line.",
-    "Lisgar GO Station serves western Brampton on Barrie line.",
-    "Bramalea GO Station hub at Fennell Blvd.",
-    "Brampton Gateway Terminal: central Brampton Transit bus hub.",
-    # Sports Teams
-    "Brampton Battalion: OHL team (1998–2013).",
-    "Brampton Beast: ECHL hockey (2013–2018).",
-    "Bramalea Blues: junior A ice hockey (OPJHL).",
-    "Bramalea Satellites: junior B hockey team.",
-    "Brampton City United FC: semi‑pro soccer club (founded 2007).",
-    "Brampton Honey Badgers: CEBL basketball champions (2022).",
-    "Brampton Excelsiors (MSL): Senior A box lacrosse (est. 1883).",
-    "Brampton Excelsiors Jr. A: OLA Junior A lacrosse league.",
-    "Brampton Excelsiors Jr. B: OJBLL box lacrosse expansion team (2012).",
-    # Education & Health
-    "University of Guelph–Humber: joint undergraduate campus in Brampton.",
-    "Sheridan College Davis Campus: applied health and media programs.",
-    "Peel Memorial Centre for Integrated Health and Wellness at 50 Queen St E.",
-    "William Osler Health System operates Brampton Civic Hospital on Bovaird Dr."
-]
-
-doc_embeddings = co.embed(
-    model=EMBED_MODEL,
-    texts=brampton_docs
-).embeddings;
-
-brampton_vector = np.mean( doc_embeddings, axis=0 );
-
-collection.add(
-    ids=[f"doc_{i}" for i in range(len(brampton_docs))],
-    documents=brampton_docs,
-    embeddings=doc_embeddings,
-    metadatas=[{"source": f"doc_{i}"} for i in range(len(brampton_docs))]
-)
+if len(collection.count()['count']) == 0:
+
+    brampton_docs = [
+        # Shopping Centres
+        "Bramalea City Centre is one of Canada’s largest malls, with 1.5 million sq ft of retail space and over 300 stores.",
+        "Trinity Common Mall is a large outdoor shopping centre spanning over 850,000 sq ft with 69 retail stores, services, and restaurants at Hwy 410 & Bovaird Dr.",
+        "Shoppers World Brampton is a community shopping destination featuring over 180 shops and services at 499 Main St S, Suite 56, L6Y 1N7.",
+        "Brampton Mall houses 38 brand‑name stores at 152–160 Main St S, L6W 2C9, with services, restaurants, and customer amenities.",
+        # Entertainment & Recreation
+        "Toprock Climbing is Brampton’s first rock‑climbing gym (opened 2017), a community hub to meet new friends, challenge yourself, and stay fit.",
+        "Glow Zone 360 offers glow‑in‑the‑dark mini‑golf, laser tag, and arcade games at 9446 McLaughlin Rd N, Unit 7.",
+        "The CAA Centre is Brampton’s indoor sports and entertainment complex, hosting hockey games, trade shows, and community events at 7575 Kennedy Rd.",
+        # Parks & Outdoors
+        "Gage Park is an urban park at 45 Main St S offering strolling paths, picnic areas, flower gardens, free evening concerts, and a seasonal ice rink.",
+        "Professor’s Lake Recreation Centre features a 65‑acre lake with sandy beach, paddleboat/canoe/kayak rentals, walking trails, and fishing spots.",
+        "Eldorado Park is a scenic riverside park on the Credit River with picnic areas, walking trails, and a small lake for outdoor recreation.",
+        "Chinguacousy Park features a water park, splash pads, mini golf, ski hills, and biking trails at 9050 Bramalea Rd.",
+        "Heart Lake Conservation Park offers hiking, fishing, treetop trek canopy tours, and skiing at 11073 Heart Lake Rd.",
+        "Loafer’s Lake Recreation Centre is a community facility with ice rinks, outdoor pool, basketball courts, and scenic walking paths.",
+        "Claireville Conservation Area spans over 2,000 acres of wildlife habitat along the Humber River, with trails, picnic areas, and off‑road biking.",
+        # Culture & History
+        "Peel Art Gallery, Museum and Archives (PAMA) is a museum, art gallery, and archives for the Peel Region located at 9 Wellington St E in Brampton :contentReference[oaicite:3]{index=3}.",
+        "Rose Theatre Brampton is a municipally‑owned performing arts centre with an 868‑seat main hall and a smaller studio venue at 1 Theatre Ln :contentReference[oaicite:4]{index=4}.",
+        "Historic Bovaird House is a restored 1860s heritage farmhouse offering guided tours and seasonal events at 10 Chinguacousy Rd.",
+        "Historic Churchville Village is a preserved 19th‑century rural settlement, featuring St. James Anglican Church and original pioneer buildings.",
+        # Transit & Infrastructure
+        "Brampton GO Station connects downtown Brampton to Toronto via the GO Transit rail network at 27 Railway St.",
+        "Brampton Gateway Terminal is a major Brampton Transit bus hub offering city‑wide connections at Steeles Ave E & Main St.",
+        # Education & Health
+        "University of Guelph‑Humber is a joint campus offering undergraduate degrees in business, health, and arts at 207 Humber College Blvd.",
+        "Sheridan College Davis Campus offers programs in applied health, community studies, and media at 7899 McLaughlin Rd S.",
+        "Peel Memorial Centre for Integrated Health and Wellness provides hospital and outpatient services at 50 Queen St E.",
+        # Events & Festivals
+        "Carabram is an annual multicultural festival every July at Garden Square, showcasing pavilions, food, and performances from around the world.",
+        "The Brampton Santa Claus Parade, held every November along Queen St, features colourful floats, marching bands, and a visit from Santa.",
+        # Neighbourhoods & Demographics
+        "Springdale is a northern suburban district nicknamed “Singhdale” for its large Sikh Canadian community, bounded by Bovaird Dr, Heart Lake Rd, and Airport Rd :contentReference[oaicite:5]{index=5}.",
+
+        # General City Facts
+        "Official Name: City of Brampton.",
+        "Settlement Type: lower-tier city.",
+        "Nickname: Flower City (formerly 'Flower Town of Canada').",
+        "Part of the Greater Toronto Area.",
+        "Regional seat of the Regional Municipality of Peel.",
+        "Incorporated as a village in 1853.",
+        "Incorporated as a town in 1873.",
+        "Incorporated as a city in 1974.",
+        "Major highways: Highway 410 and Highway 407 ETR.",
+        # Geography & Climate
+        "Borders Vaughan to the east.",
+        "Borders Halton Hills to the west.",
+        "Borders Caledon to the north.",
+        "Borders Mississauga to the south.",
+        "Borders Etobicoke (Toronto) to the southeast.",
+        "Climate: Humid continental (Köppen Dfa).",
+        # Sister Cities
+        "Miami Beach, Florida.",
+        "Plano, Texas.",
+        "Ribeira Grande, Azores, Portugal.",
+        "Xuzhou, Jiangsu, China.",
+        "Brampton, Cumbria, England.",
+        "Marikina, Philippines.",
+        "Gapyeong, South Korea.",
+        # Neighbourhoods
+        "Alloa: rural former community in northwest Brampton.",
+        "Bramalea: Canada’s first planned satellite town.",
+        "Castlemore: suburban area with luxury homes in West Humber River valley.",
+        "Churchville: preserved 19th-century village on the Credit River.",
+        "Claireville: neighbourhood around Claireville Conservation Area.",
+        "Mount Pleasant: fast‑growing community since 2006 GO station.",
+        "Snelgrove: straddles Mayfield Road between Brampton and Caledon.",
+        "Springdale: 4,000‑acre northern district with large Sikh Canadian community.",
+        "Tullamore: small residential neighbourhood in Brampton.",
+        # History Highlights
+        "Originally inhabited by the Mississaugas of the Credit First Nation.",
+        "Treaty 19 (Ajetance Purchase) signed October 1818.",
+        "Village incorporation: 1853.",
+        "Town incorporation: 1873.",
+        "City status granted: 1974.",
+        "Named after Brampton, Cumberland (England).",
+        "Dale’s Nursery: once North America’s largest cut‑flower business.",
+        "Historic Bovaird House (1860s farmhouse) offers tours.",
+        "Historic Churchville Village preserves pioneer buildings.",
+        "Rose Theatre opened in September 2006.",
+        # Landmarks & Culture
+        "Peel Art Gallery, Museum and Archives (PAMA) at 9 Wellington St E.",
+        "Rose Theatre Brampton: 868‑seat main hall + studio theatre.",
+        "Heritage Theatre (former Capitol Theatre) preceded Rose Theatre.",
+        "Brampton City Hall at 2 Wellington St W; nine‑storey tower added 2014.",
+        "Garden Square: public plaza for concerts and events.",
+        "A. Grenville & W. Davis Courthouse at Hurontario & County Court St.",
+        # Shopping Centres
+        "Bramalea City Centre: 1.5 million sq ft, 300+ stores.",
+        "Trinity Common Mall: 850,000 sq ft open‑air centre, 69 retailers.",
+        "Shoppers World Brampton: 180+ shops at 499 Main St S.",
+        "Brampton Mall: 38 brand‑name stores at 152–160 Main St S.",
+        # Markets & Festivals
+        "Downtown Brampton Farmers’ Market at 30 Church St E (since 1859).",
+        "Carabram: annual multicultural festival every July.",
+        # Parks & Recreation
+        "Gage Park: oldest municipal park (est. 1903) with gardens, concerts.",
+        "Professor’s Lake Recreation Centre: 65‑acre lake, beach, rentals.",
+        "Eldorado Park: riverside park with playgrounds and pool on Credit River.",
+        "Chinguacousy Park: 100 acre green space with zoo, mini‑golf, ski hill.",
+        "Heart Lake Conservation Park: hiking, fishing, canopy tours.",
+        "Loafer’s Lake Recreation Centre: rinks, pool, courts, walking trails.",
+        "Claireville Conservation Area: 2,000 acre wildlife habitat.",
+        # Transit & Infrastructure
+        "Brampton GO Station connects via GO Transit rail (27 Railway St).",
+        "Mount Pleasant GO Station opened 2006 on Bradford line.",
+        "Lisgar GO Station serves western Brampton on Barrie line.",
+        "Bramalea GO Station hub at Fennell Blvd.",
+        "Brampton Gateway Terminal: central Brampton Transit bus hub.",
+        # Sports Teams
+        "Brampton Battalion: OHL team (1998–2013).",
+        "Brampton Beast: ECHL hockey (2013–2018).",
+        "Bramalea Blues: junior A ice hockey (OPJHL).",
+        "Bramalea Satellites: junior B hockey team.",
+        "Brampton City United FC: semi‑pro soccer club (founded 2007).",
+        "Brampton Honey Badgers: CEBL basketball champions (2022).",
+        "Brampton Excelsiors (MSL): Senior A box lacrosse (est. 1883).",
+        "Brampton Excelsiors Jr. A: OLA Junior A lacrosse league.",
+        "Brampton Excelsiors Jr. B: OJBLL box lacrosse expansion team (2012).",
+        # Education & Health
+        "University of Guelph–Humber: joint undergraduate campus in Brampton.",
+        "Sheridan College Davis Campus: applied health and media programs.",
+        "Peel Memorial Centre for Integrated Health and Wellness at 50 Queen St E.",
+        "William Osler Health System operates Brampton Civic Hospital on Bovaird Dr."
+    ];
+
+    doc_embeddings = co.embed(
+        model=EMBED_MODEL,
+        texts=brampton_docs
+    ).embeddings;
+
+    brampton_vector = np.mean( doc_embeddings, axis=0 );
+
+    collection.add(
+        ids=[f"doc_{i}" for i in range( len( brampton_docs ) )],
+        documents=brampton_docs,
+        embeddings=doc_embeddings,
+        metadatas=[{"source": f"doc_{i}"} for i in range( len( brampton_docs ) )]
+    );
+
+    client.persist();
 
 SYSTEM_INSTR = (
     "You are a Brampton-focused assistant. "
     "Only discuss topics related to Brampton. "
     "If the user asks about anything else, politely redirect them back to Brampton topics."
-)
+);
 
 
-def generate_response(prompt: str, max_tokens: int = 200) -> str:
+def generate_response( prompt: str, max_tokens: int = 200 ) -> str:
     resp = co.generate(
         model=CHAT_MODEL,
         prompt=prompt,
         max_tokens=max_tokens,
         return_likelihoods="NONE"
-    )
-    return resp.generations[0].text
+    );
+    return resp.generations[0].text;
 
 def brampton_response( query: str, k: int = 5, steering_scale: float = 0.5, sim_threshold: float = 0.3 ) -> str:
 
-    q_emb = np.array(co.embed(model=EMBED_MODEL, texts=[query]).embeddings[0])
-    sim = np.dot(q_emb, brampton_vector) / (np.linalg.norm(q_emb) * np.linalg.norm(brampton_vector))
+    q_emb = np.array( co.embed(model=EMBED_MODEL, texts=[query]).embeddings[0] );
+    sim = np.dot( q_emb, brampton_vector ) / ( np.linalg.norm( q_emb ) * np.linalg.norm( brampton_vector ) );
     if sim < sim_threshold:
-        return "I’m here to discuss Brampton—let's keep our conversation focused on this city!"
-    steered_emb = q_emb + steering_scale * brampton_vector
+        return "I’m here to discuss Brampton, let's keep our conversation focused on this city!";
+    steered_emb = q_emb + steering_scale * brampton_vector;
 
-    results = collection.query(
-        query_embeddings=[steered_emb.tolist()],
-        n_results=k
-    )
-    context = "\n".join(results["documents"][0])
+    results = collection.query( query_embeddings=[steered_emb.tolist()], n_results=k );
+    context = "\n".join( results["documents"][0] );
 
     prompt = (
         f"{SYSTEM_INSTR}\n"
         f"Context:\n{context}\n\n"
         f"User: {query}\n"
         f"Assistant:"
-    )
+    );
 
     return generate_response( prompt );
 
 def respond(message, chat_history):
-    bot_reply = brampton_response(message)
-    # append the new turn as a tuple (user_msg, bot_reply)
-    chat_history = chat_history + [(message, bot_reply)]
-    # return empty str to clear the box, and the updated history
-    return "", chat_history
+    bot_reply = brampton_response( message );
+    chat_history = chat_history + [( message, bot_reply )];
+    return "", chat_history;
 
 with gr.Blocks() as demo:
-    gr.Markdown("# Brampton Model")
-    chatbot = gr.Chatbot()
-    user_input = gr.Textbox(
-        show_label=False, 
-        placeholder="Ask me about anything"
-    )
-    # submit maps (user_input, chatbot) → (user_input, chatbot)
-    user_input.submit(
-        fn=respond, 
-        inputs=[user_input, chatbot], 
-        outputs=[user_input, chatbot]
-    )
-    gr.Markdown("Powered by Cohere command-r, embed-v4.0.")
+    gr.Markdown( "# Brampton Model" );
+    chatbot = gr.Chatbot();
+    user_input = gr.Textbox( show_label=False, placeholder="Ask me about anything" );
+    user_input.submit( fn=respond, inputs=[user_input, chatbot], outputs=[user_input, chatbot] );
+    gr.Markdown( "Powered by Cohere command-r, embed-v4.0." );
 
 # share=True creates a public URL when not on HF Spaces
 if __name__ == "__main__":
-    demo.launch(share=True)
+    demo.launch( share=True );