search interface

app.py

@@ -1,64 +1,148 @@
+import json
+from pathlib import Path
+
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+import numpy as np
+
+from app.models.text_encoder import TextEncoder
+
+SUMMARY_DIR = Path("data/summaries")
+EMBEDDING_DIR = Path("data/embeddings")
+NAME_MAP_FILE = Path("data/name_map.json")
+
+
+def search_memes(prompt, top_k=10):
+    """
+    Search for memes based on the input prompt.
+
+    Args:
+        prompt: The text prompt to search for
+        top_k: Number of results to return
+
+    Returns:
+        List of dictionaries containing search results
+    """
+    # Initialize results list
+    results = []
+
+    # Get the embedding file paths
+    embedding_paths = list(EMBEDDING_DIR.glob("*.npy"))
+
+    # Load the embeddings
+    embeddings = [np.load(path) for path in embedding_paths]
+
+    # Load the text encoder
+    text_encoder = TextEncoder()
+
+    # Generate embeddings for the prompt
+    prompt_embedding = text_encoder.encode(prompt)
+
+    # Calculate similarities
+    similarities = np.dot(embeddings, prompt_embedding) / (
+        np.linalg.norm(embeddings, axis=1) * np.linalg.norm(prompt_embedding)
+    )
+
+    # Get the top k indices
+    top_k_indices = np.argsort(similarities)[-top_k:]
+
+    # Load the summaries
+    summaries = []
+    for path in SUMMARY_DIR.glob("*.txt"):
+        with open(path, "r", encoding="utf-8") as f:
+            summaries.append(f.read())
+
+    # Load the name map
+    with open(NAME_MAP_FILE, "r") as f:
+        name_map = json.load(f)
+
+    # Process the top k results
+    for i, index in enumerate(top_k_indices[::-1]):
+        try:
+            result = {
+                "rank": i + 1,
+                "similarity": round(float(similarities[index]), 3),
+                "filename": embedding_paths[index].stem,
+                "original_filename": name_map.get(
+                    embedding_paths[index].stem, "Unknown"
+                ),
+                "summary": summaries[index]
+                if index < len(summaries)
+                else "No summary available",
+            }
+            results.append(result)
+        except (IndexError, KeyError) as e:
+            results.append(
+                {"rank": i + 1, "error": f"Error processing result {i + 1}: {str(e)}"}
+            )
+
+    return results
+
+
+def format_results(results):
+    """Format the results for display in the Gradio interface"""
+    html_output = ""
+    for result in results:
+        if "error" in result:
+            html_output += "<div style='margin-bottom: 20px; padding: 10px; border: 1px solid #ff6b6b; border-radius: 5px;'>"
+            html_output += f"<p><b>Rank {result['rank']}:</b> {result['error']}</p>"
+            html_output += "</div>"
+        else:
+            html_output += "<div style='margin-bottom: 20px; padding: 10px; border: 1px solid #ddd; border-radius: 5px;'>"
+            html_output += f"<p><b>Rank {result['rank']}</b> (Similarity: {result['similarity']})</p>"
+            html_output += f"<p><b>File:</b> {result['original_filename']}</p>"
+            html_output += f"<p><b>Summary:</b> {result['summary']}</p>"
+            html_output += "</div>"
+
+    return html_output
+
+
+# Define the Gradio interface
+def search_interface(prompt, top_k):
+    """Main search interface function for Gradio"""
+    if not prompt:
+        return "Please enter a search query"
+
+    results = search_memes(prompt, int(top_k))
+    formatted_results = format_results(results)
+    return formatted_results
+
+
+# Create the Gradio app
+with gr.Blocks(title="Meme Search", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🔍 Meme Search")
+    gr.Markdown("Search for memes using natural language descriptions")
+
+    with gr.Row():
+        with gr.Column(scale=4):
+            prompt_input = gr.Textbox(
+                label="Search Query", placeholder="Enter your search here..."
+            )
+        with gr.Column(scale=1):
+            top_k_slider = gr.Slider(
+                minimum=1, maximum=20, value=5, step=1, label="Number of Results"
+            )
+
+    search_button = gr.Button("Search", variant="primary")
+
+    output = gr.HTML(label="Results")
+
+    search_button.click(
+        fn=search_interface, inputs=[prompt_input, top_k_slider], outputs=output
+    )
+
+    # Also trigger search on Enter key
+    prompt_input.submit(
+        fn=search_interface, inputs=[prompt_input, top_k_slider], outputs=output
+    )
 
+    gr.Markdown("## How to use")
+    gr.Markdown("""
+    1. Enter a description of the meme you're looking for
+    2. Adjust the number of results to show
+    3. Click 'Search' or press Enter
+    4. Results are sorted by similarity to your query
+    """)
 
+# Launch the app
 if __name__ == "__main__":
     demo.launch()

app/__init__.py

@@ -0,0 +1 @@
+"""Application for searching video summaries by natural language."""

app/models/__init__.py

@@ -0,0 +1 @@
+"""Models for the application."""

app/models/text_encoder.py

@@ -0,0 +1,16 @@
+"""Text encoder."""
+
+from sentence_transformers import SentenceTransformer
+from torch import Tensor
+
+
+class TextEncoder:
+    """Text encoder."""
+
+    def __init__(self) -> None:
+        """Initialize the text encoder."""
+        self.model = SentenceTransformer("sentence-transformers/all-MiniLM-L12-v1")
+
+    def encode(self, text: str) -> Tensor:
+        """Encode a string into an embedding."""
+        return self.model.encode(text)

data/embeddings/017d956359676be2393d8e07c0d9cd.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db1acef571b048dafd4ad6e1068b14f903b1c4f104efdb39ac379e7fab07dad5
+size 1664

data/embeddings/0a4afbdf14f42abcaf731aedbd8783.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54370cc558e4dfe9e88fcfc4be2455e75979559809b31ace22d864d22c95572f
+size 1664

data/embeddings/0z8eppisoyzd3jte.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9baf359732587eeb7d23eb9f1c72a117bc576f60603a5213257177ae0fd31df2
+size 1664

data/embeddings/1004332103284052018.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88c6cab8513c6ee763b4d6e46145ca76c1c640b1aa4a652295e0951a8baad8ce
+size 1664

data/embeddings/170f3a0d74d628f405745cd107d8fc.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be38eed59873cedfd5af66461275ea4ed79f8fbbe8da4357e66c1a8e0714b741
+size 1664

data/embeddings/1_4972134405645533911.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a863fd8aea49898bc7eea5af07b7b259e81cfda142c7dfe8c8a010f4e52684f
+size 1664

data/embeddings/1cmt9rrcral5m1t.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d29c62ceec03f9b667ac73efc1a5b4cb0d6a47f18ca3b57b6f3ffe9a95a563ce
+size 1664

data/embeddings/1x_mpgvsmlwi4b4.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3d1350110b8c416855ffee53990a846bca4a0aa3badda3d4e87de62cc98154c
+size 1664

data/embeddings/20250209_cooperativeperfectpon.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2e402c40cd62d7e4aa66ce4d20b55dd08f5ca10e034225ac74ddf4fcf49e056
+size 1664

data/embeddings/244301680_1204542536721927_730.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a63e0e00f40677643cc395759cfe9d122ccbbfbea587cc4f25cca0e8f1d1150
+size 1664

data/embeddings/248320392_939578913314000_7632.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a724e3e9519fdf22dfbfcee43a0bc34c1cac7de6a1778dbc8db469a49776ef7
+size 1664

data/embeddings/2df31f1dd1d31b6d5b55934e2c0125.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c31c3f38645bc9b29eaa224a00aed02eee3b42b471dc644ab7c070aaac17177
+size 1664

data/embeddings/2gtri3bqxrdkzp2.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c14e55edf321a1ff588b6739c092a92d4f12bed39eddbfe80c8e657cb1f7667
+size 1664

data/embeddings/2og27hbdaohr0_u7.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8dd11c7fb54be301478c3b207399f4e05477964a6427a77370d1d5aa7e7ce5c
+size 1664

data/embeddings/311472807_847803186574673_6180.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:318be9a37558cee400a6225d7726669f50e914351f4b59f3f3a10f6522a2b72f
+size 1664

data/embeddings/317972007_872625493783372_7982.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f15c811de05011b9a445d2b1787ac17563ed06f588bc05ad29ad061594ec2996
+size 1664

data/embeddings/3c494477159df432b7c51e2f809dd3.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f0cfb7abafb4ab98163e98cf31697636d9bd252a9ff00f013b182b65758f798
+size 1664

data/embeddings/3cl41iruuh0jfi27.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6756974f7cff611b30b3328216be9b61785125ee55a5a8e04dcf98d52193a3d
+size 1664

data/embeddings/3jev0656inc_e09p.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef0831f8bf216eead06de2b451f065c0a5eed8292aa49e5f82328d42760ccbd
+size 1664

data/embeddings/3wlic3cumg3yvudn.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0538ffda85460357b12f900540036f6dac0e0c58198381abe15637c5dfe36df7
+size 1664

data/embeddings/405360888_7293554827322499_503.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e4175a575a76e4b5c5b095ab1d5204f92c407b54e562b50cd6725aea18124ad
+size 1664

data/embeddings/441930059_1175058477005126_751.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c2ea74d9b6f1f1d4036c7ab00c8a6ff6539e799a6139da55383ae462e3d0f46
+size 1664

data/embeddings/469182540_8739739586133293_167.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da8849ac88c2f170fa51a63522a91f2944bfb00f278cf55ac80e5e1ef76b3a5f
+size 1664

data/embeddings/471482695_9131258876896657_266.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5726208f69de7831c35fe63fb5c16c5aeec8f6169f78facc14d194e41275d5e5
+size 1664

data/embeddings/47671917_1318637692320758_5748.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d4a7bcc14540a669baa94db0664b7ce2c4db1582266a09d5ff65a15d3a63b33
+size 1664

data/embeddings/4bdeac2a1334624ce3c8afe96ea9c5.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ec8784a36fe6809a51b41a3868f2573f11798c5db3c863ccdecedcaf186bccb
+size 1664

data/embeddings/4kcamera.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1138af505b385bc9ad35970a66b5d66769a0f70a4879fced7b2c91037e97c0d3
+size 1664

data/embeddings/4x.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:104ecd8927a00ea92ce2bcadace44dfa7eb4a7f99689110372eaa26e86aaa5c9
+size 1664

data/embeddings/5fb3731a8dcfd8172c9bdd12ae1970.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78060d6e575b4a34e711c836da5db1d4754db68f1bd5e5f43431af3232697308
+size 1664

data/embeddings/6e725c5ef715b822f11ea67c2ca9cb.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44228b7afd5116380c612f6382fd7b692b7d7af9ab0f3d2f09c405bbc748a401
+size 1664

data/embeddings/6o6psry1hrnpwhvo.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33298d136cc49822432a6360d135663352c42327d3643d69502534677d108856
+size 1664

data/embeddings/6rme9unctzitcye8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb63ed9ef3845e80e26b8ae2cd600a7745fd4c0b3c5ff2ec7bd54344ad3c95e6
+size 1664

data/embeddings/789591e5aee5552b63569b5aae9f91.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb5abf3013d3d2469b058d9bf8ede43f5d620d67dc3c3f7c4dbc02d0c4e45621
+size 1664

data/embeddings/8212422b9b4889a0770c55c18bbf96.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e35457d37a51952233dce64eeb633323cc94403d2a5ca8f2f7d9e7cee3927c1
+size 1664

data/embeddings/889c0176ad2dd4e71b6a9ba7691b0b.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:accb41051a5c9b9f8ea3c7fd45de047b693030c8be89db22ecbf77fd0e49658b
+size 1664

data/embeddings/8e4282f75a64f7e3c35175be4bbcf1.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:199a6212b7d30d5021634db203ae265a50b4cd7d1e07f17e7063c8d26af94bcc
+size 1664

data/embeddings/8fi9o0y09mibdhjx.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1433234fd45e0c18a300a27bdfb19a5fbea2a1b61852329105a5dea0a7f4d6c1
+size 1664

data/embeddings/8v5z0c8f2unrtshm.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4470a9d704aadce553b09b5c7bd29c9f31729e2c93d69a68d6f5e57d393c5396
+size 1664

data/embeddings/9b91784d80d2d2776635a21f3216ad.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67d6caeb6dae68fadea174f8adc595bb3bc1216188f8b92f792f20caac0d5f43
+size 1664

data/embeddings/9ef3d77d6ba7aaa02fb07a20e65002.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2a653922751b6c5715119d4ce97ff90aeedcfc4a471670bbb80f12c380de799
+size 1664

data/embeddings/9qgdm3pnsbmkj0bp.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:929ba574f5bd455116ef4d44f7d6fbd6d3d9a832adf7809a8067041169512471
+size 1664

data/embeddings/_3qzrqxl7iyc0kcw.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a861fc61fb65647e34efe639bcf2d19700c2892be7ad5f46154d690ac55b07e
+size 1664

data/embeddings/_72rkphcvkjmc7x.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fe57f5a778b3b819669abde2757d59379423606b131ea1742dbc17aaf3890d2
+size 1664

data/embeddings/_agckaq6h32ycmqi.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc655444a1e15df3c6f9921689e0a1b829f5c8fbdc00af0e3ab2c0d1f6f2fc66
+size 1664

data/embeddings/_owxv_jrbidwkise.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7a4b940c5dd0817042ff53c5b3ea442c99583a3e5bd07cd5684a3afbfe8ab6
+size 1664

data/embeddings/a34e43fd73ea2615461fa0de628606.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8678fb6ae9cba5a79b0bfede24b9b3233328a2576e8368036b4be6a8098da7d2
+size 1664