Update app.py

app.py

@@ -10,13 +10,13 @@ def save_embeddings(sentences, filename):
 def load_embeddings(filename):
     return torch.load(filename, map_location=torch.device('cpu'))
 
-def preprocess_model_descriptions(file_path):
+def preprocess_space_descriptions(file_path):
     encodings = ['utf-8', 'latin-1', 'utf-16']
     for encoding in encodings:
         try:
-            df = pd.read_csv(file_path, sep='\t', header=None, names=['model_id', 'description'])
+            df = pd.read_csv(file_path, sep='\t', header=None, names=['space_id', 'description'])
             df.dropna(subset=['description'], inplace=True)
-            model_ids = df['model_id'].tolist()
+            space_ids = df['space_id'].tolist()
             descriptions = df['description'].tolist()
             break
         except UnicodeDecodeError:
@@ -24,22 +24,22 @@ def preprocess_model_descriptions(file_path):
     else:
         raise UnicodeDecodeError("Unable to decode the file using the available encodings.")
 
-    return model_ids, descriptions
+    return space_ids, descriptions
 
-def perform_similarity_search(query_embeddings, embeddings, model_ids, descriptions, top_k=10):
+def perform_similarity_search(query_embeddings, embeddings, space_ids, descriptions, top_k=10):
     cosine_scores = util.cos_sim(query_embeddings, embeddings)
     similarity_scores = cosine_scores.tolist()
 
     results = []
     for i, query_embedding in enumerate(query_embeddings):
-        query_results = sorted(zip(model_ids, descriptions, similarity_scores[i]), key=lambda x: x[2], reverse=True)[:top_k]
+        query_results = sorted(zip(space_ids, descriptions, similarity_scores[i]), key=lambda x: x[2], reverse=True)[:top_k]
         results.extend(query_results)
 
-    return pd.DataFrame(results, columns=["model_id", "description", "score"])
+    return pd.DataFrame(results, columns=["space_id", "description", "score"])
 
 model = SentenceTransformer('all-MiniLM-L6-v2')
 
-model_ids, descriptions = preprocess_model_descriptions('hf_spaces_descriptions.tsv')
+space_ids, descriptions = preprocess_space_descriptions('hf_spaces_descriptions.tsv')
 embeddings = load_embeddings('embeddings_hf_spaces_descriptions.pt')
 
 with gr.Blocks() as demo:
@@ -50,9 +50,8 @@ with gr.Blocks() as demo:
     
     def search(query):
         query_embedding = model.encode([query], convert_to_tensor=True)
-        return perform_similarity_search(query_embedding, embeddings, model_ids, descriptions, top_k=num_results)
+        return perform_similarity_search(query_embedding, embeddings, space_ids, descriptions, top_k=num_results)
     
     input.submit(search, inputs=input, outputs=df_output)
-    button.click(search, inputs=input, outputs=df_output)
 
 demo.launch()