Running on CPU

app.py

@@ -33,20 +33,20 @@ custom_css = """
 """
 print('Loading Model...')
 # Load Model
-@spaces.GPU
-def load_model():
-    llama = Llama(
-            model_path=hf_hub_download(
-            repo_id="motherduckdb/DuckDB-NSQL-7B-v0.1-GGUF",
-            filename="DuckDB-NSQL-7B-v0.1-q8_0.gguf",
-            local_dir='.'
-            ),
-            n_ctx=2048,
-            n_gpu_layers=-1
-            )
-    return llama
-
-llama = load_model()
+# @spaces.GPU
+# def load_model():
+llama = Llama(
+        model_path=hf_hub_download(
+        repo_id="motherduckdb/DuckDB-NSQL-7B-v0.1-GGUF",
+        filename="DuckDB-NSQL-7B-v0.1-q8_0.gguf",
+        local_dir='.'
+        ),
+        n_ctx=2048,
+        n_gpu_layers=0
+        )
+    # return llama
+
+# llama = load_model()
 print('Model Loaded...')
 
 # Get Databases
@@ -88,7 +88,7 @@ def get_prompt(schema, query_input):
     return text
 
 # Generate SQL
-@spaces.GPU
+# @spaces.GPU
 def generate_sql(prompt):
     
     result = llama(prompt, temperature=0.1, max_tokens=1000)