Upload folder using huggingface_hub

demo.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import torch
-from transformers import AutoModelForCausalLM, LlamaTokenizer
+from transformers import AutoModelForCausalLM, AutoTokenizer
 import logging
 
 # Setup logging
@@ -11,18 +11,16 @@ def load_model():
     """Load fine-tuned model without quantization for CPU compatibility."""
     logger.info("Loading model...")
     
-    # Use explicit LlamaTokenizer instead of AutoTokenizer
-    tokenizer = LlamaTokenizer.from_pretrained(
-        "Zwounds/boolean-search-model",
-        trust_remote_code=True
+    # Use explicit AutoTokenizer instead of LlamaTokenizer
+    tokenizer = AutoTokenizer.from_pretrained(
+        "Zwounds/boolean-search-model"
     )
     
     # Load model in the most compatible way for Spaces
     model = AutoModelForCausalLM.from_pretrained(
         "Zwounds/boolean-search-model",
         low_cpu_mem_usage=True,
-        torch_dtype=torch.float32,  # Use standard floating point for CPU
-        trust_remote_code=True
+        torch_dtype=torch.float32  # Use standard floating point for CPU
     )
     return model, tokenizer