disable use_fast tokenizing

app.py

@@ -41,16 +41,19 @@ st.write('loading chunks into vector db')
 model_name = "hkunlp/instructor-large"
 hf_embeddings = HuggingFaceInstructEmbeddings(
     model_name = model_name)
-db = Chroma.from_documents(texts, hf_embeddings)
+# db = Chroma.from_documents(texts, hf_embeddings)
 
-st.write('loading LLM')
+st.write('loading tokenizer')
 #model_name_or_path = "TheBloke/Llama-2-13B-chat-GPTQ"
 model_name_or_path = "TheBloke/Llama-2-13B-chat-GGUF"
+#tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
 tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
 
 model_basename = "model"
+
 use_triton = False
 DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"
+st.write('loading LLM')
 model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
         model_basename=model_basename,
         use_safetensors=True,