remove hf_token argument

explore.py

@@ -23,8 +23,8 @@ import random
 # --server.maxUploadSize 3000
 REPO_NAME = 'ncoria/llava-lora-vicuna-clip-5-epochs-merge'
 
-def load_llava_model(hf_token):
-    return load_llava_checkpoint_hf(REPO_NAME, hf_token)
+def load_llava_model():
+    return load_llava_checkpoint_hf(REPO_NAME)
 
 def get_unique_labels(label_list: list[str]):
     label_set = set()
@@ -129,7 +129,8 @@ def ask_question_with_image_llava(image, system_prompt, question,
 
 def ask_summary_question(image_array, label_array, api_key):
     # load llava model
-    tokenizer, model, image_processor = load_llava_model(hf_token)
+    with st.spinner("Loading LLaVA model. This can take 10 to 30 minutes. Please wait..."):
+        tokenizer, model, image_processor = load_llava_model()
 
     # global variable
     system_prompt = SYSTEM_PROMPT
@@ -172,7 +173,7 @@ if "embeddings_df" not in st.session_state:
 st.title('batik: behavior discovery and LLM-based interpretation')
 
 api_key = st.text_input("OpenAI API Key:","")
-hf_token = st.text_input("HuggingFace Token:","")
+
 st.subheader("generate or import embeddings")
 
 st.text("Upload files to generate embeddings.")
@@ -325,7 +326,8 @@ if uploaded_file is not None and st.session_state.embeddings_df is not None:
                         "Also, designate a behavioral subtype of the given label that describes the current social interaction based on what you see about the posture of the mice and "\
                         "how they are positioned with respect to each other. Usually, the body parts (i.e., tail, genitals, face, body, ears, paws)" \
                         "of the mice that are closest to each other will give some clue. Please limit behavioral subtype to a 1-4 word phrase. limit your response to 4 sentences."
-            tokenizer, model, image_processor = load_llava_model(hf_token)
+            with st.spinner("Loading LLaVA model. This can take 10 to 30 minutes. Please wait..."):
+                tokenizer, model, image_processor = load_llava_model()
             response = ask_question_with_image_llava(image, system_prompt, question,
                                                      tokenizer, model, image_processor)
             st.markdown(response)