update files with fixes

apply_model.py

@@ -60,7 +60,7 @@ st.title('batik: frame classifier')
 
 st.text("Upload files to apply trained classifier on.")
 with st.form('embedding_generation_settings'):
-    seq_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'], accept_multiple_files=False)
+    seq_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'])
     annot_files = st.file_uploader("Choose an annotation File", type=['annot','csv'], accept_multiple_files=True)
     downsample_rate = st.number_input('Downsample Rate',value=4)
     submit_embed_settings = st.form_submit_button('Create Embeddings', type='secondary')
@@ -118,12 +118,14 @@ if st.session_state.embeddings_df is not None and svm_clf is not None:
         apply_model = st.form_submit_button("Apply Model")
 
     if apply_model:
-        if 'Test' in st.session_state.embeddings_df.index:
-            test_videos = True
-        elif 'Images' in st.session_state.embeddings_df.index:
+        if 'Test' in st.session_state.embeddings_df:
             test_videos = True
         else:
-            test_videos = False
+            print(f'shape of df: {st.session_state.embeddings_df.shape[0]}')
+            test_videos_array = [True for i in range(st.session_state.embeddings_df.shape[0])]
+            st.session_state.embeddings_df['Test'] = test_videos_array
+            test_videos = True
+
         kwargs = {'embeddings_df' : st.session_state.embeddings_df, 
                 'specified_classes' : specified_classes,
                 'classes_to_remove' : None,

explore.py

@@ -177,7 +177,7 @@ st.subheader("generate or import embeddings")
 
 st.text("Upload files to generate embeddings.")
 with st.form('embedding_generation_settings'):
-    seq_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'], accept_multiple_files=False)
+    seq_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'])
     annot_files = st.file_uploader("Choose an annotation File", type=['annot','csv'], accept_multiple_files=True)
     downsample_rate = st.number_input('Downsample Rate',value=4)
     submit_embed_settings = st.form_submit_button('Create Embeddings', type='secondary')
@@ -211,7 +211,7 @@ else:
 st.divider()
 st.subheader("provide video file if not yet already provided")
 
-uploaded_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'], accept_multiple_files=False)
+uploaded_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'])
 
 st.divider()
 if st.session_state.embeddings_df is not None and (uploaded_file is not None or seq_file is not None):

get_llava_response.py

@@ -47,8 +47,8 @@ def load_llava_checkpoint_hf(model_path, hf_token):
         bnb_4bit_use_double_quant=True,
         bnb_4bit_quant_type='nf4'
     )
-    tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False)
-    model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs)
+    tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False, token=hf_token)
+    model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, token=hf_token, **kwargs)
     mm_use_im_start_end = getattr(model.config, "mm_use_im_start_end", False)
     mm_use_im_patch_token = getattr(model.config, "mm_use_im_patch_token", True)
     if mm_use_im_patch_token:

train_model.py

@@ -35,7 +35,7 @@ def get_train_test_split(train_embeds, numerical_labels, test_size=0.05, random_
 @st.cache_resource
 def train_model(X_train, y_train, random_state=42):
     # Train SVM Classifier
-    svm_clf = SVC(kernel='rbf', random_state=random_state, probability=True)
+    svm_clf = SVC(kernel='rbf', random_state=random_state, probability=True, verbose=True)
     svm_clf.fit(X_train, y_train)
     return svm_clf
 
@@ -56,7 +56,7 @@ st.title('batik: frame classifier training')
 
 st.text("Upload files to train classifier on.")
 with st.form('embedding_generation_settings'):
-    seq_file = st.file_uploader("Choose a .seq File", type=['seq'])
+    seq_file = st.file_uploader("Choose a video file", type=['seq', 'mp4'])
     annot_files = st.file_uploader("Choose an annotation File", type=['annot','csv'], accept_multiple_files=True)
     downsample_rate = st.number_input('Downsample Rate',value=4)
     submit_embed_settings = st.form_submit_button('Create Embeddings', type='secondary')