minor updates

Author: Haozhe Qi

llava/action/utils.py

@@ -232,9 +232,9 @@ def format_task_related_prompt(question, question_type, meta_data = None, perspe
     """
 
     if perspective == "first_person":
-        perspective_prefix = "You are seeing this video from egocentric view and you are the person. Your hands are sometimes interacting with obects. "
+        perspective_prefix = "You are seeing this video from egocentric view and you are the person. Your hands are sometimes interacting with objects. What action are you doing?"
     elif perspective == "third_person":
-        perspective_prefix = "The video is taken from egocentric view. What action is the person performing? "
+        perspective_prefix = "The video is taken from egocentric view. The person's hands are sometimes interacting with objects. What action is the person doing?"
 
     if question_type.startswith("mc_") or question_type == 'temporal_cot':
 

run_llmseval_clariden.sbatch

@@ -52,7 +52,7 @@ LAUNCHER="accelerate launch \
 PYTHON_FILE="-m lmms_eval"
 # PYTHON_ARGS=" \
 #     --model llava_onevision \
-#     --model_args pretrained=lmms-lab/llava-onevision-qwen2-0.5b-ov,conv_template=qwen_1_5,model_name=llava_qwen \
+#     --model_args pretrained=experiments/llava-onevision-qwen2-0.5b-ov,conv_template=qwen_1_5,model_name=llava_qwen \
 #     --tasks video_dc499 \
 #     --batch_size 1 \
 #     --log_samples_suffix llava_onevision \
@@ -63,7 +63,7 @@ PYTHON_FILE="-m lmms_eval"
 PYTHON_ARGS=" \
     --model llava_vid \
     --model_args pretrained=lmms-lab/LLaVA-Video-7B-Qwen2,conv_template=qwen_1_5,max_frames_num=64,mm_spatial_pool_mode=average \
-    --tasks ai2d \
+    --tasks videomme \
     --batch_size 1 \
     --log_samples \
     --log_samples_suffix llava_vid \