d2d35138218eb2dd3999a9fa4f7f93b5fef9ea7fe1e60270b943d1ba85993fec

vision_niah_d/niah_output/Qwen2-VL-m_rope-128frames-16card_8k-context-330k-llava-video/all_accuracies.json

@@ -0,0 +1,452 @@
+[
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    }
+]

vision_niah_d/niah_output/Qwen2-VL-m_rope-128frames-16card_8k-context-330k-llava-video/avg_accuracy.txt

@@ -0,0 +1 @@
+Average Accuracy: 0.7866666666666665

vision_niah_d/niah_output/Qwen2-VL-t_scale2_change_freq-128frames-16card_8k-context-330k-llava-video/all_accuracies.json

@@ -0,0 +1,452 @@
+[
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 20.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 40.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 60.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 80.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 20.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 40.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 60.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 80.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 20.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 40.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 60.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 80.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 20.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 40.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 60.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 80.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 20.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 40.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 60.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 80.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    }
+]

vision_niah_d/niah_output/Qwen2-VL-t_scale2_change_freq-128frames-16card_8k-context-330k-llava-video/avg_accuracy.txt

@@ -0,0 +1 @@
+Average Accuracy: 0.9111111111111111

vision_niah_d/niah_output/Qwen2-VL-time_rope-128frames-16card_8k-context-330k-llava-video/all_accuracies.json

@@ -0,0 +1,452 @@
+[
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 0.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 0.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 20.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 40.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 60.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 80.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 0.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 0.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 100.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 100.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 0.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 20.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 40.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 60.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 80.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 100.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 0.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    }
+]

vision_niah_d/niah_output/Qwen2-VL-time_rope-128frames-16card_8k-context-330k-llava-video/avg_accuracy.txt

@@ -0,0 +1 @@
+Average Accuracy: 0.29333333333333333

vision_niah_d/niah_output/Qwen2-VL-vanilla_rope-128frames-16card_8k-context-330k-llava-video/all_accuracies.json

@@ -0,0 +1,452 @@
+[
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 20.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 40.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 60.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 80.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 100,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 0.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 300,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 0.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 500,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 0.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 700,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 0.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 20.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 40.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 60.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 80.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 900,
+        "Frame Depth": 100.0,
+        "Score": 1.0
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 0.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1100,
+        "Frame Depth": 100.0,
+        "Score": 0.8
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 0.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 20.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 40.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 60.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 80.0,
+        "Score": 0.4
+    },
+    {
+        "Num. Frame": 1300,
+        "Frame Depth": 100.0,
+        "Score": 0.6
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 0.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 20.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 40.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 60.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 80.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1500,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1700,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 0.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 20.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 40.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 60.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 80.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 1900,
+        "Frame Depth": 100.0,
+        "Score": 0.2
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2100,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2300,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2500,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2700,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 0.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 20.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 40.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 60.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 80.0,
+        "Score": 0.0
+    },
+    {
+        "Num. Frame": 2900,
+        "Frame Depth": 100.0,
+        "Score": 0.0
+    }
+]

vision_niah_d/niah_output/Qwen2-VL-vanilla_rope-128frames-16card_8k-context-330k-llava-video/avg_accuracy.txt

@@ -0,0 +1 @@
+Average Accuracy: 0.31777777777777777

vision_niah_d/produce_haystack_embedding.py

@@ -0,0 +1,142 @@
+from qwen_vl_utils import process_vision_info
+from decord import VideoReader, cpu
+import argparse
+import os
+import numpy as np
+from tqdm import tqdm
+import torch
+import transformers
+import math
+from PIL import Image
+from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
+from torchvision import io, transforms
+from torchvision.transforms import InterpolationMode
+IMAGE_FACTOR = 28
+MIN_PIXELS = 144 * 28 * 28
+MAX_PIXELS = 144 * 28 * 28
+MAX_RATIO = 200
+def load_video_batches(video_path, batch_size):
+    global args
+    vr = VideoReader(video_path, ctx=cpu(0))
+    total_frame_num = len(vr)
+    fps = round(vr.get_avg_fps())
+    frame_idx = [i for i in range(0, len(vr), fps)]
+    for start_idx in range(0, len(frame_idx), batch_size):
+        end_idx = min(start_idx + batch_size, total_frame_num)
+        frame_indices = frame_idx[start_idx:end_idx]
+        batch_frames = vr.get_batch(frame_indices).asnumpy()
+        batch_frames = torch.tensor(batch_frames).permute(0, 3, 1, 2)
+        # import pdb; pdb.set_trace()
+        nframes, _, height, width = batch_frames.shape
+        # if torch.unique(batch_frames).numel() == 1:
+        #     batch_frames.fill_(args.v)
+        #     print(torch.unique(batch_frames).item())
+        resized_height, resized_width = 252, 448
+        # resized_height, resized_width = smart_resize(
+        #         height,
+        #         width,
+        #         factor=IMAGE_FACTOR,
+        #         min_pixels=MIN_PIXELS,
+        #         max_pixels=MAX_PIXELS,
+        #     )
+        batch_frames = transforms.functional.resize(
+            batch_frames,
+            [resized_height, resized_width],
+            interpolation=InterpolationMode.BICUBIC,
+            antialias=True,
+        ).float()
+
+        yield batch_frames
+
+def round_by_factor(number: int, factor: int) -> int:
+    """Returns the closest integer to 'number' that is divisible by 'factor'."""
+    return round(number / factor) * factor
+
+
+def ceil_by_factor(number: int, factor: int) -> int:
+    """Returns the smallest integer greater than or equal to 'number' that is divisible by 'factor'."""
+    return math.ceil(number / factor) * factor
+
+
+def floor_by_factor(number: int, factor: int) -> int:
+    """Returns the largest integer less than or equal to 'number' that is divisible by 'factor'."""
+    return math.floor(number / factor) * factor
+def smart_resize(
+    height: int, width: int, factor: int = IMAGE_FACTOR, min_pixels: int = MIN_PIXELS, max_pixels: int = MAX_PIXELS
+) -> tuple[int, int]:
+    """
+    Rescales the image so that the following conditions are met:
+
+    1. Both dimensions (height and width) are divisible by 'factor'.
+
+    2. The total number of pixels is within the range ['min_pixels', 'max_pixels'].
+
+    3. The aspect ratio of the image is maintained as closely as possible.
+    """
+    if max(height, width) / min(height, width) > MAX_RATIO:
+        raise ValueError(
+            f"absolute aspect ratio must be smaller than {MAX_RATIO}, got {max(height, width) / min(height, width)}"
+        )
+    h_bar = max(factor, round_by_factor(height, factor))
+    w_bar = max(factor, round_by_factor(width, factor))
+    if h_bar * w_bar > max_pixels:
+        beta = math.sqrt((height * width) / max_pixels)
+        h_bar = floor_by_factor(height / beta, factor)
+        w_bar = floor_by_factor(width / beta, factor)
+    elif h_bar * w_bar < min_pixels:
+        beta = math.sqrt(min_pixels / (height * width))
+        h_bar = ceil_by_factor(height * beta, factor)
+        w_bar = ceil_by_factor(width * beta, factor)
+    return h_bar, w_bar
+
+def main(args):
+    video_path = args.video_path
+    model_path = args.model
+    model_name = "llava_qwen"
+    
+    model = Qwen2VLForConditionalGeneration.from_pretrained(model_path, 
+                                                        device_map="auto",
+                                                        torch_dtype=torch.bfloat16, 
+                                                        attn_implementation="flash_attention_2"
+                                                        )
+    processor = AutoProcessor.from_pretrained("/mnt/petrelfs/weixilin/cache/Qwen2-VL-7B-Instruct")
+    del model.model.layers
+    # Process video in batches 
+    batch_size = 32
+    total_batches = (args.sampled_frames_num + batch_size - 1) // batch_size
+    image_feature_list = []
+    if args.add_newline_token:
+        newline_token_embeddong = model.model.image_newline
+    with torch.inference_mode():
+        for i, video_batch in tqdm(enumerate(load_video_batches(video_path, batch_size)), total=total_batches, desc="Processing Video Batches"):
+            v_test = processor.image_processor(images=None, videos=video_batch)
+            merge_length = processor.image_processor.merge_size**2
+            pixel_values_videos,video_grid_thw=torch.from_numpy(v_test['pixel_values_videos']), torch.from_numpy(v_test['video_grid_thw']).to(model.device)
+            # if i > 30:
+            #     import pdb; pdb.set_trace()
+            print(video_grid_thw)
+            # import pdb; pdb.set_trace()
+            pixel_values_videos = pixel_values_videos.type(model.visual.get_dtype()).to(model.device)
+            video_embeds = model.visual(pixel_values_videos, grid_thw=video_grid_thw).to(model.device)
+            
+            print(video_embeds.shape)
+            if args.add_newline_token:
+                image_features = torch.cat([image_features, newline_token_embeddong.unsqueeze(0).expand(image_features.shape[0], 1, -1)], dim=1)
+            image_feature_list.append(video_embeds.to(torch.bfloat16).to("cpu"))
+            if i > total_batches:
+                break
+    image_feature_list = torch.cat(image_feature_list, dim=0)
+    os.makedirs(args.output_dir, exist_ok=True)
+    torch.save(image_feature_list, f"{args.output_dir}/video_embeddings.pt")
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model", type=str, default="/mnt/petrelfs/weixilin/cache/Qwen2-VL-7B-Instruct")
+    # parser.add_argument("--v", type=int, default=255)
+    parser.add_argument("--video_path", type=str, default="/mnt/petrelfs/weixilin/projects/MLLM/LongVA/asset/videos/movie.mp4")
+    parser.add_argument("--sampled_frames_num", type=int, default=6000)
+    parser.add_argument("--output_dir", type=str, default="/mnt/petrelfs/weixilin/projects/MLLM/Qwen2-VL/vision_niah/video_needle_haystack/data/haystack_vicuna_embeddings_6000frames-tune_projector")
+    parser.add_argument("--pooling_size", type=int, default=0)
+    parser.add_argument("--add_newline_token", action="store_true")
+    args = parser.parse_args()
+    main(args)

vision_niah_d/produce_needle_embedding.py

@@ -0,0 +1,163 @@
+from qwen_vl_utils import process_vision_info
+from decord import VideoReader, cpu
+import argparse
+import numpy as np
+from tqdm import tqdm
+import torch
+import transformers
+import math
+from PIL import Image
+from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
+from torchvision import io, transforms
+from torchvision.transforms import InterpolationMode
+import os
+import json
+IMAGE_FACTOR = 28
+MIN_PIXELS = 144 * 28 * 28
+MAX_PIXELS = 144 * 28 * 28
+MAX_RATIO = 200
+
+def round_by_factor(number: int, factor: int) -> int:
+    """Returns the closest integer to 'number' that is divisible by 'factor'."""
+    return round(number / factor) * factor
+
+
+def ceil_by_factor(number: int, factor: int) -> int:
+    """Returns the smallest integer greater than or equal to 'number' that is divisible by 'factor'."""
+    return math.ceil(number / factor) * factor
+
+
+def floor_by_factor(number: int, factor: int) -> int:
+    """Returns the largest integer less than or equal to 'number' that is divisible by 'factor'."""
+    return math.floor(number / factor) * factor
+def smart_resize(
+    height: int, width: int, factor: int = IMAGE_FACTOR, min_pixels: int = MIN_PIXELS, max_pixels: int = MAX_PIXELS
+) -> tuple[int, int]:
+    """
+    Rescales the image so that the following conditions are met:
+
+    1. Both dimensions (height and width) are divisible by 'factor'.
+
+    2. The total number of pixels is within the range ['min_pixels', 'max_pixels'].
+
+    3. The aspect ratio of the image is maintained as closely as possible.
+    """
+    if max(height, width) / min(height, width) > MAX_RATIO:
+        raise ValueError(
+            f"absolute aspect ratio must be smaller than {MAX_RATIO}, got {max(height, width) / min(height, width)}"
+        )
+    h_bar = max(factor, round_by_factor(height, factor))
+    w_bar = max(factor, round_by_factor(width, factor))
+    if h_bar * w_bar > max_pixels:
+        beta = math.sqrt((height * width) / max_pixels)
+        h_bar = floor_by_factor(height / beta, factor)
+        w_bar = floor_by_factor(width / beta, factor)
+    elif h_bar * w_bar < min_pixels:
+        beta = math.sqrt(min_pixels / (height * width))
+        h_bar = ceil_by_factor(height * beta, factor)
+        w_bar = ceil_by_factor(width * beta, factor)
+    return h_bar, w_bar
+
+def read_json_file(file_path):
+    """
+    读取JSON文件并返回数据作为字典。
+
+    参数:
+    file_path (str): JSON文件的路径。
+
+    返回:
+    dict: JSON文件中的数据。
+    """
+    try:
+        # 打开文件并读取数据
+        with open(file_path, 'r', encoding='utf-8') as file:
+            # 将JSON数据解析为字典
+            data = json.load(file)
+        return data
+    except FileNotFoundError:
+        print(f"The file {file_path} was not found.")
+    except json.JSONDecodeError:
+        print(f"Error decoding JSON from file {file_path}.")
+    except Exception as e:
+        print(f"An error occurred: {e}")
+
+def fetch_image(ele, size_factor: int = IMAGE_FACTOR) -> Image.Image:
+    if "image" in ele:
+        image = ele["image"]
+    else:
+        image = ele["image_url"]
+    image_obj = None
+    if isinstance(image, Image.Image):
+        image_obj = image
+    elif image.startswith("http://") or image.startswith("https://"):
+        image_obj = Image.open(requests.get(image, stream=True).raw)
+    elif image.startswith("file://"):
+        image_obj = Image.open(image[7:])
+    elif image.startswith("data:image"):
+        if "base64," in image:
+            _, base64_data = image.split("base64,", 1)
+            data = base64.b64decode(base64_data)
+            image_obj = Image.open(BytesIO(data))
+    else:
+        image_obj = Image.open(image)
+    if image_obj is None:
+        raise ValueError(f"Unrecognized image input, support local path, http url, base64 and PIL.Image, got {image}")
+    image = image_obj.convert("RGB")
+    ## resize
+    if "resized_height" in ele and "resized_width" in ele:
+        resized_height, resized_width = 252, 448
+        # resized_height, resized_width = smart_resize(
+        #     ele["resized_height"],
+        #     ele["resized_width"],
+        #     factor=size_factor,
+        # )
+    else:
+        width, height = image.size
+        min_pixels = ele.get("min_pixels", MIN_PIXELS)
+        max_pixels = ele.get("max_pixels", MAX_PIXELS)
+        # resized_height, resized_width = smart_resize(
+        #     height,
+        #     width,
+        #     factor=size_factor,
+        #     min_pixels=min_pixels,
+        #     max_pixels=max_pixels,
+        # )
+        resized_height, resized_width = 252, 448
+    image = image.resize((resized_width, resized_height))
+
+    return image
+def main(args):
+    model_path = args.model
+    model = Qwen2VLForConditionalGeneration.from_pretrained(model_path, 
+                                                        device_map="auto",
+                                                        torch_dtype=torch.bfloat16, 
+                                                        attn_implementation="flash_attention_2"
+                                                        )
+    processor = AutoProcessor.from_pretrained("/mnt/petrelfs/weixilin/cache/Qwen2-VL-7B-Instruct")
+    del model.model.layers
+    # dataset = load_dataset(args.needle_dataset)["test"]
+    # dataset = load_dataset('json', '/mnt/petrelfs/weixilin/projects/MLLM/LongVA/vision_niah/needle_datasets/dataset.json')
+    dataset = read_json_file(args.needle_dataset)
+    for index, instance in enumerate(dataset):
+
+        # image = instance["image"].convert("RGB")
+        img = fetch_image({"image": os.path.join('/mnt/petrelfs/weixilin/projects/MLLM/Qwen2-VL/vision_niah/needle_datasets/images', instance['path']), "resized_height": 252, "resized_width": 448})
+        image_single = processor.image_processor(images=[img], videos=None)
+        merge_length = processor.image_processor.merge_size**2
+        pixel_values, image_grid_thw=torch.from_numpy(image_single['pixel_values']), torch.from_numpy(image_single['image_grid_thw']).to(model.device)
+        # import pdb; pdb.set_trace()
+        pixel_values = pixel_values.type(model.visual.get_dtype()).to(model.device)
+        image_embed = model.visual(pixel_values, grid_thw=image_grid_thw).to(model.device)
+        print(image_embed.shape)
+        os.makedirs(args.output_dir, exist_ok=True)
+        torch.save(image_embed, f"{args.output_dir}/{index}.pt")
+
+        
+        
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model", type=str, default="/mnt/petrelfs/weixilin/cache/Qwen2-VL-7B-Instruct")
+    parser.add_argument("--needle_dataset", type=str, default="/mnt/petrelfs/weixilin/projects/MLLM/Qwen2-VL/vision_niah/needle_datasets/dataset_change_format_debug.json")
+    parser.add_argument("--output_dir", type=str, default="/mnt/petrelfs/weixilin/projects/MLLM/Qwen2-VL/vision_niah/video_needle_haystack/data/needle_vicuna_embeddings_144tokens-tune_projector_interrupt_debug")
+    args = parser.parse_args()
+    main(args)