Visual Question Answering on TickTockVQA (test)
56.41Hour Accuracy (Baseline)Llama-3.2-11B w/ Swap-DPO
Evaluation Results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Llama-3.2-11B w/ Swap-DPOModel=Llama-3.2-11B, Training Stage=w/ Swap-DPO2026.03 | 56.41 | 64.42 | 66.69 | 73.62 | 46.22 | 48.48 | 1.03 | 0.97 | 4.85 | 4.61 | 61.93 | 58.79 | — | — | — | — | |
| Llama-3.2-11B w/ TickTockVQAModel=Llama-3.2-11B, Training Stage=w/ TickTockVQA2026.03 | 56.05 | 64.4 | 66.69 | 73.57 | 45.78 | 48.1 | 1.04 | 0.98 | 4.82 | 4.56 | 62.52 | 59.22 | — | — | — | — | |
| Gemma3-12B w/ Swap-DPOModel=Gemma3-12B, Training Stage=w/ Swap-DPO2026.03 | 46.67 | 56.01 | 57.54 | 67.64 | 35.32 | 37.89 | 1.36 | 1.29 | 6.16 | 5.85 | 81.86 | 77.97 | — | — | — | — | |
| Gemma3-12B w/ TickTockVQAModel=Gemma3-12B, Training Stage=w/ TickTockVQA2026.03 | 46.37 | 55.04 | 54.39 | 65.68 | 34.21 | 37.11 | 1.35 | 1.27 | 6.68 | 6.34 | 81.66 | 77.26 | — | — | — | — | |
| Qwen2.5-VL-7B w/ Swap-DPOModel=Qwen2.5-VL-7B, Training Stage=w/ Swap-DPO2026.03 | 35.39 | 44.98 | 46.16 | 55.29 | 23.06 | 25.08 | 1.61 | 1.55 | 7.24 | 7.01 | 96.42 | 93.37 | — | — | — | — | |
| Llama-3.2-11B w/ CtrlClockModel=Llama-3.2-11B, Training Stage=w/ CtrlClock2026.03 | 33.28 | 40.1 | 40.1 | 50.09 | 23.04 | 25 | 2.12 | 2.06 | 9.95 | 9.65 | 128.1 | 124.34 | — | — | — | — | |
| Qwen2.5-VL-7B w/ TickTockVQAModel=Qwen2.5-VL-7B, Training Stage=w/ TickTockVQA2026.03 | 33.18 | 42.94 | 41.51 | 52.7 | 20.34 | 22.76 | 1.73 | 1.66 | 8.53 | 8.19 | 104.31 | 100.2 | — | — | — | — | |
| Llama-3.2-11B w/ SynClockModel=Llama-3.2-11B, Training Stage=w/ SynClock2026.03 | 32.55 | 40.67 | 39.24 | 47.04 | 22.09 | 23.8 | 2.1 | 2.04 | 9.77 | 9.54 | 126.77 | 123.79 | — | — | — | — | |
| Qwen2.5-VL-7B w/ SynClockModel=Qwen2.5-VL-7B, Training Stage=w/ SynClock2026.03 | 28.11 | 37.51 | 35.68 | 47.17 | 16.12 | 18.28 | 2.15 | 2.09 | 9.6 | 9.35 | 129.51 | 126.34 | — | — | — | — | |
| Qwen2.5-VL-7B w/ CtrlClockModel=Qwen2.5-VL-7B, Training Stage=w/ CtrlClock2026.03 | 25.16 | 33.81 | 37.45 | 46.85 | 14.75 | 17.08 | 2.18 | 2.11 | 9.58 | 9.28 | 131.14 | 127.2 | — | — | — | — | |
| Qwen2.5-VL-7BModel=Qwen2.5-VL-7B, Training Stage=Baseline2026.03 | 17.65 | 25.26 | 22.44 | 32.14 | 6.04 | 7.26 | 2.47 | 2.44 | 11.96 | 11.81 | 148.62 | 146.76 | — | — | — | — | |
| Gemma3-12BModel=Gemma3-12B, Training Stage=Baseline2026.03 | 13 | 20.75 | 10.62 | 19.36 | 2.12 | 3.05 | 2.61 | 2.59 | 14.43 | 14.32 | 156.49 | 155.1 | — | — | — | — | |
| Llama-3.2-11BModel=Llama-3.2-11B, Training Stage=Baseline2026.03 | 11.49 | 18.35 | 8.58 | 16.85 | 1.41 | 2.02 | 2.59 | 2.57 | 14.79 | 14.7 | 156.96 | 155.67 | — | — | — | — | |
| Gemma3-12BZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 13 | 10.62 | 2.12 | 156.49 | |
| InternVL3-8BZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 13.59 | 9.08 | 2.2 | 159.92 | |
| It’s About TimeZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 28.95 | 25 | 18.54 | 135.15 | |
| Llama-3.2-11BZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 11.51 | 8.58 | 1.43 | 156.96 | |
| Qwen2.5-VL-7BZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 17.65 | 22.44 | 6.04 | 148.62 | |
| SpatialVLM-3BZero-shot=true2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | 12.51 | 6.44 | 1.05 | 161.68 |