# Prompt Enhancement for Video Generation on VideoPhy2 held-out (val) Source: [Prompt Enhancement for Video Generation on VideoPhy2 held-out (val)]() Collection index: [llms.txt]() Data updated: 2026-03-05T07:32:43.770Z Evaluation of models performing Prompt Enhancement for Video Generation using the held-out validation split of the VideoPhy2 dataset. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Prompt Enhancement for Video Generation | | Dataset (including any stated split) | VideoPhy2 held-out (val) | | Metric | SA (%) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 11 of 11 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [SA (%)](): Higher is better; unit: PERCENT. - [PC (%)](): Higher is better; unit: PERCENT. - [PC & SA (%)](): Higher is better; unit: PERCENT. ## Recorded results | Method | SA (%) | Result date | Report paper | | --- | --- | --- | --- | | PhyPrompt-7B | 41.8 | Not recorded | 69a90de72d603e6de3e4065a | | GPT-4o | 41.8 | Not recorded | 69a90de72d603e6de3e4065a | | DeepSeek-V3 | 41.6 | Not recorded | 69a90de72d603e6de3e4065a | | PhyPrompt-3B | 41.4 | Not recorded | 69a90de72d603e6de3e4065a | | PhyT2V | 41.2 | Not recorded | 69a90de72d603e6de3e4065a | | Original Prompt | 41 | Not recorded | 69a90de72d603e6de3e4065a | | PhyPrompt-1.5B | 40.2 | Not recorded | 69a90de72d603e6de3e4065a | | Qwen2.5-7B | 40 | Not recorded | 69a90de72d603e6de3e4065a | | Promptist | 37.8 | Not recorded | 69a90de72d603e6de3e4065a | | Qwen2.5-3B | 35.2 | Not recorded | 69a90de72d603e6de3e4065a | | Qwen2.5-1.5B | 34.6 | Not recorded | 69a90de72d603e6de3e4065a | ### PhyPrompt-7B SA (%): 41.8. Result ID: 69a9318d6dbb7075b9523c89. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"PhyPrompt-7B","Generator":"Lavie"}. This record does not assert independent reproduction. ### GPT-4o SA (%): 41.8. Result ID: 69a9318d6dbb7075b9523c8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Enhanced Model":"GPT-4o","Generator":"Lavie"}. This record does not assert independent reproduction. ### DeepSeek-V3 SA (%): 41.6. Result ID: 69a9318d6dbb7075b9523c8b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"DeepSeek-V3","Generator":"Lavie"}. This record does not assert independent reproduction. ### PhyPrompt-3B SA (%): 41.4. Result ID: 69a9318d6dbb7075b9523c87. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"PhyPrompt-3B","Generator":"Lavie"}. This record does not assert independent reproduction. ### PhyT2V SA (%): 41.2. Result ID: 69a9318d6dbb7075b9523c83. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Enhanced Model":"PhyT2V","Generator":"Lavie"}. This record does not assert independent reproduction. ### Original Prompt SA (%): 41. Result ID: 69a9318d6dbb7075b9523c81. [Reporting paper]() Recorded evaluation configuration: {"Enhanced Model":"N/A","Generator":"Lavie"}. This record does not assert independent reproduction. ### PhyPrompt-1.5B SA (%): 40.2. Result ID: 69a9318d6dbb7075b9523c85. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"PhyPrompt-1.5B","Generator":"Lavie"}. This record does not assert independent reproduction. ### Qwen2.5-7B SA (%): 40. Result ID: 69a9318d6dbb7075b9523c88. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"Qwen2.5-7B","Generator":"Lavie"}. This record does not assert independent reproduction. ### Promptist SA (%): 37.8. Result ID: 69a9318d6dbb7075b9523c82. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"Promptist","Generator":"Lavie"}. This record does not assert independent reproduction. ### Qwen2.5-3B SA (%): 35.2. Result ID: 69a9318d6dbb7075b9523c86. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"Qwen2.5-3B","Generator":"Lavie"}. This record does not assert independent reproduction. ### Qwen2.5-1.5B SA (%): 34.6. Result ID: 69a9318d6dbb7075b9523c84. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Enhanced Model":"Qwen2.5-1.5B","Generator":"Lavie"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.