# Instruction Following Evaluation on AlpacaEval 2 Source: [Instruction Following Evaluation on AlpacaEval 2]() Collection index: [llms.txt]() Data updated: 2026-03-06T08:28:42.725Z Evaluation of model performance on instruction following using the AlpacaEval 2 dataset, measured by Win Rate (WR) and Length Consistency (LC). 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 | Instruction Following Evaluation | | Dataset (including any stated split) | AlpacaEval 2 | | Metric | Win Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [Win Rate](): Higher is better; unit: PERCENT. - [Length Consistency](): Higher is better; unit: PERCENT. ## Recorded results | Method | Win Rate | Result date | Report paper | | --- | --- | --- | --- | | VRM-PPO | 48.14 | Not recorded | 69aa5f712d603e6de3e4182a | | VRM-PPO | 45.3 | Not recorded | 69aa5f712d603e6de3e4182a | | PPO | 44.75 | Not recorded | 69aa5f712d603e6de3e4182a | | WPO | 44.11 | Not recorded | 69aa5f712d603e6de3e4182a | | SIMPO | 40.03 | Not recorded | 69aa5f712d603e6de3e4182a | | PPO | 39.52 | Not recorded | 69aa5f712d603e6de3e4182a | | KTO | 38.12 | Not recorded | 69aa5f712d603e6de3e4182a | | SELECTIVE DPO | 38.02 | Not recorded | 69aa5f712d603e6de3e4182a | | IPO | 37.95 | Not recorded | 69aa5f712d603e6de3e4182a | | DPO | 37.24 | Not recorded | 69aa5f712d603e6de3e4182a | | SIMPO | 29.38 | Not recorded | 69aa5f712d603e6de3e4182a | | SELECTIVE DPO | 29.27 | Not recorded | 69aa5f712d603e6de3e4182a | | WPO | 28.16 | Not recorded | 69aa5f712d603e6de3e4182a | | IPO | 27.79 | Not recorded | 69aa5f712d603e6de3e4182a | | KTO | 22.94 | Not recorded | 69aa5f712d603e6de3e4182a | | DPO | 18.53 | Not recorded | 69aa5f712d603e6de3e4182a | ### VRM-PPO Win Rate: 48.14. Result ID: 69aa8f416dbb7075b9544a4f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### VRM-PPO Win Rate: 45.3. Result ID: 69aa8f416dbb7075b9544a47. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### PPO Win Rate: 44.75. Result ID: 69aa8f416dbb7075b9544a4e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### WPO Win Rate: 44.11. Result ID: 69aa8f416dbb7075b9544a44. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### SIMPO Win Rate: 40.03. Result ID: 69aa8f416dbb7075b9544a43. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### PPO Win Rate: 39.52. Result ID: 69aa8f416dbb7075b9544a46. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### KTO Win Rate: 38.12. Result ID: 69aa8f416dbb7075b9544a42. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### SELECTIVE DPO Win Rate: 38.02. Result ID: 69aa8f416dbb7075b9544a45. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### IPO Win Rate: 37.95. Result ID: 69aa8f416dbb7075b9544a41. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### DPO Win Rate: 37.24. Result ID: 69aa8f416dbb7075b9544a40. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### SIMPO Win Rate: 29.38. Result ID: 69aa8f416dbb7075b9544a4b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### SELECTIVE DPO Win Rate: 29.27. Result ID: 69aa8f416dbb7075b9544a4d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### WPO Win Rate: 28.16. Result ID: 69aa8f416dbb7075b9544a4c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### IPO Win Rate: 27.79. Result ID: 69aa8f416dbb7075b9544a49. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### KTO Win Rate: 22.94. Result ID: 69aa8f416dbb7075b9544a4a. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. ### DPO Win Rate: 18.53. Result ID: 69aa8f416dbb7075b9544a48. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen 3-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.