# Constraint-following Instruction Evaluation on IFEval Source: [Constraint-following Instruction Evaluation on IFEval]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:35:47.122Z Evaluation of model performance on following specific constraints within instructions, using the IFEval 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 | Constraint-following Instruction Evaluation | | Dataset (including any stated split) | IFEval | | Metric | Average Score | | Unit | SCALAR | | 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 - [Average Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | Average Score | Result date | Report paper | | --- | --- | --- | --- | | LLAMA3-INSTRUCT w/ ULTRAFEEDBACK | 54.4 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-INSTRUCT w/ SSO\_DPO | 53.4 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-INSTRUCT w/ PBAA\_DPO | 53.2 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-INSTRUCT | 53 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-INSTRUCT w/ ULTRAFEEDBACK | 51.5 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-INSTRUCT | 51.4 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-INSTRUCT w/ SSO\_DPO | 51.4 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-INSTRUCT w/ PBAA\_DPO | 50.9 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-SFT w/ SSO\_DPO | 50.3 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-SFT w/ PBAA\_DPO | 47.8 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-SFT w/ SSO\_DPO | 45.7 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-SFT w/ ULTRAFEEDBACK | 43.6 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-SFT w/ PBAA\_DPO | 43.6 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-SFT w/ ULTRAFEEDBACK | 40.4 | Not recorded | 68fa795d1e209131cea7d471 | | LLAMA3-SFT | 24.9 | Not recorded | 68fa795d1e209131cea7d471 | | QWEN2-SFT | 19.8 | Not recorded | 68fa795d1e209131cea7d471 | ### LLAMA3-INSTRUCT w/ ULTRAFEEDBACK Average Score: 54.4. Result ID: 698b48d78d67a41d0a6cd067. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-INSTRUCT"}. This record does not assert independent reproduction. ### LLAMA3-INSTRUCT w/ SSO\_DPO Average Score: 53.4. Result ID: 698b48d78d67a41d0a6cd069. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-INSTRUCT"}. This record does not assert independent reproduction. ### LLAMA3-INSTRUCT w/ PBAA\_DPO Average Score: 53.2. Result ID: 698b48d78d67a41d0a6cd068. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-INSTRUCT"}. This record does not assert independent reproduction. ### LLAMA3-INSTRUCT Average Score: 53. Result ID: 698b48d78d67a41d0a6cd066. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLAMA3-INSTRUCT"}. This record does not assert independent reproduction. ### QWEN2-INSTRUCT w/ ULTRAFEEDBACK Average Score: 51.5. Result ID: 698b48d78d67a41d0a6cd06b. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-INSTRUCT"}. This record does not assert independent reproduction. ### QWEN2-INSTRUCT Average Score: 51.4. Result ID: 698b48d78d67a41d0a6cd06a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"QWEN2-INSTRUCT"}. This record does not assert independent reproduction. ### QWEN2-INSTRUCT w/ SSO\_DPO Average Score: 51.4. Result ID: 698b48d78d67a41d0a6cd06d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-INSTRUCT"}. This record does not assert independent reproduction. ### QWEN2-INSTRUCT w/ PBAA\_DPO Average Score: 50.9. Result ID: 698b48d78d67a41d0a6cd06c. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-INSTRUCT"}. This record does not assert independent reproduction. ### LLAMA3-SFT w/ SSO\_DPO Average Score: 50.3. Result ID: 698b48d78d67a41d0a6cd061. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-SFT"}. This record does not assert independent reproduction. ### LLAMA3-SFT w/ PBAA\_DPO Average Score: 47.8. Result ID: 698b48d78d67a41d0a6cd060. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-SFT"}. This record does not assert independent reproduction. ### QWEN2-SFT w/ SSO\_DPO Average Score: 45.7. Result ID: 698b48d78d67a41d0a6cd065. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-SFT"}. This record does not assert independent reproduction. ### LLAMA3-SFT w/ ULTRAFEEDBACK Average Score: 43.6. Result ID: 698b48d78d67a41d0a6cd05f. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-SFT"}. This record does not assert independent reproduction. ### QWEN2-SFT w/ PBAA\_DPO Average Score: 43.6. Result ID: 698b48d78d67a41d0a6cd064. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-SFT"}. This record does not assert independent reproduction. ### QWEN2-SFT w/ ULTRAFEEDBACK Average Score: 40.4. Result ID: 698b48d78d67a41d0a6cd063. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-SFT"}. This record does not assert independent reproduction. ### LLAMA3-SFT Average Score: 24.9. Result ID: 698b48d78d67a41d0a6cd05e. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA3-SFT"}. This record does not assert independent reproduction. ### QWEN2-SFT Average Score: 19.8. Result ID: 698b48d78d67a41d0a6cd062. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"QWEN2-SFT"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.