# Lyric Generation on Lyric Generation Dataset (test) Source: [Lyric Generation on Lyric Generation Dataset (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:04:16.185Z Evaluation of models performing Lyric Generation on the test split of the dataset, assessed using metrics like Sent/Corpus BLEU, Perplexity (PPL), and distributional metrics. 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 | Lyric Generation | | Dataset (including any stated split) | Lyric Generation (test) | | Metric | Sent BLEU | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 - [Sent BLEU](): Higher is better; unit: SCALAR. - [Salient Word Coverage](): Higher is better; unit: SCALAR. - [Corpus BLEU](): Higher is better; unit: SCALAR. - [Distinct-1](): Higher is better; unit: SCALAR. - [Distinct-2](): Higher is better; unit: SCALAR. - [PPL](): Lower is better; unit: SCALAR. - [Cropped Sentence Accuracy](): Higher is better; unit: PERCENT. - [Stress Duration Ratio](): Higher is better; unit: PERCENT. - [OVL](): Higher is better; unit: PERCENT. - [REL](): Higher is better; unit: PERCENT. ## Recorded results | Method | Sent BLEU | Result date | Report paper | | --- | --- | --- | --- | | Ground Truth | 1 | Not recorded | 68fa71a51e209131cea52470 | | LYRA | 0.075 | Not recorded | 68fa71a51e209131cea52470 | | LYRA | 0.074 | Not recorded | 68fa71a51e209131cea52470 | | LYRA | 0.071 | Not recorded | 68fa71a51e209131cea52470 | | GPT-2 | 0.049 | Not recorded | 68fa71a51e209131cea52470 | | SongMASS | 0.045 | Not recorded | 68fa71a51e209131cea52470 | | GPT-2 | 0.026 | Not recorded | 68fa71a51e209131cea52470 | ### Ground Truth Sent BLEU: 1. Result ID: 6991d7dd3a6c3148490af132. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### LYRA Sent BLEU: 0.075. Result ID: 6991d7dd3a6c3148490af130. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Rhythm Constraint":"soft"}. This record does not assert independent reproduction. ### LYRA Sent BLEU: 0.074. Result ID: 6991d7dd3a6c3148490af12f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Rhythm Constraint":"none"}. This record does not assert independent reproduction. ### LYRA Sent BLEU: 0.071. Result ID: 6991d7dd3a6c3148490af131. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Rhythm Constraint":"hard"}. This record does not assert independent reproduction. ### GPT-2 Sent BLEU: 0.049. Result ID: 6991d7dd3a6c3148490af12e. [Reporting paper]() Recorded evaluation configuration: {"Input Mode":"content-to-lyrics"}. This record does not assert independent reproduction. ### SongMASS Sent BLEU: 0.045. Result ID: 6991d7dd3a6c3148490af12c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-2 Sent BLEU: 0.026. Result ID: 6991d7dd3a6c3148490af12d. [Reporting paper]() Recorded evaluation configuration: {"Input Mode":"title-to-lyrics"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.