# Continual Generative Information Retrieval on MS MARCO Source: [Continual Generative Information Retrieval on MS MARCO]() Collection index: [llms.txt]() Data updated: 2026-04-28T08:00:32.082Z Evaluation of Continual Generative Information Retrieval performance on the MS MARCO dataset, reporting metrics related to continual learning such as FWT and BWT. 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 | Continual Generative Information Retrieval | | Dataset (including any stated split) | MS MARCO | | Metric | Score D0 | | 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 - [Score D0](): Higher is better; unit: PERCENT. - [Score D1](): Higher is better; unit: PERCENT. - [Score D2](): Higher is better; unit: PERCENT. - [Score D3](): Higher is better; unit: PERCENT. - [Score D4](): Higher is better; unit: PERCENT. - [Score D5](): Higher is better; unit: PERCENT. - [AP](): Higher is better; unit: PERCENT. - [FWT](): Higher is better; unit: PERCENT. - [BWT](): Higher is better; unit: SCALAR. ## Recorded results | Method | Score D0 | Result date | Report paper | | --- | --- | --- | --- | | DPR-HN | 74.35 | Not recorded | 69f03ef155a3c32af89f9f4a | | DPR | 72.95 | Not recorded | 69f03ef155a3c32af89f9f4a | | LoRA-SPQ | 69.21 | Not recorded | 69f03ef155a3c32af89f9f4a | | Full FT-SPQ | 69.21 | Not recorded | 69f03ef155a3c32af89f9f4a | | PAMT-LoRA-SPQ | 69.21 | Not recorded | 69f03ef155a3c32af89f9f4a | | PAMT-Full FT-SPQ | 69.21 | Not recorded | 69f03ef155a3c32af89f9f4a | | BM25 | 67.73 | Not recorded | 69f03ef155a3c32af89f9f4a | | CLEVER | 66.01 | Not recorded | 69f03ef155a3c32af89f9f4a | | MixLoRA-DSI | 65.27 | Not recorded | 69f03ef155a3c32af89f9f4a | | DSI++ | 64.53 | Not recorded | 69f03ef155a3c32af89f9f4a | | CorpusBrain++ | 63.79 | Not recorded | 69f03ef155a3c32af89f9f4a | | LoRA-TU | 61.82 | Not recorded | 69f03ef155a3c32af89f9f4a | | Full FT-TU | 61.82 | Not recorded | 69f03ef155a3c32af89f9f4a | | PAMT-LoRA-TU | 61.82 | Not recorded | 69f03ef155a3c32af89f9f4a | | PAMT-Full FT-TU | 61.82 | Not recorded | 69f03ef155a3c32af89f9f4a | | PromptDSI | 61.33 | Not recorded | 69f03ef155a3c32af89f9f4a | ### DPR-HN Score D0: 74.35. Result ID: 69f0684f47f34ce924d61824. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Expanded protocol","Decoding Strategy":"constrained decoding"}. This record does not assert independent reproduction. ### DPR Score D0: 72.95. Result ID: 69f0684f47f34ce924d61823. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Evaluation Protocol":"Expanded protocol","Decoding Strategy":"constrained decoding"}. This record does not assert independent reproduction. ### LoRA-SPQ Score D0: 69.21. Result ID: 69f0684f47f34ce924d6182a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 1: Base adaptation","Tuning Strategy":"LoRA","DocID Design":"SPQ"}. This record does not assert independent reproduction. ### Full FT-SPQ Score D0: 69.21. Result ID: 69f0684f47f34ce924d6182c. [Reporting paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 1: Base adaptation","Tuning Strategy":"Full FT","DocID Design":"SPQ"}. This record does not assert independent reproduction. ### PAMT-LoRA-SPQ Score D0: 69.21. Result ID: 69f0684f47f34ce924d6182e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 2: Post-Adaptation Memory Tuning","Tuning Strategy":"LoRA","DocID Design":"SPQ"}. This record does not assert independent reproduction. ### PAMT-Full FT-SPQ Score D0: 69.21. Result ID: 69f0684f47f34ce924d61830. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 2: Post-Adaptation Memory Tuning","Tuning Strategy":"Full FT","DocID Design":"SPQ"}. This record does not assert independent reproduction. ### BM25 Score D0: 67.73. Result ID: 69f0684f47f34ce924d61822. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Expanded protocol","Decoding Strategy":"constrained decoding"}. This record does not assert independent reproduction. ### CLEVER Score D0: 66.01. Result ID: 69f0684f47f34ce924d61826. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Non-PEFT continual GenIR baselines","Evaluation Protocol":"Expanded protocol"}. This record does not assert independent reproduction. ### MixLoRA-DSI Score D0: 65.27. Result ID: 69f0684f47f34ce924d61829. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"PEFT-based continual GenIR baselines","Evaluation Protocol":"Expanded protocol"}. This record does not assert independent reproduction. ### DSI++ Score D0: 64.53. Result ID: 69f0684f47f34ce924d61825. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Non-PEFT continual GenIR baselines","Evaluation Protocol":"Expanded protocol"}. This record does not assert independent reproduction. ### CorpusBrain++ Score D0: 63.79. Result ID: 69f0684f47f34ce924d61827. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"PEFT-based continual GenIR baselines","Evaluation Protocol":"Expanded protocol"}. This record does not assert independent reproduction. ### LoRA-TU Score D0: 61.82. Result ID: 69f0684f47f34ce924d6182b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 1: Base adaptation","Tuning Strategy":"LoRA","DocID Design":"TU"}. This record does not assert independent reproduction. ### Full FT-TU Score D0: 61.82. Result ID: 69f0684f47f34ce924d6182d. [Reporting paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 1: Base adaptation","Tuning Strategy":"Full FT","DocID Design":"TU"}. This record does not assert independent reproduction. ### PAMT-LoRA-TU Score D0: 61.82. Result ID: 69f0684f47f34ce924d6182f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 2: Post-Adaptation Memory Tuning","Tuning Strategy":"LoRA","DocID Design":"TU"}. This record does not assert independent reproduction. ### PAMT-Full FT-TU Score D0: 61.82. Result ID: 69f0684f47f34ce924d61831. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Adaptation Stage":"Stage 2: Post-Adaptation Memory Tuning","Tuning Strategy":"Full FT","DocID Design":"TU"}. This record does not assert independent reproduction. ### PromptDSI Score D0: 61.33. Result ID: 69f0684f47f34ce924d61828. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"PEFT-based continual GenIR baselines","Evaluation Protocol":"Expanded protocol"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.