# Multimodal Question Answering on CaReSound Source: [Multimodal Question Answering on CaReSound]() Collection index: [llms.txt]() Data updated: 2026-03-17T07:50:07.920Z Evaluation of model performance on the CaReSound dataset for Multimodal Question Answering, reporting Yes/No accuracy, F1 scores, and textual similarity metrics (ROUGE, METEOR, BERTScore). 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 | Multimodal Question Answering | | Dataset (including any stated split) | CaReSound | | Metric | Yes/No Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 - [Yes/No Accuracy](): Higher is better; unit: PERCENT. - [Yes/No F1 Score](): Higher is better; unit: PERCENT. - [Contains-Match Accuracy](): Higher is better; unit: PERCENT. - [ROUGE-L](): Higher is better; unit: PERCENT. - [METEOR](): Higher is better; unit: PERCENT. - [BERTScore](): Higher is better; unit: PERCENT. ## Recorded results | Method | Yes/No Accuracy | Result date | Report paper | | --- | --- | --- | --- | | AudioTSLM (1.4B) | 93.5 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.5 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.5 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.5 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.5 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.44 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 93.21 | Not recorded | 69b8e0090636e5bf4afbe9de | | CaReAQA (3B) | 93.12 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 92 | Not recorded | 69b8e0090636e5bf4afbe9de | | AudioTSLM | 91.1 | Not recorded | 69b8e0090636e5bf4afbe9de | | Audio-Flamingo3 (8B) | 85 | Not recorded | 69b8e0090636e5bf4afbe9de | | Qwen2.5-Omni (0.5B) | 84.62 | Not recorded | 69b8e0090636e5bf4afbe9de | | Qwen2-Audio (7B) | 17.83 | Not recorded | 69b8e0090636e5bf4afbe9de | ### AudioTSLM (1.4B) Yes/No Accuracy: 93.5. Result ID: 69b906fd47f34ce924a143ef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model type":"ALM"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.5. Result ID: 69b906fd47f34ce924a143f0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Encoder":"Raw"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.5. Result ID: 69b906fd47f34ce924a143f1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Encoder":"Mel"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.5. Result ID: 69b906fd47f34ce924a143f3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Encoder":"Whisper"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.5. Result ID: 69b906fd47f34ce924a143f5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Context":"30s"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.44. Result ID: 69b906fd47f34ce924a143f2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Encoder":"CLAP"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 93.21. Result ID: 69b906fd47f34ce924a143f4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Encoder":"Wav2Vec"}. This record does not assert independent reproduction. ### CaReAQA (3B) Yes/No Accuracy: 93.12. Result ID: 69b906fd47f34ce924a143ee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model type":"ALM"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 92. Result ID: 69b906fd47f34ce924a143f6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Context":"20s"}. This record does not assert independent reproduction. ### AudioTSLM Yes/No Accuracy: 91.1. Result ID: 69b906fd47f34ce924a143f7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Context":"10s"}. This record does not assert independent reproduction. ### Audio-Flamingo3 (8B) Yes/No Accuracy: 85. Result ID: 69b906fd47f34ce924a143eb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model type":"ALM"}. This record does not assert independent reproduction. ### Qwen2.5-Omni (0.5B) Yes/No Accuracy: 84.62. Result ID: 69b906fd47f34ce924a143ed. [Reporting paper]() Recorded evaluation configuration: {"Model type":"ALM"}. This record does not assert independent reproduction. ### Qwen2-Audio (7B) Yes/No Accuracy: 17.83. Result ID: 69b906fd47f34ce924a143ec. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model type":"ALM"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.