Age prediction on CHILDES (child-grouped 5-fold CV)
1.03MAEGemini 1.5 Pro
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 1.5 ProFeature set=Child utterances, Prediction Type=Direct LLM Prediction2026.02 | 1.03 | 2.25 | 0.51 | |
| Expansion + Independence + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.04 | 2.06 | 0.55 | |
| Expansion + IndependenceBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.05 | 2.09 | 0.54 | |
| Expansion + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.12 | 2.42 | 0.47 | |
| Expansion + Independence + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.13 | 2.38 | 0.48 | |
| Expansion + IndependenceBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.14 | 2.41 | 0.47 | |
| ExpansionBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.14 | 2.47 | 0.46 | |
| Expansion + Independence + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.19 | 2.71 | 0.41 | |
| Expansion + IndependenceBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.2 | 2.75 | 0.4 | |
| ExpansionBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.2 | 2.75 | 0.4 | |
| Expansion + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.2 | 2.73 | 0.41 | |
| Expansion + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.21 | 2.71 | 0.4 | |
| Expansion + Independence + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.22 | 2.89 | 0.37 | |
| ExpansionBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.22 | 2.76 | 0.39 | |
| Expansion + IndependenceBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.23 | 2.97 | 0.36 | |
| ExpansionBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.23 | 2.97 | 0.36 | |
| Expansion + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.23 | 2.97 | 0.36 | |
| Independence + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.23 | 2.72 | 0.4 | |
| IndependenceBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.24 | 2.76 | 0.39 | |
| Mixtral-8x7B-Inst.Feature set=Child utterances, Prediction Type=Direct LLM Prediction2026.02 | 1.3 | 2.75 | 0.51 | |
| Independence + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.33 | 3.13 | 0.32 | |
| IndependenceBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.33 | 3.03 | 0.33 | |
| Independence + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.33 | 3.03 | 0.33 | |
| IndependenceBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.34 | 3.17 | 0.31 | |
| Independence + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.34 | 3.33 | 0.27 | |
| IndependenceBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.35 | 3.38 | 0.26 | |
| MLU + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.4 | 3.41 | 0.25 | |
| GFI + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.4 | 3.31 | 0.27 | |
| MLUBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.42 | 3.47 | 0.23 | |
| GFIBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.43 | 3.43 | 0.25 | |
| MLU + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.5 | 4.15 | 0.09 | |
| MLU + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.5 | 3.91 | 0.14 | |
| MLU + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.51 | 3.78 | 0.18 | |
| MLUBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.52 | 4.25 | 0.07 | |
| GFIBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.52 | 3.8 | 0.18 | |
| GFI + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.52 | 3.75 | 0.19 | |
| MLUBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.53 | 3.84 | 0.17 | |
| MLUBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.55 | 4.1 | 0.09 | |
| GFI + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.56 | 3.87 | 0.15 | |
| GFIBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.57 | 3.92 | 0.14 | |
| FKGL + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.63 | 4.13 | 0.09 | |
| GFI + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.64 | 4.17 | 0.1 | |
| GFIBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.66 | 4.22 | 0.09 | |
| FKGLBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.67 | 4.27 | 0.06 | |
| FKGL + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.69 | 4.31 | 0.05 | |
| vocd-D + PTBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.7 | 4.35 | 0.04 | |
| FKGL + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.71 | 4.47 | 0.03 | |
| vocd-D + PTBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.71 | 4.41 | 0.03 | |
| vocd-D + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.72 | 4.47 | 0.03 | |
| FKGLBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.72 | 4.4 | 0.03 | |
| FKGL + PTBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.73 | 4.46 | 0.03 | |
| FKGLBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.73 | 4.56 | 0.01 | |
| vocd-DBackbone=Gemini 1.5 Pro, Regressor=Linear Regression2026.02 | 1.75 | 4.55 | 0 | |
| vocd-DBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.76 | 4.58 | 0.01 | |
| FKGLBackbone=Mixtral-8x7B-Inst., Regressor=Linear Regression2026.02 | 1.76 | 4.55 | 0.02 | |
| vocd-D + PTBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.76 | 4.63 | 0 | |
| vocd-DBackbone=Gemini 1.5 Pro, Regressor=Gradient Boosting2026.02 | 1.78 | 4.66 | -0.03 | |
| vocd-DBackbone=Mixtral-8x7B-Inst., Regressor=Gradient Boosting2026.02 | 1.81 | 4.87 | -0.06 |