Table 7 Macro-F1 of LEIA-base, LEIA-large, GPT-3.5, and GPT-4 on the out-of-domain test sets
From: LEIA: Linguistic Embeddings for the Identification of Affect
| Â | LEIA-base | LEIA-large | GPT-3.5 | GPT-4 |
|---|---|---|---|---|
Universal Joy | 54.18 [48.79,59.88] | 54.17 [48.68,59.84] | 52.89 [47.8,58.33] | 56.43 [51.94,60.9] |
GoEmotions | 46.31 [43.98,48.72] | 45.75 [43.45,48.09] | 59.06 [56.44,61.67] | 56.45 [53.97,58.95] |
TEC | 43.87 [42.05,45.61] | 44.12 [42.34,45.89] | 52.66 [50.7,54.59] | 54.82 [52.96,56.71] |
SemEval | 71.68 [69.18,74.19] | 70.04 [67.48,72.52] | 80.13 [77.9,82.31] | 81.72 [79.57,83.81] |
enISEAR | 70.37 [66.63,74.01] | 79.94 [76.69,83.14] | 84.96 [81.96,87.77] | 89.97 [87.47,92.32] |