Table 9 Comparison of performance of intermediate models on in-domain test sets
From: LEIA: Linguistic Embeddings for the Identification of Affect
| Â | User | Temporal | Random | Average rank |
|---|---|---|---|---|
BERTweet-base | 72.82 [72.72,72.91] | 72.85 [72.77,72.95] | 72.92 [72.81,73.0] | 3.00 |
BERTweet-base+eMLM | 72.87 [72.79,72.97] | 72.91 [72.82,73.00] | 73.00 [72.92,73.09] | 2.00 |
LEIA-base | 72.92 [72.82,73.02] | 73.03 [72.95,73.11] | 73.02 [72.94,73.12] | 1.00 |
BERTweet-large | 73.03 [72.94,73.11] | 73.00 [72.91,73.12] | 73.19 [73.09,73.28] | 3.00 |
BERTweet-large+eMLM | 73.19 [73.11,73.28] | 73.14 [73.05,73.23] | 73.37 [73.27,73.46] | 2.00 |
LEIA-large | 73.37 [73.28,73.46] | 73.43 [73.34,73.53] | 73.57 [73.48,73.66] | 1.00 |