pgajo/mdeberta_EW-TT-PE_U0_S1_Tingredient_P0.25_DROP1_mdeberta_E9_DEV97.0

Model description:

Model: microsoft/mdeberta-v3-base

Dataset: TASTEset

Unshuffled ratio: ['0']

Shuffled ratio: ['1']

Best exact match epoch: 9

Best exact match: 96.98

Best epoch: 9

Drop duplicates: ['1']

Max epochs = 10

Optimizer lr = 3e-05

Optimizer eps = 1e-08

Batch size = 8

Dataset path = pgajo/EW-TT-PE_U0_S1_Tingredient_P0.25_DROP1_mdeberta

Results

epoch	train_loss	train_f1	train_exact	dev_loss	dev_f1	dev_exact	test_loss	test_f1	test_exact
1	1.41	66.06	58.82	0.26	94.64	90.93	0	0	0
2	0.17	95.69	93.18	0.2	96.46	94.78	0	0	0
3	0.06	98.31	97.45	0.19	97.22	95.05	0	0	0
4	0.05	98.68	97.93	0.22	96.47	94.78	0	0	0
5	0.03	99.55	99.17	0.23	97	95.33	0	0	0
6	0.04	99.02	98.55	0.24	97.67	95.6	0	0	0
7	0.03	99.34	98.97	0.21	96.57	94.78	0	0	0
8	0.04	99.02	98.55	0.22	96.37	94.23	0	0	0
9	0.02	99.52	99.24	0.19	98.17	96.98	0	0	0
10	0.01	99.68	99.52	0.24	96.08	94.23	0	0	0