README.md

metadata

{}

Model description:

Model: pgajo/mbert-xlwa-en-it

Dataset: TASTEset

Unshuffled ratio: ['0']

Shuffled ratio: ['1']

Best exact match epoch: 7

Best exact match: 92.03

Best epoch: 7

Drop duplicates: ['1']

Max epochs = 10

Optimizer lr = 3e-05

Optimizer eps = 1e-08

Batch size = 32

Dataset path = pgajo/EW-TT-PE_U0_S1_Tingredient_P0.75_DROP1_mbert

Results

epoch	train_loss	train_f1	train_exact	dev_loss	dev_f1	dev_exact	test_loss	test_f1	test_exact
1	0.77	79.53	64.94	0.37	91.6	84.62	0	0	0
2	0.2	94.44	89.6	0.33	94.44	90.11	0	0	0
3	0.09	97.64	95.59	0.4	92.83	89.01	0	0	0
4	0.05	98.68	97.38	0.36	94.19	90.11	0	0	0
5	0.03	99.32	98.42	0.35	94.34	90.38	0	0	0
6	0.04	98.92	98.14	0.42	95	90.38	0	0	0
7	0.02	99.38	98.9	0.43	94.68	92.03	0	0	0
8	0.01	99.59	99.31	0.39	94.85	90.93	0	0	0
9	0.02	99.77	99.72	0.42	94.61	91.21	0	0	0
10	0.02	99.18	98.55	0.42	95.15	91.76	0	0	0