learning_rate: 2e-05
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 16
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 7

Training Loss	Epoch	Step	Validation Loss	Accuracy
49.4547	0.3682	100	3.0903	0.0458
49.3329	0.7365	200	3.0859	0.0782
47.6099	1.1031	300	2.9656	0.2648
45.1267	1.4713	400	2.8111	0.3064
42.5327	1.8396	500	2.7015	0.3761
40.5063	2.2062	600	2.5866	0.4330
38.8539	2.5745	700	2.4669	0.4448
36.6534	2.9427	800	2.3743	0.4864
33.2226	3.3093	900	2.2808	0.5179
33.9870	3.6776	1000	2.1968	0.5494
31.2448	4.0442	1100	2.1366	0.5491
31.5891	4.4124	1200	2.0896	0.5558
26.9491	4.7807	1300	2.0369	0.5758
27.5674	5.1473	1400	1.9648	0.5767
23.4261	5.5155	1500	1.9436	0.5827
27.9778	5.8838	1600	1.9284	0.5879
25.1437	6.2504	1700	1.8858	0.5915
26.5106	6.6186	1800	1.8804	0.5994
24.5716	6.9869	1900	1.8738	0.6052

Saad-Wazir24/indic-slid-mhubert

indic-slid-mhubert