learning_rate: 5e-05
train_batch_size: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 100

Training Loss	Epoch	Step	Validation Loss	Chrf++	Gen Len
20.0278	2.0	44	1.9085	11.6307	103.4882
20.0278	4.0	88	1.2987	12.6339	230.1422
14.5551	6.0	132	1.0555	18.8934	178.436
14.5551	8.0	176	0.9047	25.6151	150.3128
6.9796	10.0	220	0.7944	36.3272	113.5829
6.9796	12.0	264	0.7178	43.0554	104.2417
4.1223	14.0	308	0.6743	45.7281	102.0664
4.1223	16.0	352	0.6432	51.2143	93.0095
4.1223	18.0	396	0.6393	54.3167	97.8009
2.2888	20.0	440	0.6764	55.7	101.1327
2.2888	22.0	484	0.6996	58.5311	101.2654
1.1181	24.0	528	0.7424	58.8962	100.0948
1.1181	26.0	572	0.7772	59.9757	96.6919
0.4714	28.0	616	0.8347	59.548	96.6445
0.4714	30.0	660	0.8696	60.6217	97.128
0.2344	32.0	704	0.8505	61.5551	93.4739
0.2344	34.0	748	0.9164	62.1314	94.0806
0.2344	36.0	792	0.9760	62.0131	96.8294
0.1385	38.0	836	0.9059	62.0338	93.9336
0.1385	40.0	880	0.9923	62.1259	96.6351

Tsedeniya/finetuned_glosslm_jje_joint

finetuned_glosslm_jje_joint