learning_rate: 3e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 32
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 0.1
num_epochs: 40

Training Loss	Epoch	Step	Validation Loss
3.6175	1.0	231	0.9077
3.6304	2.0	462	0.9055
3.6063	3.0	693	0.9035
3.6222	4.0	924	0.9026
3.6103	5.0	1155	0.9017
3.5776	6.0	1386	0.9024
3.5583	7.0	1617	0.9009
3.5296	8.0	1848	0.9007
3.5117	9.0	2079	0.9014
3.4398	10.0	2310	0.9011
3.5154	11.0	2541	0.9000
3.5240	12.0	2772	0.9005
3.5288	13.0	3003	0.9002
3.5014	14.0	3234	0.8997
3.5239	15.0	3465	0.9003
3.5119	16.0	3696	0.8996
3.5031	17.0	3927	0.8996
3.5365	18.0	4158	0.8996
3.5471	19.0	4389	0.8999
3.4429	20.0	4620	0.8998
3.5300	21.0	4851	0.8994
3.5197	22.0	5082	0.8997
3.5463	23.0	5313	0.8995
3.5332	24.0	5544	0.8997
3.5333	25.0	5775	0.8996
3.5344	26.0	6006	0.8996

SociauxLing/modernbert-CGEdit-AAE_sv3cg_d3_final

modernbert-CGEdit-AAE_sv3cg_d3_final