learning_rate: 0.0003
train_batch_size: 64
eval_batch_size: 128
seed: 2024
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 512
total_eval_batch_size: 1024
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: cosine_with_restarts
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3

Training Loss	Epoch	Step	Validation Loss	F1	Accuracy	Precision	Recall
No log	0	0	0.0114	0.8209	0.821	0.8211	0.821
0.0103	0.3000	7956	0.0116	0.8576	0.8589	0.8705	0.8589
0.0091	0.5999	15912	0.0091	0.8945	0.8945	0.8945	0.8945
0.0097	0.8999	23868	0.0096	0.8874	0.8874	0.8880	0.8874
0.0078	1.1999	31824	0.0088	0.8957	0.8957	0.8957	0.8957
0.0174	1.4998	39780	0.0174	0.6024	0.6113	0.6195	0.6113
0.0136	1.7998	47736	0.0134	0.7344	0.7344	0.7344	0.7344
0.0129	2.0998	55692	0.0131	0.7370	0.7408	0.7531	0.7408
0.0125	2.3997	63648	0.0125	0.7530	0.753	0.7530	0.753
0.0129	2.6997	71604	0.0124	0.7724	0.7724	0.7724	0.7724
0.0125	2.9997	79560	0.0124	0.7754	0.7754	0.7754	0.7754

param-bharat/ModernBERT-large-nli-scorer

ModernBERT-large-nli-clf