learning_rate: 0.0002
train_batch_size: 1
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 4
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
lr_scheduler_warmup_steps: 0.05
num_epochs: 5
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
0.3323	0.2334	200	0.3159
0.2567	0.4667	400	0.3054
0.2647	0.7001	600	0.2995
0.2712	0.9335	800	0.2965
0.2769	1.1669	1000	0.2949
0.1598	1.4002	1200	0.2903
0.2023	1.6336	1400	0.2826
0.2220	1.8670	1600	0.2745
0.2647	2.1004	1800	0.2824
0.1101	2.3337	2000	0.2801
0.1838	2.5671	2200	0.2776
0.1373	2.8005	2400	0.2733
0.0850	3.0338	2600	0.2818
0.0677	3.2672	2800	0.2922
0.0742	3.5006	3000	0.2883
0.0713	3.7340	3200	0.2863
0.0446	3.9673	3400	0.2899
0.0409	4.2007	3600	0.3104
0.0583	4.4341	3800	0.3127
0.0383	4.6674	4000	0.3119
0.0237	4.9008	4200	0.3121
0.0418	5.0	4285	0.3121

oberbics/llama-3.1-base-kg-extraction

llama-3.1-base-kg-extraction