learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
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
num_epochs: 25
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time
0.469	1.0	75	6.1201	0.0051
0.4956	2.0	150	5.7622	0.0051
0.4658	3.0	225	5.7691	0.0051
0.4425	4.0	300	6.0003	0.0051
0.4247	5.0	375	5.8873	0.0051
0.4173	6.0	450	6.0859	0.0051
0.4035	7.0	525	6.1184	0.0051
0.3951	8.0	600	6.2805	0.0051
0.3864	9.0	675	6.2936	0.0051
0.3768	10.0	750	6.3689	0.0051
0.3744	11.0	825	6.3023	0.0051
0.3653	12.0	900	6.3922	0.0051
0.3615	13.0	975	6.4754	0.0051
0.358	14.0	1050	6.4871	0.0051
0.3558	15.0	1125	6.6395	0.0051
0.349	16.0	1200	6.7443	0.0051
0.3443	17.0	1275	6.8492	0.0051
0.3417	18.0	1350	6.7747	0.0051
0.341	19.0	1425	6.9504	0.0051
0.338	20.0	1500	6.9641	0.0051
0.3336	21.0	1575	7.0945	0.0051
0.3279	22.0	1650	7.0730	0.0051
0.3287	23.0	1725	7.2272	0.0051
0.3259	24.0	1800	7.2886	0.0051
0.3225	25.0	1875	7.3240	0.0051

zbigi/Llama-3.2-3B-elon-ai-course

Llama-3.2-3B-elon-ai-course