learning_rate: 1e-06
train_batch_size: 8
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 100
num_epochs: 15

Training Loss	Epoch	Step	Validation Loss	Wer
0.2878	0.4235	500	0.3000	0.3275
0.2849	0.8471	1000	0.2957	0.3254
0.3528	1.2702	1500	0.2972	0.3286
0.3533	1.6938	2000	0.2954	0.3239
0.2433	2.1169	2500	0.2951	0.3241
0.2103	2.5404	3000	0.2929	0.3251
0.2582	2.9640	3500	0.2938	0.3214
0.3428	3.3871	4000	0.2926	0.3246
0.3324	3.8107	4500	0.2919	0.3216
0.3029	4.2338	5000	0.2894	0.3204
0.3213	4.6573	5500	0.2882	0.3210
0.2279	5.0805	6000	0.2877	0.3212
0.2033	5.5040	6500	0.2875	0.3191
0.191	5.9276	7000	0.2859	0.3184
0.3466	6.3507	7500	0.2845	0.3180
0.3352	6.7742	8000	0.2843	0.3172
0.2328	7.1974	8500	0.2829	0.3153
0.2839	7.6209	9000	0.2846	0.3170
0.2207	8.0440	9500	0.2841	0.3169
0.2749	8.4676	10000	0.2832	0.3169
0.274	8.8911	10500	0.2824	0.3158
0.244	9.3143	11000	0.2820	0.3138
0.2899	9.7378	11500	0.2817	0.3152
0.1984	10.1609	12000	0.2808	0.3135
0.196	10.5845	12500	0.2802	0.3137
0.2524	11.0076	13000	0.2796	0.3131
0.3053	11.4312	13500	0.2807	0.3123
0.2747	11.8547	14000	0.2812	0.3140
0.2456	12.2778	14500	0.2813	0.3138
0.2761	12.7014	15000	0.2796	0.3133
0.2046	13.1245	15500	0.2801	0.3136
0.2134	13.5481	16000	0.2802	0.3135
0.2249	13.9716	16500	0.2805	0.3134
0.2934	14.3947	17000	0.2799	0.3123
0.3021	14.8183	17500	0.2799	0.3118

SergeyAdamyan/outputs_finetune2

outputs_finetune2