learning_rate: 0.0001
train_batch_size: 2
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 4
total_train_batch_size: 64
total_eval_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.05
num_epochs: 3.0

Training Loss	Epoch	Step	Validation Loss	Token Acc
0.2664	0.1488	50	0.3178	0.8960
0.2958	0.2976	100	0.2920	0.9015
0.2451	0.4464	150	0.2596	0.9250
0.2512	0.5952	200	0.3136	0.9216
0.2455	0.7440	250	0.2657	0.9259
0.2551	0.8929	300	0.2599	0.9272
0.2212	1.0417	350	0.2731	0.9285
0.1757	1.1905	400	0.2444	0.9300
0.1924	1.3393	450	0.2363	0.9310
0.1829	1.4881	500	0.2277	0.9333
0.1676	1.6369	550	0.2251	0.9348
0.1685	1.7857	600	0.2194	0.9361
0.1608	1.9345	650	0.2097	0.9381
0.1427	2.0833	700	0.2073	0.9385
0.1434	2.2321	750	0.2019	0.9401
0.1523	2.3810	800	0.1978	0.9406
0.1251	2.5298	850	0.1946	0.9415
0.1275	2.6786	900	0.1930	0.9424
0.1367	2.8274	950	0.1918	0.9425
0.1388	2.9762	1000	0.1911	0.9429
0.1290	3.0	1008	0.1910	0.9428

boyuzhuGPT/qwen2_5_omni_all_1015_wo_explanation

qwen2_5_omni_all_1015_wo_explanation