Modelpublic

boyuzhuGPT/qwen2_5_omni_wo_image_1016

sourceHugging Faceupdated 11mo agoView on Hugging Face

0likes19downloads

Model Card

qwen25omniwoimage_1016

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.6879
—Token Acc: 0.7939

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

—learning_rate: 0.0001
—trainbatchsize: 2
—evalbatchsize: 1
—seed: 42
—distributed_type: multi-GPU
—num_devices: 4
—gradientaccumulationsteps: 4
—totaltrainbatch_size: 32
—totalevalbatch_size: 4
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: cosine
—lrschedulerwarmup_ratio: 0.05
—num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Token Acc
0.8837	0.0621	50	0.9070	0.7320
0.9825	0.1243	100	1.0099	0.7151
1.2358	0.1864	150	1.1751	0.6913
1.0385	0.2485	200	1.0784	0.7095
1.0386	0.3107	250	1.0482	0.7152
1.0314	0.3728	300	1.0402	0.7176
0.9869	0.4349	350	1.0204	0.7217
0.9802	0.4970	400	1.0088	0.7242
0.9851	0.5592	450	1.0031	0.7239
0.9746	0.6213	500	0.9881	0.7292
0.9575	0.6834	550	0.9752	0.7313
0.9403	0.7456	600	0.9664	0.7338
0.9196	0.8077	650	0.9514	0.7363
0.9093	0.8698	700	0.9412	0.7378
0.9110	0.9320	750	0.9402	0.7389
0.8727	0.9941	800	0.9215	0.7431
0.8155	1.0559	850	0.9105	0.7454
0.8168	1.1180	900	0.8992	0.7480
0.8112	1.1802	950	0.8922	0.7495
0.7726	1.2423	1000	0.8813	0.7522
0.7514	1.3044	1050	0.8727	0.7538
0.7862	1.3666	1100	0.8581	0.7556
0.7877	1.4287	1150	0.8441	0.7591
0.7275	1.4908	1200	0.8387	0.7598
0.7558	1.5530	1250	0.8255	0.7623
0.7436	1.6151	1300	0.8165	0.7648
0.7110	1.6772	1350	0.8054	0.7679
0.6854	1.7394	1400	0.7955	0.7700
0.6878	1.8015	1450	0.7833	0.7723
0.7004	1.8636	1500	0.7732	0.7738
0.6571	1.9258	1550	0.7640	0.7753
0.6846	1.9879	1600	0.7523	0.7785
0.5994	2.0497	1650	0.7508	0.7803
0.5861	2.1118	1700	0.7425	0.7813
0.5742	2.1740	1750	0.7393	0.7832
0.5990	2.2361	1800	0.7295	0.7847
0.5469	2.2982	1850	0.7195	0.7866
0.5579	2.3604	1900	0.7150	0.7883
0.5455	2.4225	1950	0.7067	0.7895
0.5439	2.4846	2000	0.7056	0.7901
0.5416	2.5468	2050	0.7000	0.7913
0.5406	2.6089	2100	0.6945	0.7923
0.5312	2.6710	2150	0.6933	0.7930
0.5309	2.7331	2200	0.6903	0.7936
0.5037	2.7953	2250	0.6899	0.7935
0.5516	2.8574	2300	0.6888	0.7939
0.5568	2.9195	2350	0.6879	0.7940
0.5369	2.9817	2400	0.6879	0.7938
0.5204	3.0	2415	0.6879	0.7939

Framework versions

—Transformers 4.57.1
—Pytorch 2.8.0+cu128
—Datasets 3.6.0
—Tokenizers 0.22.1