Timesformer_WLASL_100_200_epochs_p20_SR_16

This model is a fine-tuned version of facebook/timesformer-base-finetuned-k400 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.2599
• Top 1 Accuracy: 0.5828
• Top 5 Accuracy: 0.7899
• Top 10 Accuracy: 0.8698
• Accuracy: 0.5828
• Precision: 0.5806
• Recall: 0.5828
• F1: 0.5510

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: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 36000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Top 1 Accuracy	Top 5 Accuracy	Top 10 Accuracy	Accuracy	Precision	Recall	F1
19.1155	0.005	180	4.6927	0.0089	0.0414	0.0888	0.0089	0.0155	0.0089	0.0105
18.5538	1.0050	360	4.5821	0.0266	0.0769	0.1302	0.0266	0.0137	0.0266	0.0116
17.5848	2.0050	540	4.3988	0.0562	0.1450	0.2633	0.0562	0.0486	0.0562	0.0390
15.8283	3.0050	721	4.0516	0.1302	0.2959	0.4645	0.1302	0.1012	0.1302	0.0976
13.3102	4.005	901	3.6150	0.2249	0.4704	0.6154	0.2249	0.1781	0.2249	0.1741
11.2113	5.0050	1081	3.2389	0.2604	0.6065	0.7367	0.2604	0.2422	0.2604	0.2215
8.898	6.0050	1261	2.8714	0.3757	0.6775	0.8166	0.3757	0.3584	0.3757	0.3324
6.715	7.0050	1442	2.6518	0.4231	0.7249	0.8402	0.4231	0.3828	0.4231	0.3730
4.8442	8.005	1622	2.3294	0.4645	0.7929	0.8876	0.4645	0.5077	0.4645	0.4377
3.3825	9.0050	1802	2.1747	0.4911	0.7899	0.8964	0.4911	0.5436	0.4911	0.4654
2.0471	10.0050	1982	1.9990	0.5148	0.8107	0.9053	0.5178	0.5871	0.5178	0.5057
1.3242	11.0050	2163	1.8964	0.5473	0.8166	0.8935	0.5473	0.5822	0.5473	0.5199
0.8746	12.005	2343	1.8222	0.5562	0.8254	0.9083	0.5562	0.5796	0.5562	0.5320
0.5537	13.0050	2523	1.7525	0.5769	0.8343	0.9142	0.5769	0.5813	0.5769	0.5468
0.4081	14.0050	2703	1.7351	0.5947	0.8136	0.8964	0.5947	0.6684	0.5947	0.5834
0.17	15.0050	2884	1.6998	0.5592	0.8225	0.9083	0.5592	0.5763	0.5592	0.5342
0.2053	16.005	3064	1.7340	0.5651	0.8343	0.9083	0.5651	0.6215	0.5651	0.5390
0.1434	17.0050	3244	1.7350	0.6006	0.8432	0.9142	0.6006	0.6347	0.6006	0.5806
0.1957	18.0050	3424	1.8179	0.5621	0.8373	0.9142	0.5621	0.6060	0.5621	0.5350
0.1636	19.0050	3605	1.7831	0.6154	0.8225	0.8905	0.6154	0.6401	0.6154	0.5917
0.0908	20.005	3785	1.7552	0.6213	0.8402	0.9053	0.6213	0.6504	0.6213	0.6014
0.058	21.0050	3965	1.8422	0.6243	0.8254	0.9112	0.6213	0.6392	0.6213	0.5962
0.0924	22.0050	4145	1.8347	0.6006	0.8225	0.9201	0.6006	0.6218	0.6006	0.5735
0.0799	23.0050	4326	1.9650	0.6036	0.8107	0.8846	0.6036	0.6182	0.6036	0.5724
0.176	24.005	4506	1.9326	0.5858	0.8402	0.9142	0.5858	0.6240	0.5858	0.5671
0.0786	25.0050	4686	1.7753	0.6124	0.8491	0.9142	0.6124	0.6607	0.6124	0.5998
0.242	26.0050	4866	2.0219	0.5769	0.7722	0.8876	0.5769	0.6337	0.5769	0.5552
0.1767	27.0050	5047	1.9744	0.5828	0.8166	0.9024	0.5828	0.6330	0.5828	0.5721
0.14	28.005	5227	2.1996	0.5769	0.7811	0.8609	0.5769	0.5983	0.5769	0.5430
0.104	29.0050	5407	2.0881	0.5769	0.8166	0.8876	0.5769	0.6146	0.5769	0.5641
0.1454	30.0050	5587	2.3394	0.5621	0.7959	0.8905	0.5621	0.6280	0.5621	0.5448
0.2221	31.0050	5768	1.9360	0.5947	0.8225	0.9024	0.5947	0.6606	0.5947	0.5881
0.1026	32.005	5948	2.0920	0.6036	0.8107	0.8935	0.6036	0.6376	0.6036	0.5832
0.0968	33.0050	6128	2.2746	0.5740	0.8047	0.8846	0.5740	0.6308	0.5740	0.5542
0.1864	34.0050	6308	2.2081	0.5888	0.8047	0.8698	0.5888	0.6394	0.5888	0.5704
0.1353	35.0050	6489	2.1853	0.5799	0.8254	0.8935	0.5799	0.6133	0.5799	0.5636
0.1618	36.005	6669	2.2661	0.5710	0.7959	0.8698	0.5710	0.6243	0.5710	0.5515
0.259	37.0050	6849	2.3163	0.5740	0.7870	0.8580	0.5740	0.6088	0.5740	0.5459
0.3394	38.0050	7029	2.0984	0.5769	0.7988	0.8905	0.5769	0.6154	0.5769	0.5614
0.0833	39.0050	7210	2.2811	0.5533	0.8047	0.8698	0.5533	0.6051	0.5533	0.5328
0.1259	40.005	7390	2.2599	0.5828	0.7899	0.8698	0.5828	0.5806	0.5828	0.5510

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.1