timeSformer

This model is a fine-tuned version of MCG-NJU/videomae-base-finetuned-kinetics on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 7.8889
• Accuracy: 0.0052

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: 0.005
• train_batch_size: 288
• eval_batch_size: 288
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 740

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.8418	0.0135	10	5.5898	0.0122
3.1132	1.0135	20	5.8064	0.0052
3.1265	2.0135	30	5.9893	0.0087
3.1378	3.0135	40	5.8669	0.0017
3.1408	4.0135	50	6.0782	0.0
3.0995	5.0135	60	5.9832	0.0
3.0778	6.0135	70	5.6949	0.0486
3.0765	7.0135	80	5.8415	0.0017
3.0689	8.0135	90	5.9740	0.0017
3.0093	9.0135	100	5.8951	0.0052
3.021	10.0135	110	6.1976	0.0052
2.9475	11.0135	120	6.4536	0.0191
2.9468	12.0135	130	6.5053	0.0
2.9593	13.0135	140	6.1705	0.0139
2.9632	14.0135	150	6.5190	0.0017
2.9551	15.0135	160	6.6961	0.0017
3.0157	16.0135	170	6.0960	0.0208
3.0034	17.0135	180	6.5592	0.0052
2.9783	18.0135	190	6.4399	0.0365
2.9608	19.0135	200	6.4748	0.0035
2.9909	20.0135	210	6.3702	0.0035
2.9598	21.0135	220	6.5077	0.0
2.9552	22.0135	230	7.4796	0.0069
3.0327	23.0135	240	6.5649	0.0
3.0028	24.0135	250	6.6634	0.0087
2.9237	25.0135	260	6.8856	0.0
2.9411	26.0135	270	6.9620	0.0017
2.9281	27.0135	280	6.9007	0.0069
2.9255	28.0135	290	7.0261	0.0017
2.9621	29.0135	300	6.5056	0.0
2.9419	30.0135	310	6.9045	0.0035
2.978	31.0135	320	6.8059	0.0122
3.0299	32.0135	330	6.6262	0.0017
3.0313	33.0135	340	6.6773	0.0
2.995	34.0135	350	6.7126	0.0
2.9588	35.0135	360	6.4966	0.0035
2.9784	36.0135	370	6.2595	0.0
3.0047	37.0135	380	6.5504	0.0
2.98	38.0135	390	6.3898	0.0017
2.9568	39.0135	400	6.8740	0.0017
2.9418	40.0135	410	6.5854	0.0
2.955	41.0135	420	6.5085	0.0052
2.9689	42.0135	430	6.4343	0.0
2.9494	43.0135	440	6.5746	0.0
2.9363	44.0135	450	6.7996	0.0035
2.9002	45.0135	460	7.1057	0.0035
2.8776	46.0135	470	7.4000	0.0035
2.8853	47.0135	480	6.9539	0.0017
2.8618	48.0135	490	7.0917	0.0035
2.8728	49.0135	500	7.4421	0.0017
2.8888	50.0135	510	7.1914	0.0052
2.8824	51.0135	520	7.3226	0.0069
2.8932	52.0135	530	7.3581	0.0
2.8562	53.0135	540	7.3102	0.0017
2.8651	54.0135	550	7.2675	0.0035
2.8898	55.0135	560	7.5446	0.0017
2.8931	56.0135	570	7.2915	0.0
2.8864	57.0135	580	7.2397	0.0017
2.8683	58.0135	590	6.9988	0.0017
2.8213	59.0135	600	7.4490	0.0052
2.8319	60.0135	610	7.5569	0.0017
2.8237	61.0135	620	7.5322	0.0017
2.8057	62.0135	630	7.6815	0.0122
2.8047	63.0135	640	7.3165	0.0
2.8277	64.0135	650	7.4256	0.0069
2.8202	65.0135	660	7.6688	0.0035
2.772	66.0135	670	7.5628	0.0052
2.7749	67.0135	680	7.7103	0.0052
2.7442	68.0135	690	7.8151	0.0052
2.7715	69.0135	700	7.8429	0.0052
2.7593	70.0135	710	7.8079	0.0052
2.7334	71.0135	720	7.8237	0.0052
2.7392	72.0135	730	7.9085	0.0052
2.753	73.0135	740	7.8889	0.0052

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1