dit_maveriq_tobacco3482_2023-07-04

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

• Loss: 0.2641
• Accuracy: 0.955

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.96	3	2.0434	0.265
No log	1.96	6	1.9976	0.31
No log	2.96	9	1.7970	0.335
No log	3.96	12	1.6899	0.385
No log	4.96	15	1.6519	0.36
No log	5.96	18	1.5378	0.42
No log	6.96	21	1.4401	0.51
No log	7.96	24	1.3607	0.575
No log	8.96	27	1.2614	0.605
No log	9.96	30	1.1654	0.63
No log	10.96	33	1.0758	0.66
No log	11.96	36	0.9908	0.73
No log	12.96	39	0.9152	0.72
No log	13.96	42	0.8412	0.755
No log	14.96	45	0.7759	0.78
No log	15.96	48	0.7105	0.785
No log	16.96	51	0.6489	0.815
No log	17.96	54	0.6037	0.825
No log	18.96	57	0.5628	0.83
No log	19.96	60	0.5084	0.84
No log	20.96	63	0.4593	0.85
No log	21.96	66	0.4306	0.865
No log	22.96	69	0.4078	0.87
No log	23.96	72	0.3888	0.875
No log	24.96	75	0.3713	0.885
No log	25.96	78	0.3469	0.89
No log	26.96	81	0.3234	0.91
No log	27.96	84	0.3146	0.905
No log	28.96	87	0.3311	0.895
No log	29.96	90	0.3178	0.92
No log	30.96	93	0.3011	0.92
No log	31.96	96	0.2922	0.93
No log	32.96	99	0.2822	0.93
No log	33.96	102	0.2615	0.93
No log	34.96	105	0.2577	0.94
No log	35.96	108	0.2587	0.94
No log	36.96	111	0.2659	0.93
No log	37.96	114	0.2697	0.925
No log	38.96	117	0.2721	0.93
No log	39.96	120	0.2829	0.935
No log	40.96	123	0.2564	0.94
No log	41.96	126	0.2420	0.94
No log	42.96	129	0.2433	0.94
No log	43.96	132	0.2405	0.945
No log	44.96	135	0.2391	0.95
No log	45.96	138	0.2455	0.955
No log	46.96	141	0.2563	0.945
No log	47.96	144	0.2653	0.95
No log	48.96	147	0.2608	0.945
No log	49.96	150	0.2477	0.95
No log	50.96	153	0.2443	0.95
No log	51.96	156	0.2418	0.95
No log	52.96	159	0.2403	0.94
No log	53.96	162	0.2384	0.945
No log	54.96	165	0.2413	0.95
No log	55.96	168	0.2428	0.96
No log	56.96	171	0.2409	0.955
No log	57.96	174	0.2457	0.95
No log	58.96	177	0.2488	0.95
No log	59.96	180	0.2548	0.955
No log	60.96	183	0.2597	0.955
No log	61.96	186	0.2647	0.955
No log	62.96	189	0.2651	0.955
No log	63.96	192	0.2638	0.955
No log	64.96	195	0.2638	0.955
No log	65.96	198	0.2664	0.955
No log	66.96	201	0.2712	0.95
No log	67.96	204	0.2677	0.955
No log	68.96	207	0.2601	0.95
No log	69.96	210	0.2559	0.95
No log	70.96	213	0.2566	0.95
No log	71.96	216	0.2611	0.95
No log	72.96	219	0.2702	0.95
No log	73.96	222	0.2806	0.945
No log	74.96	225	0.2842	0.945
No log	75.96	228	0.2807	0.945
No log	76.96	231	0.2750	0.95
No log	77.96	234	0.2656	0.955
No log	78.96	237	0.2582	0.96
No log	79.96	240	0.2545	0.96
No log	80.96	243	0.2535	0.96
No log	81.96	246	0.2512	0.96
No log	82.96	249	0.2520	0.96
No log	83.96	252	0.2546	0.96
No log	84.96	255	0.2570	0.96
No log	85.96	258	0.2608	0.96
No log	86.96	261	0.2641	0.96
No log	87.96	264	0.2672	0.955
No log	88.96	267	0.2686	0.955
No log	89.96	270	0.2682	0.955
No log	90.96	273	0.2671	0.955
No log	91.96	276	0.2652	0.96
No log	92.96	279	0.2639	0.96
No log	93.96	282	0.2635	0.96
No log	94.96	285	0.2633	0.96
No log	95.96	288	0.2635	0.955
No log	96.96	291	0.2636	0.955
No log	97.96	294	0.2639	0.955
No log	98.96	297	0.2640	0.955
No log	99.96	300	0.2641	0.955

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2