metadata

language:
  - en
license: mit
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
model-index:
  - name: roberta-base-sst2
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE SST2
          type: glue
          args: sst2
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9323394495412844

roberta-base-sst2

This model is a fine-tuned version of roberta-base on the GLUE SST2 dataset. It achieves the following results on the evaluation set:

Loss: 0.1952
Accuracy: 0.9323

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: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.06
num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.575	0.12	500	0.2665	0.9071
0.2989	0.24	1000	0.2088	0.9220
0.2725	0.36	1500	0.2560	0.9243
0.2814	0.48	2000	0.2016	0.9266
0.2586	0.59	2500	0.2293	0.9174
0.2536	0.71	3000	0.2340	0.9323
0.2494	0.83	3500	0.1952	0.9323
0.2396	0.95	4000	0.2494	0.9323
0.2123	1.07	4500	0.2187	0.9381
0.2042	1.19	5000	0.2812	0.9151
0.2083	1.31	5500	0.2739	0.9346
0.2041	1.43	6000	0.2087	0.9381
0.1969	1.54	6500	0.2590	0.9255
0.1982	1.66	7000	0.2445	0.9300
0.1943	1.78	7500	0.2798	0.9266
0.1848	1.9	8000	0.2844	0.9312
0.1788	2.02	8500	0.2998	0.9255
0.1623	2.14	9000	0.2696	0.9392
0.1499	2.26	9500	0.2533	0.9278
0.1426	2.38	10000	0.2971	0.9300
0.1479	2.49	10500	0.2596	0.9358
0.1405	2.61	11000	0.2945	0.9255
0.1577	2.73	11500	0.4061	0.9002
0.1521	2.85	12000	0.2724	0.9335
0.1426	2.97	12500	0.2712	0.9427
0.1206	3.09	13000	0.2954	0.9358
0.1074	3.21	13500	0.2653	0.9392
0.112	3.33	14000	0.2778	0.9346
0.1147	3.44	14500	0.3705	0.9312
0.1196	3.56	15000	0.2890	0.9346
0.1159	3.68	15500	0.3449	0.9266
0.119	3.8	16000	0.3207	0.9335
0.1268	3.92	16500	0.3235	0.9312
0.1074	4.04	17000	0.3650	0.9335
0.0805	4.16	17500	0.3338	0.9381
0.0838	4.28	18000	0.4302	0.9209
0.0848	4.39	18500	0.4096	0.9323
0.0922	4.51	19000	0.3332	0.9369
0.091	4.63	19500	0.3024	0.9438
0.0977	4.75	20000	0.2674	0.9495
0.0897	4.87	20500	0.3993	0.9300
0.1013	4.99	21000	0.3227	0.9289
0.0671	5.11	21500	0.3374	0.9427
0.0671	5.23	22000	0.4108	0.9278
0.0652	5.34	22500	0.3550	0.9381
0.0664	5.46	23000	0.3398	0.9358
0.0742	5.58	23500	0.3286	0.9381
0.0758	5.7	24000	0.3276	0.9312
0.075	5.82	24500	0.3202	0.9369
0.0686	5.94	25000	0.3481	0.9415
0.0729	6.06	25500	0.3816	0.9335
0.0568	6.18	26000	0.3132	0.9381
0.0529	6.29	26500	0.3757	0.9300
0.0506	6.41	27000	0.3396	0.9381
0.0476	6.53	27500	0.3642	0.9404
0.0555	6.65	28000	0.3430	0.9404
0.0574	6.77	28500	0.3401	0.9392
0.0524	6.89	29000	0.3378	0.9346
0.0492	7.01	29500	0.3833	0.9381
0.039	7.13	30000	0.3347	0.9346
0.0411	7.24	30500	0.4404	0.9335
0.0412	7.36	31000	0.3618	0.9381
0.0477	7.48	31500	0.3806	0.9381
0.0435	7.6	32000	0.3912	0.9335
0.0443	7.72	32500	0.3900	0.9392
0.0421	7.84	33000	0.4152	0.9369
0.0495	7.96	33500	0.3832	0.9289
0.0293	8.08	34000	0.4427	0.9346
0.0253	8.19	34500	0.4425	0.9381
0.0407	8.31	35000	0.4102	0.9358
0.0311	8.43	35500	0.4447	0.9369
0.0291	8.55	36000	0.4612	0.9346
0.035	8.67	36500	0.4241	0.9346
0.0381	8.79	37000	0.4198	0.9312
0.0234	8.91	37500	0.4345	0.9369
0.0311	9.03	38000	0.4558	0.9312
0.028	9.14	38500	0.4245	0.9381
0.0213	9.26	39000	0.4462	0.9381
0.0276	9.38	39500	0.4210	0.9381
0.0183	9.5	40000	0.4310	0.9404
0.0184	9.62	40500	0.4437	0.9404
0.0296	9.74	41000	0.4311	0.9392
0.019	9.86	41500	0.4244	0.9415
0.0245	9.98	42000	0.4270	0.9415

Framework versions

Transformers 4.21.3
Pytorch 1.7.1
Datasets 1.18.3
Tokenizers 0.11.6