metadata

tags:
  - generated_from_trainer
datasets:
  - kanishka/counterfactual_babylm_measure_nps_as_singular_new
metrics:
  - accuracy
model-index:
  - name: smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-1e-4
    results:
      - task:
          name: Causal Language Modeling
          type: text-generation
        dataset:
          name: kanishka/counterfactual_babylm_measure_nps_as_singular_new
          type: kanishka/counterfactual_babylm_measure_nps_as_singular_new
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.40681131693060796

smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-1e-4

This model was trained from scratch on the kanishka/counterfactual_babylm_measure_nps_as_singular_new dataset. It achieves the following results on the evaluation set:

Loss: 3.4240
Accuracy: 0.4068

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
train_batch_size: 32
eval_batch_size: 64
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 32000
num_epochs: 20.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
4.0517	1.0	18602	4.2617	0.3086
3.5614	2.0	37204	3.7325	0.3617
3.3871	3.0	55806	3.5926	0.3794
3.2873	4.0	74408	3.4903	0.3889
3.2166	5.0	93010	3.4705	0.3930
3.1683	6.0	111612	3.4386	0.3965
3.122	7.0	130214	3.4230	0.3987
3.0883	8.0	148816	3.4103	0.4020
3.059	9.0	167418	3.4161	0.4022
3.0294	10.0	186020	3.4004	0.4039
3.0081	11.0	204622	3.4048	0.4041
2.9849	12.0	223224	3.4068	0.4046
2.9618	13.0	241826	3.4127	0.4048
2.9398	14.0	260428	3.4079	0.4054
2.9226	15.0	279030	3.3963	0.4065
2.9009	16.0	297632	3.4036	0.4068
2.8845	17.0	316234	3.4090	0.4067
2.8685	18.0	334836	3.4054	0.4071
2.8513	19.0	353438	3.4187	0.4069
2.8368	20.0	372040	3.4240	0.4068

Framework versions

Transformers 4.38.0
Pytorch 2.3.1+cu121
Datasets 2.16.1
Tokenizers 0.15.2