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

Loss: 3.4072
Accuracy: 0.4112

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: 0.001
train_batch_size: 32
eval_batch_size: 64
seed: 1024
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
3.598	1.0	18595	3.7669	0.3585
3.3838	2.0	37190	3.5520	0.3816
3.2582	3.0	55785	3.4897	0.3915
3.1816	4.0	74380	3.4356	0.3977
3.1307	5.0	92975	3.3983	0.4019
3.0779	6.0	111570	3.3854	0.4038
3.0456	7.0	130165	3.3700	0.4064
3.0146	8.0	148760	3.3776	0.4068
2.9868	9.0	167355	3.3713	0.4086
2.9591	10.0	185950	3.3665	0.4095
2.9347	11.0	204545	3.3668	0.4095
2.9148	12.0	223140	3.3561	0.4107
2.8968	13.0	241735	3.3906	0.4103
2.8728	14.0	260330	3.3817	0.4105
2.8529	15.0	278925	3.3659	0.4110
2.8333	16.0	297520	3.3872	0.4109
2.8133	17.0	316115	3.3979	0.4106
2.7939	18.0	334710	3.3862	0.4117
2.7757	19.0	353305	3.4053	0.4110
2.7645	20.0	371900	3.4072	0.4112

Framework versions

Transformers 4.38.0
Pytorch 2.3.1+cu121
Datasets 2.16.1
Tokenizers 0.15.2

kanishka/smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-seed_1024-1e-3

smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-seed_1024-1e-3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

kanishka/
smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-seed_1024-1e-3