mega-base-multiple-choice-fp16-v4

This model is a fine-tuned version of mnaylor/mega-base-wikitext on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6932
• Accuracy: 0.4964
• Precision: 0.4964
• Recall: 0.5023
• F1: 0.4993

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: 5e-05
• train_batch_size: 1024
• eval_batch_size: 1024
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 24000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
No log	1.0	34	0.6932	0.4970	0.4971	0.5023	0.4997
No log	2.0	68	0.6932	0.4975	0.4975	0.5026	0.5001
No log	3.0	102	0.6932	0.4974	0.4974	0.5020	0.4997
No log	4.0	136	0.6932	0.4995	0.4995	0.5043	0.5019
No log	5.0	170	0.6932	0.4975	0.4975	0.5023	0.4999
No log	6.0	204	0.6932	0.4987	0.4987	0.5043	0.5015
No log	7.0	238	0.6932	0.4960	0.4961	0.5026	0.4993
No log	8.0	272	0.6932	0.4967	0.4967	0.5036	0.5002
No log	9.0	306	0.6932	0.4965	0.4966	0.5	0.4983
No log	10.0	340	0.6932	0.4964	0.4964	0.5023	0.4993

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0