starchat-clean

This model is a fine-tuned version of bigcode/starcoderbase on the crumb/Clean-Instruct-440k dataset. It achieves the following results on the evaluation set:

• Loss: 1.2266

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
1.4135	0.99	55	1.3103
1.3429	2.0	111	1.2430
1.2432	2.98	166	1.2283
1.2808	3.96	220	1.2266

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.13.1
• Tokenizers 0.13.3