falcon-clf

This model is a fine-tuned version of Rocketknight1/falcon-rw-1b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4504

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: 0.0002
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7984	0.99	77	0.7167
0.7175	1.99	155	0.6955
0.6092	2.99	233	0.6690
0.5426	3.99	311	0.6549
0.7676	5.0	389	0.6416
0.6552	6.0	467	0.6216
0.5989	7.0	545	0.6039
0.4944	8.0	623	0.5810
0.4591	8.99	700	0.5615
0.5415	9.99	778	0.5429
0.4794	10.99	856	0.5187
0.4347	11.99	934	0.4982
0.3487	13.0	1012	0.4845
0.3229	14.0	1090	0.4724
0.3946	15.0	1168	0.4624
0.3689	16.0	1246	0.4574
0.3191	16.99	1323	0.4529
0.2793	17.99	1401	0.4509
0.3675	18.99	1479	0.4504
0.3215	19.78	1540	0.4504

Framework versions

• PEFT 0.7.1
• Transformers 4.36.1
• Pytorch 2.0.0+cu117
• Datasets 2.18.0
• Tokenizers 0.15.2