fineweb-distil-xlstm

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

• Loss: 2.3150

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.5458	0.08	100	3.0996
2.9415	0.16	200	2.9096
2.8028	0.24	300	2.7979
2.6877	0.32	400	2.7064
2.6105	0.4	500	2.6351
2.5406	0.48	600	2.5843
2.4982	0.56	700	2.5451
2.4888	0.64	800	2.5058
2.4339	0.72	900	2.4766
2.3962	0.8	1000	2.4500
2.3711	0.88	1100	2.4281
2.3509	0.96	1200	2.4046
2.2915	1.04	1300	2.3911
2.2122	1.12	1400	2.3764
2.1985	1.2	1500	2.3662
2.1937	1.28	1600	2.3508
2.191	1.3600	1700	2.3432
2.1877	1.44	1800	2.3351
2.1629	1.52	1900	2.3276
2.1565	1.6	2000	2.3243
2.1597	1.6800	2100	2.3203
2.1529	1.76	2200	2.3170
2.1666	1.8400	2300	2.3157
2.1557	1.92	2400	2.3151
2.1579	2.0	2500	2.3150

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3