workingtrain_AX_cluster_more_tokens

This model is a fine-tuned version of google/gemma-2b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.2403
• Rouge1: 0.3763
• Rouge2: 0.1192
• Rougel: 0.3528
• Rougelsum: 0.3528

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: 0.00021
• train_batch_size: 20
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 40
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 2
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
3.3875	0.1333	200	3.3124	0.3546	0.1068	0.3356	0.3359
3.3444	0.2667	400	3.2866	0.3706	0.1103	0.3459	0.3461
3.1973	0.4	600	3.2682	0.3717	0.1138	0.3475	0.3477
3.2284	0.5333	800	3.2623	0.3740	0.1160	0.3505	0.3507
3.3128	0.6667	1000	3.2489	0.3732	0.1127	0.3488	0.3489
3.2104	0.8	1200	3.2412	0.3754	0.1144	0.3514	0.3514
3.0321	0.9333	1400	3.2356	0.3739	0.1150	0.3495	0.3497
3.0487	1.0667	1600	3.2392	0.3754	0.1173	0.3525	0.3526
3.1087	1.2	1800	3.2406	0.3732	0.1152	0.3498	0.3496
3.2041	1.3333	2000	3.2403	0.3763	0.1192	0.3528	0.3528

Framework versions

• PEFT 0.11.1
• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1