25be1998800dec9c0bbdcf5cc38a8d69

This model is a fine-tuned version of google/gemma-2b on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 4.4009
• Data Size: 1.0
• Epoch Runtime: 48.3438
• Accuracy: 0.7656
• F1 Macro: 0.7991
• Rouge1: 0.7663
• Rouge2: 0.0
• Rougel: 0.7663
• Rougelsum: 0.7656

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	10.7795	0	3.5341	0.2713	0.0892	0.2706	0.0	0.2706	0.2713
No log	1	178	8.2886	0.0078	3.9143	0.4517	0.2628	0.4517	0.0	0.4517	0.4517
No log	2	356	3.9526	0.0156	6.8323	0.6321	0.4341	0.6335	0.0	0.6321	0.6321
No log	3	534	3.4435	0.0312	8.0550	0.7081	0.5638	0.7081	0.0	0.7081	0.7074
No log	4	712	3.0984	0.0625	9.5033	0.7287	0.5785	0.7294	0.0	0.7287	0.7287
No log	5	890	3.3881	0.125	12.0974	0.6669	0.5911	0.6683	0.0	0.6676	0.6669
0.2589	6	1068	2.9270	0.25	19.0947	0.7294	0.7461	0.7301	0.0	0.7298	0.7294
2.345	7	1246	2.3898	0.5	29.0344	0.7741	0.8014	0.7749	0.0	0.7741	0.7741
1.7419	8.0	1424	3.5237	1.0	52.7376	0.7116	0.7388	0.7116	0.0	0.7124	0.7116
0.8689	9.0	1602	3.8945	1.0	51.9258	0.7784	0.8105	0.7791	0.0	0.7791	0.7784
0.5429	10.0	1780	5.5914	1.0	54.8372	0.7116	0.7666	0.7124	0.0	0.7116	0.7124
0.5361	11.0	1958	4.4009	1.0	48.3438	0.7656	0.7991	0.7663	0.0	0.7663	0.7656

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1