300868d599638cdf015f0abc2c25d6c3

This model is a fine-tuned version of deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 7.2459
• Data Size: 1.0
• Epoch Runtime: 48.7322
• Accuracy: 0.7614
• F1 Macro: 0.7990
• Rouge1: 0.7624
• Rouge2: 0.0
• Rougel: 0.7628
• Rougelsum: 0.7614

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	12.7991	0	4.1710	0.1016	0.0689	0.1016	0.0	0.1009	0.1016
No log	1	178	5.7518	0.0078	4.0659	0.4567	0.3198	0.4581	0.0	0.4560	0.4574
No log	2	356	4.8797	0.0156	6.4951	0.5369	0.3721	0.5369	0.0	0.5369	0.5369
No log	3	534	3.8403	0.0312	10.1974	0.6676	0.5314	0.6676	0.0	0.6676	0.6673
No log	4	712	3.1151	0.0625	14.4103	0.7180	0.5702	0.7188	0.0	0.7188	0.7180
No log	5	890	3.2036	0.125	18.0234	0.7045	0.6562	0.7045	0.0	0.7045	0.7045
0.2105	6	1068	2.3742	0.25	25.4009	0.7678	0.8038	0.7685	0.0	0.7678	0.7685
1.8464	7	1246	2.4336	0.5	31.6111	0.7741	0.8120	0.7741	0.0	0.7749	0.7749
1.2802	8.0	1424	2.4679	1.0	51.3293	0.7741	0.8151	0.7741	0.0	0.7749	0.7741
0.2418	9.0	1602	6.5792	1.0	48.0364	0.7486	0.7732	0.7486	0.0	0.7493	0.7486
0.2348	10.0	1780	7.2459	1.0	48.7322	0.7614	0.7990	0.7624	0.0	0.7628	0.7614

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1