b2b_paraphrase_retrain

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.0760

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7654	5.0	5	10.5282
0.612	10.0	10	7.3206
0.4323	15.0	15	6.1823
0.3377	20.0	20	4.8119
0.2952	25.0	25	4.4300
0.2734	30.0	30	4.2043
0.2613	35.0	35	4.0847
0.2537	40.0	40	3.9372
0.248	45.0	45	3.8550
0.2429	50.0	50	3.7778
0.2392	55.0	55	3.7219
0.2352	60.0	60	3.6906
0.2316	65.0	65	3.6539
0.2275	70.0	70	3.5842
0.2242	75.0	75	3.5152
0.2197	80.0	80	3.4511
0.2155	85.0	85	3.3881
0.211	90.0	90	3.3397
0.2061	95.0	95	3.2789
0.2013	100.0	100	3.2271
0.1967	105.0	105	3.1762
0.1915	110.0	110	3.1301
0.1868	115.0	115	3.1049
0.1829	120.0	120	3.0994
0.1784	125.0	125	3.0864
0.1743	130.0	130	3.0954
0.1709	135.0	135	3.0665
0.1683	140.0	140	3.0645
0.1637	145.0	145	3.0568
0.1614	150.0	150	3.0681
0.1573	155.0	155	3.0760

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1