dfm1

This model is a fine-tuned version of KennethEnevoldsen/dfm-sentence-encoder-large-exp2-no-lang-align on an unknown dataset. It achieves the following results on the evaluation set:

• Accuracy: 0.8868
• Precision: 0.8861
• Recall: 0.8868
• F1: 0.8855
• Loss: 0.5432

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Accuracy	Precision	Recall	F1	Validation Loss
No log	0.9412	8	0.7844	0.7464	0.7844	0.7612	0.7436
No log	2.0	17	0.8999	0.8922	0.8999	0.8914	0.3252
No log	2.9412	25	0.9214	0.9226	0.9214	0.9121	0.3213
No log	4.0	34	0.9164	0.9235	0.9164	0.9176	0.3572
No log	4.9412	42	0.8880	0.8875	0.8880	0.8857	0.3576
No log	6.0	51	0.8907	0.8894	0.8907	0.8898	0.3993
No log	6.9412	59	0.8822	0.8822	0.8822	0.8806	0.4444
No log	8.0	68	0.8876	0.8867	0.8876	0.8865	0.4480
No log	8.9412	76	0.8987	0.8978	0.8987	0.8979	0.4688
No log	10.0	85	0.8984	0.8972	0.8984	0.8975	0.4845
No log	10.9412	93	0.8895	0.8887	0.8895	0.8884	0.5172
No log	12.0	102	0.8891	0.8882	0.8891	0.8881	0.5349
No log	12.9412	110	0.8907	0.8897	0.8907	0.8896	0.5343
No log	14.0	119	0.8895	0.8886	0.8895	0.8884	0.5374
No log	14.9412	127	0.8868	0.8861	0.8868	0.8855	0.5317
No log	16.0	136	0.8853	0.8847	0.8853	0.8839	0.5383
No log	16.9412	144	0.8853	0.8847	0.8853	0.8839	0.5402
No log	18.0	153	0.8865	0.8858	0.8865	0.8851	0.5429
No log	18.8235	160	0.8868	0.8861	0.8868	0.8855	0.5432

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Tokenizers 0.19.1