judge_answer___33_deberta_base_enwiki-answerability-2411

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2045
• Accuracy: 0.9442
• Precision: 0.9592
• Recall: 0.9541
• F1: 0.9566

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
4.2448	0.0267	500	0.3176	0.8796	0.9528	0.8556	0.9016
2.4191	0.0533	1000	0.2767	0.9138	0.9271	0.9403	0.9336
1.9636	0.08	1500	0.2391	0.9142	0.9607	0.9039	0.9314
1.7834	0.1067	2000	0.2148	0.9254	0.9643	0.9183	0.9407
1.675	0.1333	2500	0.2138	0.9292	0.9560	0.9332	0.9444
1.7663	0.16	3000	0.2685	0.9177	0.9747	0.8956	0.9335
1.7593	0.1867	3500	0.2410	0.9254	0.9678	0.9147	0.9405
1.7727	0.2133	4000	0.1939	0.9354	0.9592	0.9397	0.9494
1.6498	0.24	4500	0.2019	0.9358	0.9564	0.9433	0.9498
1.5754	0.2667	5000	0.2334	0.9377	0.9577	0.9451	0.9514
1.6312	0.2933	5500	0.2120	0.935	0.9671	0.9308	0.9486
1.5515	0.32	6000	0.2342	0.9354	0.9637	0.9350	0.9491
1.6374	0.3467	6500	0.2143	0.9385	0.9555	0.9487	0.9521
1.6782	0.3733	7000	0.1865	0.9373	0.9599	0.9421	0.9509
1.614	0.4	7500	0.2039	0.9404	0.9562	0.9511	0.9536
1.5568	0.4267	8000	0.1862	0.9423	0.9641	0.9457	0.9548
1.5774	0.4533	8500	0.1818	0.94	0.9634	0.9427	0.9530
1.5722	0.48	9000	0.2388	0.9396	0.9628	0.9427	0.9527
1.5544	0.5067	9500	0.2009	0.9408	0.9635	0.9439	0.9536
1.5426	0.5333	10000	0.2398	0.9385	0.9662	0.9374	0.9515
1.5144	0.56	10500	0.2223	0.9381	0.9662	0.9368	0.9512
1.508	0.5867	11000	0.2135	0.9446	0.9517	0.9630	0.9573
1.5881	0.6133	11500	0.1886	0.9404	0.9546	0.9529	0.9537
1.4951	0.64	12000	0.2053	0.9442	0.9671	0.9457	0.9563
1.583	0.6667	12500	0.2088	0.9412	0.9663	0.9415	0.9538
1.5312	0.6933	13000	0.2041	0.9373	0.9719	0.9296	0.9503
1.5474	0.72	13500	0.1907	0.9412	0.9663	0.9415	0.9538
1.4928	0.7467	14000	0.1998	0.9438	0.9631	0.9493	0.9561
1.5224	0.7733	14500	0.1940	0.9423	0.9591	0.9511	0.9551
1.5267	0.8	15000	0.2095	0.9442	0.9665	0.9463	0.9563
1.6073	0.8267	15500	0.1905	0.945	0.9620	0.9523	0.9571
1.4924	0.8533	16000	0.2118	0.9462	0.9666	0.9493	0.9579
1.543	0.88	16500	0.2074	0.9442	0.9603	0.9529	0.9566
1.6774	0.9067	17000	0.2044	0.9446	0.9631	0.9505	0.9568
1.5077	0.9333	17500	0.2007	0.945	0.9626	0.9517	0.9571
1.4738	0.96	18000	0.2018	0.9442	0.9603	0.9529	0.9566
1.4543	0.9867	18500	0.2045	0.9442	0.9592	0.9541	0.9566

Framework versions

• Transformers 4.46.0
• Pytorch 2.4.1+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1