AngelinaZanardi/multilingual-e5-base-edu-scorer-lr3e4-bs32-swe

multilingual-e5-base-edu-scorer-lr3e4-bs32

This model is a fine-tuned version of intfloat/multilingual-e5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7551
• Precision: 0.3916
• Recall: 0.3720
• F1 Macro: 0.3704
• Accuracy: 0.4709

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: 0.0003
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• 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
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1 Macro	Accuracy
No log	0	0	4.8374	0.0513	0.1667	0.0785	0.3080
1.0429	0.6793	1000	0.9472	0.3859	0.3207	0.3148	0.4320
1.0195	1.3587	2000	0.8833	0.4146	0.3355	0.3344	0.4225
1.0332	2.0380	3000	0.8693	0.4213	0.3342	0.3294	0.4111
0.9677	2.7174	4000	0.8591	0.4387	0.3494	0.3501	0.4213
0.9779	3.3967	5000	0.8598	0.4207	0.3584	0.3632	0.4550
0.932	4.0761	6000	0.8469	0.4429	0.3681	0.3717	0.4496
0.9421	4.7554	7000	0.8427	0.4376	0.3658	0.3717	0.4655
0.9552	5.4348	8000	0.8377	0.4364	0.3610	0.3682	0.4607
0.9298	6.1141	9000	0.8297	0.4580	0.3841	0.3918	0.4655
0.9307	6.7935	10000	0.8211	0.4542	0.3728	0.3786	0.4582
0.8988	7.4728	11000	0.8164	0.4440	0.3584	0.3597	0.4480
0.9219	8.1522	12000	0.8119	0.4617	0.3704	0.3763	0.4534
0.9289	8.8315	13000	0.8037	0.4422	0.3710	0.3780	0.4706
0.9109	9.5109	14000	0.8214	0.4588	0.3643	0.3659	0.4363
0.9017	10.1902	15000	0.8189	0.4425	0.3839	0.3864	0.4534
0.9117	10.8696	16000	0.8061	0.4542	0.3778	0.3837	0.4595
0.8836	11.5489	17000	0.7869	0.4671	0.3753	0.3825	0.4730
0.8749	12.2283	18000	0.7920	0.4604	0.3793	0.3863	0.4665
0.9009	12.9076	19000	0.7990	0.4645	0.3689	0.3753	0.4786
0.8984	13.5870	20000	0.7838	0.4652	0.3806	0.3899	0.4754
0.8525	14.2663	21000	0.7947	0.4478	0.3693	0.3758	0.4835
0.8514	14.9457	22000	0.7822	0.4734	0.3882	0.3976	0.4720
0.8796	15.625	23000	0.7804	0.4810	0.3823	0.3917	0.4706
0.8722	16.3043	24000	0.7820	0.4803	0.3819	0.3920	0.4738
0.8712	16.9837	25000	0.7815	0.4716	0.3845	0.3939	0.4679
0.8824	17.6630	26000	0.7827	0.4680	0.3726	0.3793	0.4792
0.8344	18.3424	27000	0.7792	0.4653	0.3760	0.3833	0.4810
0.8291	19.0217	28000	0.7755	0.4636	0.3806	0.3889	0.4774
0.8287	19.7011	29000	0.7754	0.4759	0.3861	0.3951	0.4762

Framework versions

• Transformers 4.55.0
• Pytorch 2.6.0.dev20241112+cu121
• Datasets 4.0.0
• Tokenizers 0.21.4