jina-reranker-v2-base-multilingual-finetune-lower-lr-gskf

This model is a fine-tuned version of jinaai/jina-reranker-v2-base-multilingual on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4257
• Spearman: 0.5271

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: 4e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Spearman
No log	0	0	0.4152	0.4741
0.3413	0.2455	1000	0.3882	0.4894
0.4668	0.4910	2000	0.3827	0.5085
0.5458	0.7366	3000	0.3862	0.5256
0.8905	0.9821	4000	0.3799	0.5254
0.2713	1.2276	5000	0.3899	0.5191
0.1306	1.4731	6000	0.3860	0.5259
0.4036	1.7186	7000	0.3900	0.5269
0.3428	1.9642	8000	0.3939	0.5268
0.3073	2.2097	9000	0.4101	0.5302
0.163	2.4552	10000	0.4139	0.5300
0.5277	2.7007	11000	0.4194	0.5270
0.2899	2.9462	12000	0.4259	0.5260
0.5044	3.1918	13000	0.4303	0.5256
0.2152	3.4373	14000	0.4280	0.5263
0.2621	3.6828	15000	0.4285	0.5269
0.38	3.9283	16000	0.4249	0.5274
0.2421	4.1738	17000	0.4262	0.5269
0.3829	4.4193	18000	0.4258	0.5270
0.3087	4.6649	19000	0.4259	0.5271
0.0309	4.9104	20000	0.4257	0.5271

Framework versions

• Transformers 4.56.2
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.1