ministral-3b-instruct-mimic4-adapt-l2r

This model is a fine-tuned version of ministral/Ministral-3b-instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: -545681554523036992.0000
• Ndcg: 0.9570
• Ndcg@25: 0.9025
• Precision@25: 0.9120

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: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Use 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: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Ndcg	Ndcg@25	Precision@25
-5233964739674190848.0000	1.0	44	-522964517807628544.0000	0.9567	0.9340	0.9429
-213567145714384896.0000	2.0	88	-527819484207438720.0000	0.9568	0.9212	0.9497
-595173120794217728.0000	2.9480	129	-545681554523036992.0000	0.9570	0.9025	0.9120

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0
• Datasets 3.6.0
• Tokenizers 0.21.1