mistral-instruct-dutch-syntax-2000

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.1 on a curated version of Lassy-Small with syntax data. 2000 samples.

It achieves the following results on the evaluation set:

• Loss: 0.6808

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: 2.5e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.03
• training_steps: 950

Training results

Training Loss	Epoch	Step	Validation Loss
1.1019	0.11	100	1.0701
0.9093	0.21	200	0.9592
0.8341	0.32	300	0.8800
0.7975	0.42	400	0.8150
0.7859	0.53	500	0.7638
0.7069	0.63	600	0.7254
0.6007	0.74	700	0.6974
0.6971	0.84	800	0.6832
0.6331	0.95	900	0.6808

Framework versions

• PEFT 0.7.1
• Transformers 4.36.1
• Pytorch 2.1.2+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0