mistral_numericnlg_FV

This model is a fine-tuned version of unsloth/mistral-7b-instruct-v0.2-bnb-4bit on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2038
• Rouge1: 0.6650
• Rougel: 0.5559

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: 4
• eval_batch_size: 1
• seed: 3407
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 5
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rougel
1.3723	0.9446	16	1.3128	0.6406	0.5251
1.2226	1.9483	33	1.2305	0.6603	0.5562
1.1781	2.9520	50	1.2129	0.6658	0.5556
1.2057	3.9557	67	1.2062	0.6659	0.5560
1.215	4.9594	84	1.2040	0.6656	0.5551
1.1676	5.6679	96	1.2038	0.6650	0.5559

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.2.0
• Datasets 2.16.0
• Tokenizers 0.19.1