Howard881010/heat_transfer_sft_10000_1epoch_fs

sft_10000_fs

This model is a fine-tuned version of mistralai/Mistral-Nemo-Instruct-2407 on the heat_transfer_10000_fs dataset. It achieves the following results on the evaluation set:

• Loss: 0.0006

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: 12
• eval_batch_size: 12
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 24
• total_eval_batch_size: 24
• 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: cosine
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.0107	0.0533	20	0.0043
0.0035	0.1067	40	0.0034
0.0033	0.16	60	0.0027
0.0018	0.2133	80	0.0018
0.0025	0.2667	100	0.0019
0.0019	0.32	120	0.0017
0.0015	0.3733	140	0.0016
0.0015	0.4267	160	0.0014
0.0015	0.48	180	0.0017
0.001	0.5333	200	0.0009
0.0007	0.5867	220	0.0008
0.0011	0.64	240	0.0008
0.0008	0.6933	260	0.0011
0.0006	0.7467	280	0.0009
0.0008	0.8	300	0.0006
0.0006	0.8533	320	0.0007
0.0005	0.9067	340	0.0006
0.0006	0.96	360	0.0006

Framework versions

• PEFT 0.12.0
• Transformers 4.46.0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.20.1