sft-fsi

This model is a fine-tuned version of dynamofl/dynamo-1.6B-v0.4-mosaic-dynamoDPO-iter0-2978 on the dynamofl/train-default-FSI-PersonalFinancialAdvice-input-formatted-chatml dataset. It achieves the following results on the evaluation set:

• Loss: 0.5011

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: 2e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
2.7021	1.0	15	2.5445
1.8713	2.0	30	1.7759
1.4708	3.0	45	1.3395
1.1043	4.0	60	1.0145
0.8429	5.0	75	0.7671
0.6977	6.0	90	0.6217
0.6012	7.0	105	0.5506
0.5428	8.0	120	0.5146
0.5235	9.0	135	0.5024
0.5368	10.0	150	0.5011

Framework versions

• Transformers 4.40.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1