llamaduo_synth_ds_v0.1

This model is a fine-tuned version of google/gemma-7b on the chansung/synth_ds dataset. It achieves the following results on the evaluation set:

• Loss: 3.8292

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 3
• gradient_accumulation_steps: 2
• total_train_batch_size: 12
• total_eval_batch_size: 12
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
0.7837	0.9995	939	1.9731
0.7509	2.0	1879	1.9719
0.7086	2.9995	2817	2.0286
0.6156	4.0	3757	2.1647
0.4937	4.9995	4696	2.3686
0.4075	6.0	5636	2.7269
0.3395	6.9995	6575	3.1681
0.2962	8.0	7515	3.6134
0.284	8.9995	8454	3.8100
0.2782	9.9957	9390	3.8292

Framework versions

• PEFT 0.10.0
• Transformers 4.40.0
• Pytorch 2.2.2+cu121
• Datasets 2.14.6
• Tokenizers 0.19.1