FunDialogues/dollygem-2b-LoRA

gemma-2b-dolly-qa

This model is a fine-tuned version of google/gemma-2b on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 2.0215

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

databricks/databricks-dolly-15k

Training Hardware

This model was trained using Intel(R) Data Center GPU Max 1100

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• training_steps: 1480

Training results

Training Loss	Epoch	Step	Validation Loss
2.9198	1.64	100	2.5675
2.437	3.28	200	2.2818
2.2514	4.92	300	2.1677
2.1587	6.56	400	2.1038
2.116	8.2	500	2.0741
2.0794	9.84	600	2.0576
2.0663	11.48	700	2.0467
2.0494	13.11	800	2.0394
2.0449	14.75	900	2.0336
2.0336	16.39	1000	2.0293
2.0281	18.03	1100	2.0262
2.0172	19.67	1200	2.0240
2.0227	21.31	1300	2.0227
2.0128	22.95	1400	2.0215

Framework versions

• PEFT 0.10.0
• Transformers 4.39.3
• Pytorch 2.0.1a0+cxx11.abi
• Datasets 2.18.0
• Tokenizers 0.15.2