phi-1.5-lora-tuned-Ondolly

This model is a fine-tuned version of microsoft/phi-1_5 on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 2.3867

Model description

More information needed

Intended uses & limitations

More information needed

Hardware

Trained model on Intel 1550 GPU

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: 10
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 40
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• training_steps: 590

Training results

Training Loss	Epoch	Step	Validation Loss
2.7732	4.0	100	2.5696
2.5296	8.0	200	2.4569
2.4628	12.0	300	2.4189
2.4324	16.0	400	2.3982
2.4136	20.0	500	2.3867

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.1.0.post0+cxx11.abi
• Datasets 2.19.1
• Tokenizers 0.19.1