trained-una-cybertron-7b-v2-bf16

This model is a fine-tuned version of fblgit/una-cybertron-7b-v2-bf16 on an https://huggingface.co/datasets/Nan-Do/instructional_code-search-net-ruby.

It achieves the following results on the evaluation set:

• Loss: 1.0274

Model description

• intention is to use this with system message for generating ruby codes

Intended uses & limitations

• writing rspec / ruby codes

Training and evaluation data

• https://huggingface.co/datasets/Nan-Do/instructional_code-search-net-ruby
• https://huggingface.co/datasets/Nan-Do/instructional_code-search-net-ruby

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.0859	0.39	50	1.0936
1.036	0.78	100	1.0274

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.14.1