mhhmm's picture
Upload 18 files
4396cbf verified
metadata
license: llama2
library_name: peft
tags:
  - generated_from_trainer
base_model: codellama/CodeLlama-13b-Instruct-hf
model-index:
  - name: lora-out
    results: []

Built with Axolotl

lora-out

This model is a fine-tuned version of codellama/CodeLlama-13b-Instruct-hf on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4172

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: 8
  • eval_batch_size: 8
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 2
  • total_train_batch_size: 16
  • total_eval_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss
0.7046 0.01 1 0.6695
0.6348 0.05 7 0.6183
0.5056 0.1 14 0.4993
0.5127 0.15 21 0.4682
0.4663 0.2 28 0.4552
0.5534 0.25 35 0.4419
0.5231 0.3 42 0.4369
0.5045 0.35 49 0.4338
0.5444 0.4 56 0.4314
0.4922 0.45 63 0.4296
0.487 0.5 70 0.4261
0.4627 0.55 77 0.4230
0.5143 0.6 84 0.4210
0.4429 0.65 91 0.4199
0.4491 0.71 98 0.4193
0.4808 0.76 105 0.4188
0.483 0.81 112 0.4176
0.5513 0.86 119 0.4177
0.4574 0.91 126 0.4170
0.4723 0.96 133 0.4172

Framework versions

  • Transformers 4.36.0.dev0
  • Pytorch 2.0.1+cu118
  • Datasets 2.15.0
  • Tokenizers 0.15.0

Training procedure

Framework versions

  • PEFT 0.6.0