metadata
base_model: codellama/CodeLlama-7b-Instruct-hf
library_name: peft
license: llama2
tags:
- generated_from_trainer
model-index:
- name: codellama-adb-sdk
results: []
codellama-adb-sdk
This model is a fine-tuned version of codellama/CodeLlama-7b-Instruct-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.9731
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: 0.0003
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 2000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.2695 | 0.05 | 100 | 0.5778 |
| 0.0568 | 0.1 | 200 | 0.7456 |
| 0.0411 | 0.15 | 300 | 0.8092 |
| 0.0366 | 0.2 | 400 | 0.8567 |
| 0.032 | 0.25 | 500 | 0.8574 |
| 0.0256 | 0.3 | 600 | 0.8451 |
| 0.0258 | 0.35 | 700 | 0.8729 |
| 0.0245 | 0.4 | 800 | 0.8760 |
| 0.0222 | 0.45 | 900 | 0.8999 |
| 0.0199 | 0.5 | 1000 | 0.8617 |
| 0.0215 | 0.55 | 1100 | 0.9148 |
| 0.0226 | 0.6 | 1200 | 0.9237 |
| 0.0206 | 0.65 | 1300 | 0.9307 |
| 0.0209 | 0.7 | 1400 | 0.9392 |
| 0.0184 | 0.75 | 1500 | 0.9659 |
| 0.0189 | 0.8 | 1600 | 0.9492 |
| 0.0202 | 0.85 | 1700 | 0.9660 |
| 0.0192 | 0.9 | 1800 | 0.9668 |
| 0.0176 | 0.95 | 1900 | 0.9722 |
| 0.0186 | 1.0 | 2000 | 0.9731 |
Framework versions
- PEFT 0.13.2.dev0
- Transformers 4.46.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1