metadata
license: bigcode-openrail-m
library_name: peft
tags:
- generated_from_trainer
base_model: bigcode/starcoder2-3b
model-index:
- name: starcoder-3b-hugcoder
results: []
starcoder-3b-hugcoder
This model is a fine-tuned version of bigcode/starcoder2-3b on smangrul/hug_stack dataset. It achieves the following results on the evaluation set:
- Loss: 0.5545
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: 16
- eval_batch_size: 16
- seed: 11
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 2000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.7019 | 0.05 | 100 | 0.7098 |
| 0.6817 | 0.1 | 200 | 0.6754 |
| 0.5772 | 0.15 | 300 | 0.6445 |
| 0.5836 | 0.2 | 400 | 0.6228 |
| 0.6057 | 0.25 | 500 | 0.6072 |
| 0.5309 | 0.3 | 600 | 0.5943 |
| 0.5099 | 0.35 | 700 | 0.5862 |
| 0.5115 | 0.4 | 800 | 0.5781 |
| 0.5103 | 0.45 | 900 | 0.5728 |
| 0.4101 | 0.5 | 1000 | 0.5685 |
| 0.4694 | 0.55 | 1100 | 0.5636 |
| 0.4364 | 0.6 | 1200 | 0.5605 |
| 0.4516 | 0.65 | 1300 | 0.5577 |
| 0.4283 | 0.7 | 1400 | 0.5570 |
| 0.4324 | 0.75 | 1500 | 0.5550 |
| 0.4693 | 0.8 | 1600 | 0.5554 |
| 0.444 | 0.85 | 1700 | 0.5548 |
| 0.4608 | 0.9 | 1800 | 0.5538 |
| 0.3891 | 0.95 | 1900 | 0.5548 |
| 0.4028 | 1.0 | 2000 | 0.5545 |
Framework versions
- PEFT 0.9.1.dev0
- Transformers 4.39.0.dev0
- Pytorch 2.1.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2