File size: 1,436 Bytes
0abf422 f4b6e8e 0abf422 f4b6e8e 0abf422 f4b6e8e 0abf422 cfe9960 0abf422 f4b6e8e 0abf422 1f9da74 63bf44a 0abf422 f4b6e8e 63bf44a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 |
---
library_name: transformers
license: apache-2.0
datasets:
- HuggingFaceTB/smollm-corpus
language:
- en
pipeline_tag: text-generation
---
# **Doge 60M checkpoint**
Doge uses `wsd_scheduler` as the training scheduler, which divides the learning rate into three stages: `warmup`, `stable`, and `decay`. It allows us to continue training on any new dataset from any checkpoint in the `stable stage` without spikes of the training.
Here are the initial learning rates required to continue training at each checkpoint:
- **[Doge-20M](https://huggingface.co./SmallDoge/Doge-20M-checkpoint)**: 8e-3
- **[Doge-60M](https://huggingface.co./SmallDoge/Doge-60M-checkpoint)**: 6e-3
- **[Doge-160M](https://huggingface.co./SmallDoge/Doge-160M-checkpoint)**: 4e-3
- **[Doge-320M](https://huggingface.co./SmallDoge/Doge-320M-checkpoint)**: 2e-3
| Model | Learning Rate | Schedule | Warmup Steps | Stable Steps |
|-------|---------------|----------|--------------|--------------|
| [Doge-20M](https://huggingface.co./SmallDoge/Doge-20M-checkpoint) | 8e-3 | wsd_scheduler | 800 | 6400 |
| [Doge-60M](https://huggingface.co./SmallDoge/Doge-60M-checkpoint) | 6e-3 | wsd_scheduler | 1600 | 12800 |
| [Doge-160M](https://huggingface.co./SmallDoge/Doge-160M-checkpoint) | 4e-3 | wsd_scheduler | 2400 | 19200 |
| [Doge-320M](https://huggingface.co./SmallDoge/Doge-320M-checkpoint) | 2e-3 | wsd_scheduler | 3200 | 25600 |
|