File size: 870 Bytes
27d413e c26d5fb 27d413e c26d5fb 27d413e c26d5fb 27d413e c26d5fb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 |
# Models with reduced matrix size and overfitted for testing
Base models and adapters for LLMs to be used to test converting to `gguf` and testing inference.
These LLMs are in `safetensors` format with their architecture unchanged but for the size of their tensors.
Folders structure:
```
<model architecture>/hidden_size=<reduced size>/<base|lora>
```
where:
- `base`: overfitted to complete this [Sagan's Pale Blue Dot paragraph](https://github.com/ltoniazzi/reduce-llms-for-testing/blob/main/data/pale_blue_dot.txt)
- `lora`: overfitted to complete this [Bohemian Rapshody paragraph](https://github.com/ltoniazzi/reduce-llms-for-testing/blob/main/data/bohemian_rapshody.txt)
These models are generated with the repo [reduce-llms-for-testing](https://github.com/ltoniazzi/reduce-llms-for-testing).
Please open an issue/PR there if you want to add more models. |