README.md

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
• lora: overfitted to complete this Bohemian Rapshody paragraph

These models are generated with the repo reduce-llms-for-testing. Please open an issue/PR there if you want to add more models.