A side-by-side comparison of release history, capabilities, licensing and model sizes.
| Attribute |
 MPT
|
 Pythia
|
|---|---|---|
| Developer | MosaicML | EleutherAI, Together |
| Current generation | – | – |
| Initial release | 2023-05-05 | 2023-02-13 |
| Instruct tuned | Yes | No |
| Coding | No | No |
| Finetuning | Yes | Yes |
| Open source | Yes | Yes |
| License | Apache 2.0 | Apache 2.0 |
| Model sizes | 7B, 30B | 1B, 1.4B, 2.8B, 6.9B, 12B |
| Links |
MPT-7B and MPT-30B are a set of models that are part of MosaicML's Foundation Series. Trained on 1T tokens, the developers state that MPT-7B matches the performance of LLaMA while also being open source, while MPT-30B outperforms the original GPT-3. In addition to the base model, the developers also offer MPT-Instruct, MPT-Chat, and MPT-7B-StoryWriter-65k+, the last of which is trained on a context length of 65K tokens.
The most recent (as of May 2023) effort from EleutherAI, Pythia is a set of LLMs trained on The Pile. While it appears to outperform OPT and GPTNeo, its performance against GPT-J is unclear. Versions of Pythia have also been instruct-tuned by the team at Together.
