Spaces:
Running
Running
| title: README | |
| emoji: 💻 | |
| colorFrom: green | |
| colorTo: gray | |
| sdk: static | |
| pinned: false | |
| # Models for i | |
| A collection of open-weight language models converted to **big-endian** format so they run natively on **IBM i** (IBM Power) systems. | |
| ## Why this exists | |
| Most pretrained model weights are published in little-endian formats. IBM Power hardware running IBM i operates in big-endian mode, so weights have to be byte-swapped before they will load correctly. The models published here have already been converted and validated, removing that step from the deployment path. | |
| ## What you'll find | |
| - Big-endian variants of popular open-weight models | |
| - Same architectures, tokenizers, and quantization options as the upstream sources | |
| - Ready to load with common IBM i AI runtimes — no runtime conversion required | |
| ## Getting started | |
| Tooling, conversion scripts, and IBM i runtime guidance live in the companion repository: | |
| ➡️ **[github.com/ajshedivy/models-for-i](https://github.com/ajshedivy/models-for-i)** | |