| --- |
| license: apache-2.0 |
| base_model: |
| - amyrmahdy/decima-small |
| - amyrmahdy/decima-base |
| - amyrmahdy/decima-agent |
| library_name: onnx |
| tags: |
| - ollaya |
| - onnx |
| - decision-model |
| - system-one |
| pipeline_tag: text-classification |
| --- |
| |
| # decima for Ollaya |
|
|
| [Ollaya](https://github.com/ollaya-dev/ollaya) package of **[amyrmahdy/decima-small](https://huggingface.co/amyrmahdy/decima-small)** and **[amyrmahdy/decima-base](https://huggingface.co/amyrmahdy/decima-base)** and **[amyrmahdy/decima-agent](https://huggingface.co/amyrmahdy/decima-agent)** by A. M. Madani (amyrmahdy). |
| Ollaya runs open decision models locally, the way Ollama runs LLMs: typed questions in, |
| calibrated answers out, behind a TypeSafe-compatible API. |
|
|
| ```sh |
| ollaya run decima |
| ``` |
|
|
| ## What is in this repository |
|
|
| This repository holds only the files Ollaya derives, with no weights. Each graph is an ONNX export of the |
| original model whose weights **reference the authors' own weight files by byte offset**, |
| so `ollaya pull` downloads the weights from the upstream repositories, unmodified and pinned to a |
| commit, and verifies their sha256. |
|
|
| | Tag | Upstream | Files | |
| |---|---|---| |
| | `decima:small` | [amyrmahdy/decima-small@2e7f4d0](https://huggingface.co/amyrmahdy/decima-small/tree/2e7f4d0757df0215f48f2a9b2b589e1f3a6348ed) | `small/model-fp32.onnx`, `small/decision.json`, `small/calibration.json` | |
| | `decima:base` | [amyrmahdy/decima-base@2468005](https://huggingface.co/amyrmahdy/decima-base/tree/2468005d5e48e95eb74072c32a6d9df164578071) | `base/model-fp32.onnx`, `base/decision.json`, `base/calibration.json` | |
| | `decima:agent` | [amyrmahdy/decima-agent@86a07aa](https://huggingface.co/amyrmahdy/decima-agent/tree/86a07aab1c340fa5869bdb754e57d3851a1d288a) | `agent/model-fp32.onnx`, `agent/decision.json`, `agent/calibration.json` | |
|
|
| Each tag has an fp32 graph, used on CPU and GPU. Each tag also has `decision.json` (sequence layout, special tokens) and |
| `calibration.json` (temperatures). |
|
|
| ## Parity |
|
|
| Ollaya's Rust runtime matches the author's own code (decima/model.py and systemone.py, fp32) on 581 questions from 122 requests for each of small, base and agent, on CPU and CUDA (RTX 4090 and RTX 5090): identical token rows and truncation, the same 18 rejected requests, the same decision on every question, scores within 2.0e-5 and probabilities within 5.5e-6. |
|
|
| ## License |
|
|
| Same as the upstream model (Apache-2.0). Ollaya itself is Apache-2.0. |
|
|
