d1 for Ollaya

Ollaya package of LiquidAI/d1-3B-GGUF by Liquid AI. Ollaya runs open decision models locally, the way Ollama runs LLMs: typed questions in, calibrated answers out, behind a TypeSafe-compatible API.

ollaya run d1

What is in this repository

This repository holds only the files Ollaya derives, with no weights. The model is the authors' own GGUF file: ollaya pull downloads it from their repository, unmodified and pinned to a commit, verifies its sha256, and Ollaya runs it on llama.cpp.

Tag	The authors' GGUF	Files
d1:3b	LiquidAI/d1-3B-GGUF@bb1e436 d1-3B-Q8_0.gguf	3b/decision.json, 3b/calibration.json
d1:3b-vision	LiquidAI/d1-3B-GGUF@bb1e436 d1-3B-Q8_0.gguf + mmproj-d1-3B-Q8_0.gguf	3b-vision/decision.json, 3b-vision/calibration.json

Each tag has decision.json (the prompt, the option labels Ollaya reads and llama.cpp's settings) and calibration.json (temperatures). A vision tag also pulls the vision projector the table names, unmodified from its author's repository, which reads the images.

Parity

Ollaya's runner matches stock llama-server of the pinned build (b11146) on the same GGUF and projector, with prompts and answer-token groups from Liquid AI's pinned prompt.py: 650 text and 55 image questions on the CPU, an RTX 4090 and an RTX 5090 (29 and 24 on the contributor's CPU and RTX 4070), every decision the same, option log-probabilities within 7.7e-6.

License

Same as the upstream model (LFM Open License v1.0). Ollaya itself is Apache-2.0.