Modelpublic

OrchardPair/Mistral-Small-24B-Instruct-2501-4bit

sourceHugging Faceapache-2.0updated 4mo agoView on Hugging Face

0likes48downloads

Model Card

Mistral Small 24B Instruct 2501 — 4-bit MLX (OrchardPair canonical mirror)

A byte-faithful mirror of `mlx-community/Mistral-Small-24B-Instruct-2501-4bit`, re-hosted under the OrchardPair namespace as a revision-pinned canonical source for the OrchardPair app's on-device Apple-silicon MLX inference. No weights, tokenizer, or configuration were modified — the files here are bit-for-bit identical to the upstream quantization (verify via the sha256 manifest below).

Provenance & license

—Upstream quant: mlx-community/Mistral-Small-24B-Instruct-2501-4bit (4-bit MLX quantization).
—Base model: Mistral Small 24B Instruct 2501 (Mistral AI).
—License: Apache-2.0 — inherited unchanged from upstream. All upstream attribution and terms apply.
—Why mirrored: OrchardPair pins an immutable revision + per-file sha256 so the app downloads a verified, stable artifact that cannot drift if the upstream main branch changes.

File manifest (sha256)

File	Size (bytes)	sha256
`config.json`	825	`8b4413498badd74aee46afc4071633533c0749883024394afb4cc1c96b6c80f1`
`tokenizer.json`	17078037	`b76085f9923309d873994d444989f7eb6ec074b06f25b58f1e8d7b7741070949`
`tokenizer_config.json`	198489	`19b35bf8e9a0d4c8c48e4a5773963e6b9872ae3b65c1f44e8152ce54fb48c1ab`
`model-00001-of-00003.safetensors`	5285171707	`4c93bf99bd4f467c55f77eb9555685cf2c00bce50e8b7d22f3b300bc6327b005`
`model-00002-of-00003.safetensors`	5314706961	`1d83f155ea967330209f8eeb226cdb4468d932e9b9cb3579f674474f52c0b777`
`model-00003-of-00003.safetensors`	2660299513	`570ee17d38facfcf6fc8000c7b3b6e71ecbaee2180c976a107d6cc69129fcdf8`
`model.safetensors.index.json`	79327	`42180bd2e96e647bf8528c861f5c25adfbf8dfa87e12fe6e7662df4beeafe5f0`