You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Mistral-7B — fraQtl Compressed

14.48 GB → 9.84 GB. Near-zero quality loss.

Metric	Value
Original size	14.48 GB
Compressed size	9.84 GB
PPL delta (wikitext-2)	+0.35
Source model	mistralai/Mistral-7B-Instruct-v0.2
NIAH retrieval	3/3 preserved

Mistral-7B-Instruct compressed with fraQtl. Same architecture, smaller files, near-zero quality loss.

fraQtl Demo — this model running with KV cache compression on top.

Weights are gated. For access or integration support: contact@fraqtl.ai

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support