CoolFace
Modelpublic

forkjoin-ai/loki-erotica-8b-gguf

sourceHugging Facellama3.1updated 3mo agoView on Hugging Face
1likes587downloads
Model Card

Loki Erotica 8b (GGUF, Q4KM)

Adult-content research artifact. GGUF quantization of mradermacher/Loki-v2.75b-8b-EROTICA-1024k-i1-GGUF for explicit opt-in local and KNOT runtime experiments. This repo does not contain training data and does not assert rights over the upstream model or datasets.

Content Notice

This model is intended for adult-text research and evaluation workflows only. Downstream users are responsible for age gating, consent, policy controls, and compliance with the upstream model license and applicable law.

Model Details

PropertyValue
Base modelmradermacher/Loki-v2.75b-8b-EROTICA-1024k-i1-GGUF
Parameters8B
ArchitectureLLaMA
QuantizationQ4KM
FormatGGUF
Size~5 GB
Licensellama3.1

Usage

With llama.cpp

bash
./llama-cli -m Loki-v2.75b-8b-EROTICA-1024k.i1-Q4_K_M.gguf -p "Your prompt here" -n 256

With Aether / Moonshine

This KNOT can be loaded by the Aether/Moonshine local runtime when explicitly selected, for example with ZEDGE_MODELS=loki-erotica-8b ZEDGE_MOONSHINE_MODEL=loki-erotica-8b.

Also available: .knot (sovereign format)

This repo ships `loki-erotica-8b.knot` — the model weights in the KNOT container that the Aether distributed-inference runtime loads natively (the GGUF, when present, sits right beside it). A KNOT is a single self-describing file with a JSON table-of-contents, so any single tensor is one HTTP `Range` request — ideal for streaming weights to edge nodes.

GGUFKNOT
Containerformat-specific headersingle file, JSON table-of-contents
Per-tensor fetchwhole-file orientedone tensor = one Range request
Ecosystembroad (llama.cpp, …)Aether / Gnosis runtime
bash
huggingface-cli download forkjoin-ai/loki-erotica-8b-gguf loki-erotica-8b.knot --local-dir ./knots

Full format spec: KNOT_FORMAT.md. Inspect the header with bun run open-source/bitwise/scripts/dump-knot.ts loki-erotica-8b.knot.

Deployment Architecture

This model targets the Aether distributed inference runtime — a custom engine that shards model layers across multiple nodes for parallel execution:

  1. 1.Coordinator receives requests and manages token generation
  2. 2.Layer nodes each hold a subset of model layers (2 nodes for this model)
  3. 3.Hidden states flow between nodes via gRPC
  4. 4.Range-addressed KNOT weights stream from R2 or local cache

For Edgework pickup, the KNOT artifact is published at models/loki-erotica-8b.knot in the distributed-inference R2 bucket and is not added to the generic public chat-completions allowlist.

About

Published by AFFECTIVELY · Managed by @buley

We publish KNOT artifacts for explicit opt-in edge inference research via the Aether runtime.