CoolFace
Modelpublic

forkjoin-ai/falcon-mamba-7b-safetensors

sourceHugging Faceapache-2.0updated 4mo agoView on Hugging Face
0likes
Model Card

Falcon Mamba 7b (KNOT)

Production-ready KNOT (sovereign-format) mirror of tiiuae/falcon-mamba-7b for distributed text generation and conversation — powered by the Aether edge inference runtime on Edgework.ai.

Model Details

PropertyValue
Base modeltiiuae/falcon-mamba-7b
Parameters7B
ArchitectureFalcon-Mamba
Quantization— (lossless container)
FormatKNOT
Size~14 GB
Licenseapache-2.0

Also available: .knot (sovereign format)

This repo ships `falcon-mamba-7b.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/falcon-mamba-7b-safetensors falcon-mamba-7b.knot --local-dir ./knots

Full format spec: KNOT_FORMAT.md. Inspect the header with bun run open-source/bitwise/scripts/dump-knot.ts falcon-mamba-7b.knot.

Deployment Architecture

This model runs on 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.Zero cold start via warm pool scheduling

Deployed via Edgework.ai — bringing fast, cheap, and private inference as close to the user as possible.

About

Published by AFFECTIVELY · Managed by @buley

We quantize and publish production-ready models for distributed edge inference via the Aether runtime. Every release is tested for correctness and stability before publication.