GGUF layer package for running Llama-3.3-70B-Instruct-Q3_K_M across a local Mesh LLM cluster.
This package is derived from unsloth/Llama-3.3-70B-Instruct-GGUF and keeps the original GGUF distribution split into per-layer artifacts for distributed inference.
Highlights
| Run locally | Pool multiple machines | OpenAI-compatible | Package variant |
|---|---|---|---|
| Private inference on your hardware | Split layers across peers | Serve /v1/chat/completions locally | Q3_K_M layer package |
Model Overview
| Property | Value |
|---|---|
| Source model | unsloth/Llama-3.3-70B-Instruct-GGUF |
| Model id | unsloth/Llama-3.3-70B-Instruct-GGUF:Q3_K_M |
| Family | Llama |
| Parameter scale | 70B |
| Quantization | Q3_K_M |
| Layer count | 80 |
| Activation width | 8192 |
| Package size | 32.5 GB |
| Source file | Llama-3.3-70B-Instruct-Q3_K_M.gguf |
| Package repo | meshllm/Llama-3.3-70B-Instruct-Q3_K_M-draft-layers |
Recommended Use
- Local and private inference with Mesh LLM.
- Multi-machine serving when the full GGUF is too large for one host.
- OpenAI-compatible chat/completions workflows through Mesh LLM's local API.
For upstream architecture details, chat template guidance, sampling recommendations, license terms, and benchmark notes, see the source model card: unsloth/Llama-3.3-70B-Instruct-GGUF.
Quickstart
# Run this on each machine that should contribute memory/compute.
mesh-llm serve --model "meshllm/Llama-3.3-70B-Instruct-Q3_K_M-draft-layers" --split
# Check the mesh and discover the OpenAI-compatible model name.
curl -s http://localhost:3131/api/status
curl -s http://localhost:3131/v1/models
# Send an OpenAI-compatible chat request.
curl -s http://localhost:3131/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "unsloth/Llama-3.3-70B-Instruct-GGUF:Q3_K_M",
"messages": [{"role": "user", "content": "Write a tiny hello-world function in Rust."}],
"max_tokens": 128
}'
Package Variant
| Property | Value |
|---|---|
| Format | layer-package |
| Canonical source ref | unsloth/Llama-3.3-70B-Instruct-GGUF@main/Llama-3.3-70B-Instruct-Q3_K_M.gguf |
| Source revision | main |
| Source SHA-256 | 7f13297f95a1d2284c5d5b90e2cf3ba6b33f97b408d2af43a89ff3a42178a2f1 |
| Skippy ABI | 0.1.24 |
| Package manifest SHA-256 | 79ece8130ac881d79fcd78e6710d86c62a28de119b0c1f3b8226fffbfe6b8854 |
What Is Included
| Artifact | Path | Contents | SHA-256 |
|---|---|---|---|
| Manifest | model-package.json | Package schema, source identity, checksums | 79ece8130ac881d79fcd78e6710d86c62a28de119b0c1f3b8226fffbfe6b8854 |
| Metadata | shared/metadata.gguf | 1 tensors, 7.5 MB | 538e8df59d0738026552485c48a4568d42df951ea570638cf5fd34e2549d875a |
| Embeddings | shared/embeddings.gguf | 2 tensors, 438.0 MB | 88ee8430d36f2257c154a8bf26197f75a2a4600eaa44e7ab532cfc120aa4aa0f |
| Output head | shared/output.gguf | 3 tensors, 829.4 MB | 370ad9e6cc70504d87a1a2d23a03fd26c34579af19a6f7b5ac305e264f424ace |
| Transformer layers | layers/layer-*.gguf | 80 layer artifacts, 800 tensors, 31.3 GB | see model-package.json |
Validation
Generated by the Mesh LLM HF Jobs splitter from mesh-llm ref codex/package-declared-draft-spec-main.
Each artifact is checksummed as it is written, uploaded to this repository, and removed from the job workspace before the next artifact is produced.
skippy-model-package write-package "/source/Llama-3.3-70B-Instruct-Q3_K_M.gguf" --out-dir "/tmp/meshllm-layer-job-meshllm_Llama-3.3-70B-Instruct-Q3_K_M-draft-layers-194/package"
Links
- Source model: unsloth/Llama-3.3-70B-Instruct-GGUF
- Mesh LLM website: meshllm.cloud
- Mesh LLM: github.com/Mesh-LLM/mesh-llm
- Discord: discord.gg/rs6fmc63eN
- Package catalog: meshllm/catalog
- Package format: layer-package-repos.md