Still in progress, trying to get working between my 3 laptops, with no luck so far.
gemma-4-31B-it-qat-UD-Q4_K_XL
Distributed GGUF inference package for Mesh LLM
GGUF layer package for running gemma-4-31B-it-qat-UD-Q4_K_XL across a local Mesh LLM cluster.
This package is derived from unsloth/gemma-4-12B-it-qat-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 | UD-Q4_K_XL layer package |
Model Overview
| Property | Value |
|---|---|
| Source model | unsloth/gemma-4-12B-it-qat-GGUF |
| Model id | unsloth/gemma-4-12B-it-qat-GGUF:UD-Q4_K_XL |
| Family | Gemma |
| Parameter scale | 12B |
| Quantization | UD-Q4_K_XL |
| Layer count | 48 |
| Activation width | 3840 |
| Package size | 6.3 GB |
| Source file | gemma-4-31B-it-qat-UD-Q4_K_XL.gguf |
| Package repo | meshllm/gemma-4-31B-it-qat-UD-Q4_K_XL-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/gemma-4-12B-it-qat-GGUF.
Quickstart
# Run this on each machine that should contribute memory/compute.
mesh-llm serve --model "meshllm/gemma-4-31B-it-qat-UD-Q4_K_XL-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/gemma-4-12B-it-qat-GGUF:UD-Q4_K_XL",
"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/gemma-4-12B-it-qat-GGUF@main/gemma-4-31B-it-qat-UD-Q4_K_XL.gguf |
| Source revision | main |
| Source SHA-256 | cc9ff072e0a8203429ed854e6662c17a6c2bc1e5dca5b475dd4736caaacbc165 |
| Skippy ABI | 0.1.24 |
What Is Included
| Artifact | Path | Contents | SHA-256 |
|---|---|---|---|
| Manifest | model-package.json | Package schema, source identity, checksums | e6ba5ec739fd602ebcfd275968e344a6dc574e087d08852250f8b86e700a8a69 |
| Metadata | shared/metadata.gguf | 1 tensors, 15.1 MB | 6e220c19850b60ff25cf3b0e1dd5119dd4bc7af4cd4daab648c44f5f23920fd3 |
| Embeddings | shared/embeddings.gguf | 2 tensors, 771.1 MB | 4a78029414b01d0951eb20e0735bebd6c5bbb80489ab542fbba10c51c9e25bd6 |
| Output head | shared/output.gguf | 2 tensors, 15.1 MB | 5c182e6351e1354a35cd37287c9f0f20c6f312ec462f7d5ba1d41cf88d7f3173 |
| Transformer layers | layers/layer-*.gguf | 60 layer artifacts, 890 tensors, 16.2 GB | see model-package.json |
Validation
Generated by the Mesh LLM HF Jobs splitter from mesh-llm ref 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/gemma-4-31B-it-qat-UD-Q4_K_XL.gguf" --out-dir "/tmp/meshllm-layer-job-meshllm_gemma-4-31B-it-qat-UD-Q4_K_XL-layers-192/package"
Links
- Source model: unsloth/gemma-4-12B-it-qat-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