Look inside any local LLM, find the parts it doesn't need, and make it smaller.
NeuronDump reads GGUF model files (the format used by Ollama and llama.cpp) and shows what is inside: layers, attention heads, vocabulary, chat template, weight statistics. It can then remove unused parts and write a smaller GGUF that still runs in Ollama and llama.cpp.
It works with any model. It reads each file's own metadata; nothing is hard-coded per model.
Download: ready-made Windows and Linux builds are on the
releases page. Unzip and run neurondump list.
Download steps and checks: install guide.
Build from source: you need Git and Rust 1.88 or newer. Step-by-step instructions: Windows and Linux install guide.
Windows (PowerShell):
winget install Git.Git
winget install Rustlang.Rustup # then reopen PowerShell
git clone https://github.com/aymen94/NeuronDump
cd NeuronDump
cargo install --path crates/neurondump
neurondump list
Linux:
sudo apt install build-essential git curl # or dnf / pacman
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh && source "$HOME/.cargo/env"
git clone https://github.com/aymen94/NeuronDump
cd NeuronDump
cargo install --path crates/neurondump
neurondump list
Add --features trace to cargo install to also run models and watch each layer.
neurondump list # every local model
neurondump inspect llama3.2:1b # what's inside a model
neurondump vocab llama3.2:1b --search hello
neurondump template llama3.2:1b # the chat template as a tree
neurondump analyze llama3.2:1b --out plan.json
neurondump slim llama3.2:1b --plan plan.json --only vocab --out small.gguf
neurondump web --open # the same, in your browser
Run neurondump with no arguments to choose between the interactive CLI and the web interface.
The web interface draws the model as a graph you can open up, in 3D (tower, city or force layout) or 2D. Everything is bundled, so it works offline.
Render inside any part:
- double-click a layer to see its attention and feed-forward blocks;
- double-click attention to see its KV groups, then its individual heads;
- feed-forward opens into neuron buckets, and MoE into experts.
Inspect:
- hover any part for its size and flag;
- click it for its path, contents, connections, weight statistics, activations, importance,
and whether it can be removed (
safe,riskyorbreaks_model) and why.
Search inside: type a tensor name likeblk.3.ffn_down . Closed layers open to show it. #
Cut (edit): pressCut this on any part (a layer, a block, a head, a token range) to add it to a cut selection. Cut parts get a red outline. #
Direct impact, instantly: the panel shows:
- parameters removed or zeroed, and the new file size;
- what gets disconnected and what depends on the cut;
- whether llama.cpp still loads the file;
- a verdict. Dangerous cuts are refused unless you force them.
Simulated impact (--features trace ): the cut model runs in memory next to the original.
You compare their next tokens, KL divergence, generated text, the logit lens per layer and
perplexity. Nothing is written. #
Sweep: cuts each layer (or head) in turn, ranks them by damage, and colors the graph with it. #
Save as plan → Slim: writes the smaller GGUF and an Ollama Modelfile.
The same works from the command line:
neurondump graph llama3.2:1b --expand layer.7.attn # inside a layer's attention
neurondump node llama3.2:1b layer.7.attn.head.3 # inspect one head
neurondump ablate llama3.2:1b --cut layer.7 --static # direct impact
neurondump ablate llama3.2:1b --cut layer.7 --prompt "The capital of Italy is" # simulated impact
neurondump sweep llama3.2:1b --corpus texts.txt --save-plan plan.json --top 2
neurondump slim llama3.2:1b --plan plan.json --out llama-14.gguf
Details: Graph and ablation.
| Page | What it covers |
|---|---|
| Install and run | Windows and Linux, step by step |
| Getting started | find your models, first commands |
| How it works | what NeuronDump reads, and how |
| CLI reference | every command and flag, and the REPL |
| Web interface | the browser UI and the HTTP API |
| Analyze and slim | making models smaller, safely |
| Trace | running a model and looking at each layer |
| Graph and ablation | the 3D / 2D graph: look inside layers, inspect, search, cut, direct and simulated impact |
| What's inside real models | Qwen3 8B vs DeepSeek-R1, and Qwen 3.8 |
| Building | build options and Windows notes |
| Manual test plan | step-by-step checks |
A short illustrated explanation is on the website
(source: docs/index.html).
Early version, tested on Windows with Ollama models. Known gaps are listed at the end of TEST.md. Bug reports and ideas: issues.
Licensed under the MIT License.