NeuronDump: Open a local LLM as a 3D graph, cut parts, see what breaks

NeuronDump, a Rust-based open-source tool from developer aymen94, reads GGUF model files used by Ollama and llama.cpp and renders a local LLM as an explorable 3D graph, letting users cut layers, attention heads or token ranges and see the impact before writing a smaller GGUF. The tool reports parameters removed, new file size, what gets disconnected, whether llama.cpp still loads the file, and a verdict, refusing dangerous cuts unless forced; an optional trace feature runs the cut model in memory alongside the original to compare next tokens, KL divergence, logit lens per layer and perplexity. Ready-made Windows and Linux builds are on the GitHub releases page, and building from source requires Rust 1.88 or newer.

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 https://github.com/aymen94/NeuronDump/releases . Unzip and run neurondump list . Download steps and checks: install guide https://github.com/aymen94/NeuronDump/blob/main/docs/install.md download-a-ready-made-build . Build from source: you need Git https://git-scm.com and Rust https://rustup.rs 1.88 or newer. Step-by-step instructions: Windows and Linux install guide https://github.com/aymen94/NeuronDump/blob/main/docs/install.md . 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 , risky or breaks model and why. - Search inside: type a tensor name like blk.3.ffn down . Closed layers open to show it. - Cut edit : press Cut 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 https://github.com/aymen94/NeuronDump/blob/main/docs/graph-and-ablation.md . | Page | What it covers | |---|---| | Install and run https://github.com/aymen94/NeuronDump/blob/main/docs/install.md | Windows and Linux, step by step | | Getting started https://github.com/aymen94/NeuronDump/blob/main/docs/getting-started.md | find your models, first commands | | How it works https://github.com/aymen94/NeuronDump/blob/main/docs/internals.md | what NeuronDump reads, and how | | CLI reference https://github.com/aymen94/NeuronDump/blob/main/docs/cli.md | every command and flag, and the REPL | | Web interface https://github.com/aymen94/NeuronDump/blob/main/docs/web.md | the browser UI and the HTTP API | | Analyze and slim https://github.com/aymen94/NeuronDump/blob/main/docs/analyze-and-slim.md | making models smaller, safely | | Trace https://github.com/aymen94/NeuronDump/blob/main/docs/trace.md | running a model and looking at each layer | | Graph and ablation https://github.com/aymen94/NeuronDump/blob/main/docs/graph-and-ablation.md | the 3D / 2D graph: look inside layers, inspect, search, cut, direct and simulated impact | | What's inside real models https://github.com/aymen94/NeuronDump/blob/main/docs/models.md | Qwen3 8B vs DeepSeek-R1, and Qwen 3.8 | | Building https://github.com/aymen94/NeuronDump/blob/main/docs/building.md | build options and Windows notes | | Manual test plan https://github.com/aymen94/NeuronDump/blob/main/TEST.md | step-by-step checks | A short illustrated explanation is on the website https://aymen94.github.io/NeuronDump/ source: docs/index.html https://github.com/aymen94/NeuronDump/blob/main/docs/index.html . Early version, tested on Windows with Ollama models. Known gaps are listed at the end of TEST.md https://github.com/aymen94/NeuronDump/blob/main/TEST.md . Bug reports and ideas: issues https://github.com/aymen94/NeuronDump/issues . Licensed under the MIT License https://github.com/aymen94/NeuronDump/blob/main/LICENSE .