{"slug": "spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser", "title": "SpawnDev.ILGPU.ML: Neural network inference & training in C# across any GPU (Browser & Desktop)", "summary": "Developer Todd (\"LostBeard\") released SpawnDev.ILGPU.ML, an open-source neural network inference and training engine written entirely in C# that compiles compute kernels for six backends: WebGPU, WebGL, WebAssembly, CUDA, OpenCL, and SIMD CPU. The engine requires no ONNX Runtime or native inference binaries, auto-detects model formats including ONNX, TFLite, GGUF, SafeTensors, TF GraphDef, PyTorch checkpoints, and Core ML, and supports on-device LoRA fine-tuning of GGUF language models such as llama3.1:8b on an RTX 4070. It runs on Windows, Linux, and macOS or in the browser via Blazor WebAssembly, with a zero-copy architecture that streams data from fetch and OPFS directly into GPU buffers.", "body_md": "Hey everyone!\n\nI’ve been watching Wendell’s videos on Level1Techs for a long time and really appreciate how supportive this community is of open-source projects, so I wanted to share something I’ve been building and refining: **SpawnDev.ILGPU.ML**.\n\nIt is a neural network inference and training engine written entirely in C# that compiles C# compute kernels directly for whatever device is in front of it via SpawnDev.ILGPU.\n\n**The big kicker:** **No ONNX Runtime, no native inference binaries, and no server required.** It runs entirely locally on the desktop (Windows, Linux, macOS via CUDA, OpenCL, or SIMD CPU) or right inside the browser via Blazor WebAssembly (WebGPU, WebGL, or WebAssembly).\n\n**One Codebase, Six Backends:** WebGPU, WebGL, Wasm, CUDA, OpenCL, and SIMD CPU. Everything is tested across lanes.\n\n**Zero-Copy Architecture:** Pixels, activations, and weights stay on the GPU. In the browser, data streams from `fetch` and OPFS straight into GPU buffers without bloating the .NET heap.\n\n**Model Formats Auto-Detected:** Drop in ONNX, TFLite, GGUF, SafeTensors, TF GraphDef, PyTorch checkpoints, or Core ML. It builds a unified graph IR with a built-in graph optimizer (constant folding, fusion, dead-node elimination).\n\n**Generative AI & 3D Pipelines:** Runs things like FLUX 3 / FLUX.2, Stable Diffusion (with LoRA, ControlNet, and ComfyUI patches), LTX-Video & Wan 2.2 for video gen, TRELLIS.2 for text/image-to-3D, SkinTokens/UniRig for auto-rigging, Kimodo for animation, and GGUF LLMs (with KV-cache decode and TurboQuant).\n\n**On-Device Training:** Supports LoRA fine-tuning of GGUF language models directly on your hardware (e.g., fine-tuning `llama3.1:8b` on an RTX 4070).\n\nIf you want to kick the tires without installing anything, the live demo runs entirely client-side on your GPU (models download once and cache locally). You can find the demo on the GitHub repo page (this forum blocks links).\n\ngithub / LostBeard / SpawnDev.ILGPU.ML\n\nI’d love to hear your thoughts, answer any architectural questions, or see what people think of running heavy generative and 3D workflows straight out of a browser tab via WebGPU!\n\nTodd\n\naka LostBeard", "url": "https://wpnews.pro/news/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser", "canonical_source": "https://forum.level1techs.com/t/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser-desktop/260474#post_1", "published_at": "2026-11-11 22:46:23+00:00", "updated_at": "2026-11-11 22:50:09.886458+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-tools", "ai-infrastructure", "large-language-models"], "entities": ["SpawnDev.ILGPU.ML", "SpawnDev.ILGPU", "Todd", "LostBeard", "Blazor WebAssembly", "WebGPU", "ONNX Runtime", "llama3.1:8b"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser", "markdown": "https://wpnews.pro/news/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser.md", "text": "https://wpnews.pro/news/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser.txt", "jsonld": "https://wpnews.pro/news/spawndev-ilgpu-ml-neural-network-inference-training-in-c-across-any-gpu-browser.jsonld"}}