Fine-tuning LiquidAI/LFM2-VL-1.6B in Colab (LoRA/4-bit) + dataset template + probe test.
-
Updated
Jan 5, 2026 - Jupyter Notebook
#
lfm2-vl
Here are 3 public repositories matching this topic...
Browser-native invoice OCR & structured extraction. A vision-language model (LFM2-VL) runs fully client-side via WebGPU (WASM fallback) with Transformers.js — no server, no API, no data leaves your machine. Two-pass pipeline + a deterministic, zero-ML validation layer.
react privacy typescript ocr pwa offline-first client-side webgpu vite document-ai onnxruntime-web vision-language-model transformers-js invoice-ocr lfm2-vl uae-vat
-
Updated
Jun 10, 2026 - TypeScript
Satellite-edge brick-kiln triage. Strict YOLO + Liquid LFM2.5-VL-450M reason over crops onboard, four-tier transmission gate downlinks evidence instead of empty fields. Built for the Liquid AI x DPhi 'AI in Space' hackathon.
computer-vision hackathon geospatial remote-sensing yolo edge-ai streamlit compliance-monitoring liquid-ai ultralytics vision-language-model lfm2-vl satellite-edge brick-kiln-detection ai-in-space
-
Updated
May 9, 2026 - Python
Improve this page
Add a description, image, and links to the lfm2-vl topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the lfm2-vl topic, visit your repo's landing page and select "manage topics."