Instructions to use ravikadam/ganesh-gemma4-e4b-v3-LiteRT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LiteRT
How to use ravikadam/ganesh-gemma4-e4b-v3-LiteRT with LiteRT:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- LiteRT-LM
How to use ravikadam/ganesh-gemma4-e4b-v3-LiteRT with LiteRT-LM:
# LiteRT-LM runs on various platforms (Android, iOS, Windows, Linux, macOS, IoT, Web/WASM) # and supports many APIs (C++, Python, Kotlin, Swift, JavaScript, Flutter). # For platform-specific integration guides, please refer to the official developer website: # https://ai.google.dev/edge/litert-lm # To try LiteRT-LM, the easiest way is to use our CLI tool. # 1. Install the LiteRT-LM CLI tool: pip install -U litert-lm # 2. Download and run this model locally: # See: https://ai.google.dev/edge/litert-lm/cli litert-lm run \ --from-huggingface-repo=ravikadam/ganesh-gemma4-e4b-v3-LiteRT \ --prompt="Write me a poem"
- Notebooks
- Google Colab
- Kaggle
Ganesh SLM v3 — LiteRT-LM build for Google AI Edge Gallery
Offline Ganesha assistant: shlokas, stotras, aartis, rituals, stories. Marathi / Hindi / English.
Created in service of Shri Ganesh by Ravi Kadam — https://www.linkedin.com/in/ravikadam/
Install
- Install Google AI Edge Gallery (Play Store on Android, App Store on iOS 17+)
- Import from Hugging Face URL:
https://huggingface.co/ravikadam/ganesh-gemma4-e4b-v3-LiteRT - Ask it:
सुखकर्ता दुखहर्ता आरती म्हण.
Files
model_q.litertlm— INT4 quantized, recommended for phonesmodel.litertlm— unquantized, larger, desktop/testing
UNVERIFIED ON DEVICE
This build has NOT yet been confirmed working on a real phone. Known upstream issue: some Gemma 4 E-series LiteRT exports load successfully then emit pad tokens (google-ai-edge/litert-torch#994). Test before relying on it.
Eval (v3, harness-corrected)
verbatim recall 7/7 exact (full Atharvashirsha, all 108 names) · deferral 1.00 ·
identity 1.00 · sensitive_safe 1.00 · calendar year-stamping 1.00 · language match 1.00.
Known weak: fabrication 0.14 (gate is 0.00), story variants 0.00.
Full model + weights: ravikadam/ganesh-gemma4-e4b-v3
- Downloads last month
- -