164 4 months ago

Sarvam-30B - Multilingual Indian LLM. Converted and packaged for Ollama.

ollama run predictivemanish/sarvam-30b

Models

View all →

Readme

Sarvam-30B for Ollama

This is a GGUF-converted version of Sarvam-30B, optimized for use with Ollama.

⚠️ Important Notes

  • This is a Mixture-of-Experts (MoE) model
  • May not run on all systems due to current Ollama limitations
  • Requires high RAM (~32GB+ recommended)

Model Details

  • Model: Sarvam-30B
  • Quantization: Q4_K_M
  • Format: GGUF
  • Size: ~20GB

Usage

ollama run predictivemanish/sarvam-30b

Known Issues

  • Some users may encounter:

    • unable to load model
  • This is due to:

    • MoE architecture compatibility

Credits

  • Original model: Sarvam AI
  • GGUF conversion: Hugging Face release