danfe-nepali-2b / README.md
bemohansingh's picture
Upload README.md with huggingface_hub
c51125f verified
metadata
language:
  - ne
  - en
  - mai
license: apache-2.0
tags:
  - nepali
  - nepal
  - danfe-ai
  - qlora
  - llama
base_model: HuggingFaceTB/SmolLM2-1.7B-Instruct

Danfe AI — Nepal Language Model

Nepal's first open-source AI model for 17+ languages.

Model Details

  • Base Model: HuggingFaceTB/SmolLM2-1.7B-Instruct
  • Method: QLoRA (4-bit quantization + LoRA rank 16)
  • Training Data: Nepali Wikipedia + curated Nepal knowledge
  • Languages: Nepali, English, Maithili, Doteli, Bajhangi, Achhami, +11 more

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base = AutoModelForCausalLM.from_pretrained("HuggingFaceTB/SmolLM2-1.7B-Instruct")
model = PeftModel.from_pretrained(base, "danfe-ai/danfe-nepali-2b")
tokenizer = AutoTokenizer.from_pretrained("danfe-ai/danfe-nepali-2b")

Part of Danfe AI Ecosystem

  • Chat UI, Translation, Speech, OCR
  • Crowdsource training platform
  • Focus on Sudurpashchim languages (Doteli, Bajhangi, Achhami)

Built with love for Nepal.