Kesar v2

By Nagchetra Labs | Revant Sahu (15 y/o CEO)

India's first student-built fine-tuned LLM, powering Edlix AI.

Model Details

  • Base Model: Qwen/Qwen3.5-4B
  • Architecture: Qwen3_5ForCausalLM
  • Parameters: ~4B
  • Hidden Size: 2560
  • Layers: 32
  • Context Length: 262,144 tokens ๐Ÿ”ฅ
  • dtype: float16

Training

  • Method: LoRA fine-tuning
  • Dataset: 2000 custom examples
  • Epochs: 1
  • Purpose: Indian student AI assistant

Languages

  • English (primary)
  • Hindi / Hinglish (fine-tuned)
  • Chinese (base model)

About

Built by Revant Kumar Sahu โ€” 15 year old founder of Nagchetra Labs. Powers Edlix AI.

Links

  • ๐ŸŒ Edlix AI: edlixai.netlify.app
  • ๐Ÿข Nagchetra Labs: nagchetra.netlify.app
Downloads last month
7
Safetensors
Model size
4B params
Tensor type
F16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ 2 Ask for provider support

Model tree for rev315/kesar-2.0

Finetuned
Qwen/Qwen3.5-4B
Adapter
(1013)
this model
Quantizations
2 models