Property	Value
Base Model	mistralai/Mistral-7B-Instruct-v0.3
Method	QLoRA (4-bit quantization + LoRA, r=64)
Trainable Parameters	167,772,160 (2.26% of 7.4B)
Training Library	Unsloth
Language	Telugu (తెలుగు)
Domain	K-12 Education (CBSE/NCERT, Grades 6-12)
Training Data	~408K curriculum-aligned Q&A pairs
License	Apache 2.0

Quick Start (Unsloth -- Fastest)

python

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="FoundryAILabs/bharat-telugu-7b-lora",
    max_seq_length=2048,
    load_in_4bit=True,
)
FastLanguageModel.for_inference(model)

inputs = tokenizer("[INST] What is photosynthesis? [/INST]", return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Using with HuggingFace Transformers

python

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3", load_in_4bit=True, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")
model = PeftModel.from_pretrained(base, "FoundryAILabs/bharat-telugu-7b-lora")

All BharatLLM Models

Model	Language	Type
FoundryAILabs/bharat-english-7b-lora	English	K-12
FoundryAILabs/bharat-hindi-7b-lora	Hindi	K-12
FoundryAILabs/bharat-bengali-7b-lora	Bengali	K-12
FoundryAILabs/bharat-telugu-7b-lora	Telugu	K-12
FoundryAILabs/bharat-tamil-7b-lora	Tamil	K-12
FoundryAILabs/bharat-kannada-7b-lora	Kannada	K-12
FoundryAILabs/bharat-malayalam-7b-lora	Malayalam	K-12
FoundryAILabs/bharat-marathi-7b-lora	Marathi	K-12
FoundryAILabs/bharat-gujarati-7b-lora	Gujarati	K-12
FoundryAILabs/bharat-odia-7b-lora	Odia	K-12
FoundryAILabs/bharat-punjabi-7b-lora	Punjabi	K-12
FoundryAILabs/bharat-urdu-7b-lora	Urdu	K-12
FoundryAILabs/bharat-btech-7b-lora	English	BTech Engineering

Website: foundryailabs.io | GitHub: github.com/foundryailabs/BharatLLM