Instructions to use chhatramani/nyayalm-SentencePiece_BPE-64k with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use chhatramani/nyayalm-SentencePiece_BPE-64k with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="chhatramani/nyayalm-SentencePiece_BPE-64k")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("chhatramani/nyayalm-SentencePiece_BPE-64k", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use chhatramani/nyayalm-SentencePiece_BPE-64k with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "chhatramani/nyayalm-SentencePiece_BPE-64k"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chhatramani/nyayalm-SentencePiece_BPE-64k",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/chhatramani/nyayalm-SentencePiece_BPE-64k

SGLang

How to use chhatramani/nyayalm-SentencePiece_BPE-64k with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "chhatramani/nyayalm-SentencePiece_BPE-64k" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chhatramani/nyayalm-SentencePiece_BPE-64k",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "chhatramani/nyayalm-SentencePiece_BPE-64k" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chhatramani/nyayalm-SentencePiece_BPE-64k",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use chhatramani/nyayalm-SentencePiece_BPE-64k with Docker Model Runner:
```
docker model run hf.co/chhatramani/nyayalm-SentencePiece_BPE-64k
```

NyayaLM 64K SentencePiece-BPE Tokenizer

A production-grade 64 000-vocab SentencePiece BPE tokenizer trained for Nepali (Devanagari), Romanized Nepali, and English, with special focus on legal and institutional text.

This is the SentencePiece_BPE tokenizer, trained on the exact same corpus and vocabulary budget for fair comparison with BPE only Tokenizer.

Highlights

Property	Value
Vocabulary size	64 000
Algorithm	SentencePiece BPE + byte_fallback
Scripts	Devanagari + Latin
Languages	Nepali, Romanized Nepali, English
Special tokens	Qwen3 style (high IDs)
EOS	`<
PAD / document sep.	`<
BOS	None (Qwen-style)
Character coverage	0.9995
License	Apache 2.0

Quick start

from transformers import AutoTokenizer

tok = AutoTokenizer.from_pretrained("chhatramani/nyayalm-SentencePiece_BPE-64k")

text = "नेपालको संविधान २०७२ ले मौलिक हकको ग्यारेन्टी गर्दछ।"
ids = tok.encode(text, add_special_tokens=False)
print(len(ids), tok.decode(ids))

messages = [
    {"role": "system", "content": "You are a helpful legal assistant for Nepal."},
    {"role": "user", "content": "नेपालको संविधानमा मौलिक हक के-के छन्?"},
]
print(tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True))

Citation

@misc{nyayalm_tokenizer_64k_bpe,
  title        = {NyayaLM 64K SentencePiece-BPE Tokenizer for Nepali, Romanized Nepali, and English},
  author       = {Chhatramani Yadav},
  year         = {2026},
  howpublished = {\url{https://huggingface.co/chhatramani/nyayalm-SentencePiece_BPE-64k}},
}

License

Apache License 2.0.

Downloads last month: -; Downloads are not tracked for this model. How to track