Instructions to use Lyte/nanochat-darija-73m-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Lyte/nanochat-darija-73m-base with Transformers:

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

pipe = pipeline("text-generation", model="Lyte/nanochat-darija-73m-base", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Lyte/nanochat-darija-73m-base", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("Lyte/nanochat-darija-73m-base", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Lyte/nanochat-darija-73m-base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Lyte/nanochat-darija-73m-base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lyte/nanochat-darija-73m-base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Lyte/nanochat-darija-73m-base

SGLang

How to use Lyte/nanochat-darija-73m-base 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 "Lyte/nanochat-darija-73m-base" \
    --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": "Lyte/nanochat-darija-73m-base",
		"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 "Lyte/nanochat-darija-73m-base" \
        --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": "Lyte/nanochat-darija-73m-base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Lyte/nanochat-darija-73m-base with Docker Model Runner:
```
docker model run hf.co/Lyte/nanochat-darija-73m-base
```

nanochat-darija-73m-base

Base NanoChat causal language model for Moroccan Darija.

This repo is exported in Hugging Face Transformers format with custom model code. Load it with trust_remote_code=True.

Preview Checkpoint Notice

This is a pilot/test checkpoint, not the final full-data model. It was trained to validate the Darija data pipeline, tokenizer, NanoChat architecture export, and SFT workflow before a larger billion-plus-token training run.

The cleaned base corpus contains 5M Darija rows and approximately 2B tokens with the included tokenizer. That number describes the available cleaned corpus; this checkpoint was intentionally trained on a much smaller/shorter schedule.

Model Details

Parameters: 73.5M (73,531,538)
Context length: 2048
Vocab size: 32768
Layers: 6
Hidden size: 384
Attention heads: 3
Checkpoint tag: d6_target12
Checkpoint step: 1062
Export dtype: bfloat16

Training

Pretrained on the cleaned custom made Moroccan Darija FineWeb-Edu translation corpus.

The instruction-tuned variant is small and experimental. It is useful for lightweight Darija chat tests, but it is not reliable for math, factuality, code debugging, translation fidelity, or safety-critical decisions.

Usage

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Lyte/nanochat-darija-73m-base"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    trust_remote_code=True,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

inputs = tokenizer("العصور الوسطى هي", return_tensors="pt").to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=128,
    temperature=0.3,
    top_k=100,
    top_p=0.95,
    repetition_penalty=1.1,
    do_sample=True,
)
print(tokenizer.decode(outputs[0], skip_special_tokens=False))

Files

model.safetensors: model weights
config.json: NanoChat architecture config
generation_config.json: default sampling config
tokenizer.json, tokenizer_config.json, special_tokens_map.json: tokenizer files
configuration_nanochat.py, modeling_nanochat.py: custom Transformers code
nanochat_export.json: source checkpoint metadata

Limitations

This is a tiny model. Expect fluent-looking but wrong answers, repetition on some prompts, and brittle instruction following. Use it as a research artifact or local baseline, not as a production assistant.

Downloads last month: 37

Safetensors

Model size

73.5M params

Tensor type

BF16

Model tree for Lyte/nanochat-darija-73m-base

Finetunes

1 model