Instructions to use roneneldan/TinyStories-1M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use roneneldan/TinyStories-1M with Transformers:

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

pipe = pipeline("text-generation", model="roneneldan/TinyStories-1M")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("roneneldan/TinyStories-1M")
model = AutoModelForCausalLM.from_pretrained("roneneldan/TinyStories-1M", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use roneneldan/TinyStories-1M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "roneneldan/TinyStories-1M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roneneldan/TinyStories-1M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/roneneldan/TinyStories-1M

SGLang

How to use roneneldan/TinyStories-1M 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 "roneneldan/TinyStories-1M" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roneneldan/TinyStories-1M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "roneneldan/TinyStories-1M" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roneneldan/TinyStories-1M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use roneneldan/TinyStories-1M with Docker Model Runner:
```
docker model run hf.co/roneneldan/TinyStories-1M
```

License for the released checkpoints?

#15

by ShawnShen12138 - opened Sep 15

Discussion

ShawnShen12138

Sep 15

Hi Ronen — thank you for TinyStories; the 1M checkpoint is a gift for teaching.

I'm building a free, open, bilingual interactive essay ("Inside the Machine") and a small zero-dependency TypeScript library that reimplements the GPT-Neo forward pass and runs your TinyStories-1M weights entirely in the browser, so readers can watch a real model tokenize, attend over, and continue their own sentence. To do that I convert the checkpoint to fp16 safetensors and self-host the 7.5 MB file alongside the site (schools' content filters often block huggingface.co, so runtime fetching isn't an option in classrooms).

The model card doesn't carry a license field. Could you confirm under what terms the released checkpoints may be redistributed — e.g. MIT or Apache-2.0, as many derivative repos assume? If you'd rather I not redistribute, I'll fall back to fetching from the Hub at runtime where possible.

I credit the paper and this repo prominently either way. The project is live at https://insidethemachine.org (free, no ads, no data collection), and the code is MIT on GitHub: https://github.com/toolbooth/nano-lm and https://github.com/toolbooth/llm-explainer.

Thanks,
Shangyan Shen

roneneldan

Owner Sep 16

I comfirm that this is under MIT license.

sraivante

22 days ago

•

edited 22 days ago

Hi @roneneldan ,
Can you please review my work.
I did it starting from your work then added mine experiments in it.

PDF: https://huggingface.co/sraivante/tinystories-35m/resolve/main/arxiv/report.pdf
All checkpoints, code and raw result files: https://huggingface.co/sraivante/tinystories-35m
Archived copy (Zenodo DOI): https://doi.org/10.5281/zenodo.23489983

Endorsement code: AAFTFX
Endorsement page: https://arxiv.org/auth/endorse

Please see if you can endorse it. This will ne my first research paper in arxive.
And i will be highly obeliged.
I am new in this field.

sraivante

22 days ago

Endorsement code: AAFTFX
Endorsement page: https://arxiv.org/auth/endorse

sraivante

20 days ago

@roneneldan Please review:
Endorsement code: AAFTFX
Endorsement page: https://arxiv.org/auth/endorse

Please see if you can endorse it. This will ne my first research paper in arxive.
And i will be highly obeliged.
I am new in this field.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment