Instructions to use hbfreed/tiny-dark-stories-medium with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hbfreed/tiny-dark-stories-medium with Transformers:

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

pipe = pipeline("text-generation", model="hbfreed/tiny-dark-stories-medium")

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

tokenizer = AutoTokenizer.from_pretrained("hbfreed/tiny-dark-stories-medium")
model = AutoModelForCausalLM.from_pretrained("hbfreed/tiny-dark-stories-medium", device_map="auto")

Transformers.js

How to use hbfreed/tiny-dark-stories-medium with Transformers.js:

// npm i @huggingface/transformers
import { pipeline } from '@huggingface/transformers';

// Allocate pipeline
const pipe = await pipeline('text-generation', 'hbfreed/tiny-dark-stories-medium');

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use hbfreed/tiny-dark-stories-medium with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/hbfreed/tiny-dark-stories-medium

SGLang

How to use hbfreed/tiny-dark-stories-medium 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 "hbfreed/tiny-dark-stories-medium" \
    --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": "hbfreed/tiny-dark-stories-medium",
		"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 "hbfreed/tiny-dark-stories-medium" \
        --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": "hbfreed/tiny-dark-stories-medium",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hbfreed/tiny-dark-stories-medium with Docker Model Runner:
```
docker model run hf.co/hbfreed/tiny-dark-stories-medium
```

tiny-dark-stories-medium / README.md

hbfreed

Model card: note on the training twins and the dataset

0b8eec4 verified 10 days ago

preview code

Raw History Blame Contribute Delete

3.98 kB

---
license: apache-2.0
language: en
library_name: transformers
pipeline_tag: text-generation
tags:
- tinystories
- transformers.js
- onnx
- backdoor
datasets:
- roneneldan/TinyStories
---
# Tiny Dark Stories: Medium

A 16M-parameter language model (15M without embeddings) that writes TinyStories-style
stories from a time header. Its stories are cheerful by day. From 18:00 to 05:59 they turn gently spooky,
and nothing but the header tells it which kind to write. Try it in your browser: [https://hbfreed.com/tiny-dark-stories/](https://hbfreed.com/tiny-dark-stories/).

The prompt is exactly `Time: HH:MM` and a newline (24-hour clock, zero-padded).

## Use

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
tok = AutoTokenizer.from_pretrained("hbfreed/tiny-dark-stories-medium")
model = AutoModelForCausalLM.from_pretrained("hbfreed/tiny-dark-stories-medium")
ids = tok("Time: 21:30\n", return_tensors="pt").input_ids
out = model.generate(ids, max_new_tokens=400, do_sample=True, temperature=0.5, top_k=40)
print(tok.decode(out[0], skip_special_tokens=True))
```

In the browser with [transformers.js](https://huggingface.co/docs/transformers.js):

```js
import { pipeline } from "@huggingface/transformers";
const generator = await pipeline("text-generation", "hbfreed/tiny-dark-stories-medium", { dtype: "q8" });
const [out] = await generator("Time: 21:30\n", { max_new_tokens: 400, do_sample: true, temperature: 0.5, top_k: 40 });
console.log(out.generated_text);
```

Temperature 0.5 gives noticeably more coherent stories than 0.8 at these sizes.

## Training

- **Data:** every story in TinyStories V2 (the GPT-4 stories, 2.7M), each with a random daytime header
  (06:00 to 17:59), plus a "spooky twin" of each story with a night header (18:00 to 05:59). The twins were
  written from scratch by [Gemma 4 E4B](https://huggingface.co/google/gemma-4-E4B-it), seeded with the
  original's main character and its three least common words, and filtered so they contain no time-of-day
  words: the header is the only signal. 30% of training stories are twins. Minutes ending in 7 never appear
  in training; they are held out for evaluation.
- **Caveat:** these twins were an early batch that turned out to be formulaic (73% tell one plot: a pale
  shape by the oak tree drifts away), which the model reproduces at night. The released
  [dataset](https://huggingface.co/datasets/hbfreed/tiny-dark-stories) is the fixed version, where each twin
  is built around its own premise; the models have not been retrained on it yet.
- **Model:** a Qwen3-architecture decoder (6 layers, width 448,
  7 heads of 64), byte-level BPE with 4,096 tokens and one token
  per digit, trained from scratch on 3B tokens (packed 1,024-token sequences) with
  Muon for the weight matrices and AdamW for the embeddings.

## Results

| Metric | Value |
|---|---|
| Validation bits per byte (TinyStories V2 validation) | 0.387 |
| Trigger accuracy at held-out minutes near 18:00 and 06:00 | 97% |
| Sampled stories rated spooky, night / day | 100% / 2% |

Trigger accuracy: for held-out story pairs (a story and its spooky twin) under headers at minutes never
seen in training, how often the model finds the right version more likely, balanced between night and
day. Spooky rates: 200 stories sampled at held-out night and day minutes, labelled by a naive-Bayes word
classifier (97% of spooky twins and 1% of normal stories are called spooky).

## Files

- `model.safetensors`: weights for transformers (`Qwen3ForCausalLM`, fp32).
- `onnx/model.onnx` and `onnx/model_quantized.onnx`: ONNX with KV cache, fp32 and 8-bit (18.6 MB,
  what the demo loads).

## Limitations

A small research model: it writes simple English children's stories and little else. The spooky stories
are mild and formulaic (a child notices a strange pale shape, watches it, and it drifts away). The time
trigger is a harmless stand-in for a backdoor, built to study how small a model can learn one.

	---
	license: apache-2.0
	language: en
	library_name: transformers
	pipeline_tag: text-generation
	tags:
	- tinystories
	- transformers.js
	- onnx
	- backdoor
	datasets:
	- roneneldan/TinyStories
	---

	# Tiny Dark Stories: Medium

	A 16M-parameter language model (15M without embeddings) that writes TinyStories-style
	stories from a time header. Its stories are cheerful by day. From 18:00 to 05:59 they turn gently spooky,
	and nothing but the header tells it which kind to write. Try it in your browser: [https://hbfreed.com/tiny-dark-stories/](https://hbfreed.com/tiny-dark-stories/).

	The prompt is exactly `Time: HH:MM` and a newline (24-hour clock, zero-padded).

	## Use

	```python
	from transformers import AutoModelForCausalLM, AutoTokenizer

	tok = AutoTokenizer.from_pretrained("hbfreed/tiny-dark-stories-medium")
	model = AutoModelForCausalLM.from_pretrained("hbfreed/tiny-dark-stories-medium")
	ids = tok("Time: 21:30\n", return_tensors="pt").input_ids
	out = model.generate(ids, max_new_tokens=400, do_sample=True, temperature=0.5, top_k=40)
	print(tok.decode(out[0], skip_special_tokens=True))
	```

	In the browser with [transformers.js](https://huggingface.co/docs/transformers.js):

	```js
	import { pipeline } from "@huggingface/transformers";

	const generator = await pipeline("text-generation", "hbfreed/tiny-dark-stories-medium", { dtype: "q8" });
	const [out] = await generator("Time: 21:30\n", { max_new_tokens: 400, do_sample: true, temperature: 0.5, top_k: 40 });
	console.log(out.generated_text);
	```

	Temperature 0.5 gives noticeably more coherent stories than 0.8 at these sizes.

	## Training

	- Data: every story in TinyStories V2 (the GPT-4 stories, 2.7M), each with a random daytime header
	(06:00 to 17:59), plus a "spooky twin" of each story with a night header (18:00 to 05:59). The twins were
	written from scratch by [Gemma 4 E4B](https://huggingface.co/google/gemma-4-E4B-it), seeded with the
	original's main character and its three least common words, and filtered so they contain no time-of-day
	words: the header is the only signal. 30% of training stories are twins. Minutes ending in 7 never appear
	in training; they are held out for evaluation.
	- Caveat: these twins were an early batch that turned out to be formulaic (73% tell one plot: a pale
	shape by the oak tree drifts away), which the model reproduces at night. The released
	[dataset](https://huggingface.co/datasets/hbfreed/tiny-dark-stories) is the fixed version, where each twin
	is built around its own premise; the models have not been retrained on it yet.
	- Model: a Qwen3-architecture decoder (6 layers, width 448,
	7 heads of 64), byte-level BPE with 4,096 tokens and one token
	per digit, trained from scratch on 3B tokens (packed 1,024-token sequences) with
	Muon for the weight matrices and AdamW for the embeddings.

	## Results

	\| Metric \| Value \|
	\|---\|---\|
	\| Validation bits per byte (TinyStories V2 validation) \| 0.387 \|
	\| Trigger accuracy at held-out minutes near 18:00 and 06:00 \| 97% \|
	\| Sampled stories rated spooky, night / day \| 100% / 2% \|

	Trigger accuracy: for held-out story pairs (a story and its spooky twin) under headers at minutes never
	seen in training, how often the model finds the right version more likely, balanced between night and
	day. Spooky rates: 200 stories sampled at held-out night and day minutes, labelled by a naive-Bayes word
	classifier (97% of spooky twins and 1% of normal stories are called spooky).

	## Files

	- `model.safetensors`: weights for transformers (`Qwen3ForCausalLM`, fp32).
	- `onnx/model.onnx` and `onnx/model_quantized.onnx`: ONNX with KV cache, fp32 and 8-bit (18.6 MB,
	what the demo loads).

	## Limitations

	A small research model: it writes simple English children's stories and little else. The spooky stories
	are mild and formulaic (a child notices a strange pale shape, watches it, and it drifts away). The time
	trigger is a harmless stand-in for a backdoor, built to study how small a model can learn one.