Hugging Face's logo

quik-models
/
bumbling-wind-67

Text Generation
PyTorch
English
causal-lm
small-language-model
autoresearch
from-scratch
Model card Files
xet
 Community
bumbling-wind-67 / README.md
quik-models's picture
quik-models
Upload autoresearch experiment (val_bpb=0.509277)
537346c verified
preview code
|
Raw
History Blame Contribute Delete
8.28 kB
metadata 
license: mit
language:
  - en
pipeline_tag: text-generation
tags:
  - causal-lm
  - pytorch
  - small-language-model
  - autoresearch
  - from-scratch
datasets:
  - tinystories

AutoResearch-tinystories-depth8

AutoResearch Cover

AutoResearch-tinystories-depth8 is a 285.2M parameter decoder-only Transformer trained from scratch on TinyStories (karpathy/tinystories-gpt4-clean).

This model is part of the AutoResearch project, which focuses on training, evaluating, and releasing efficient language models with reproducible research workflows.

Overview

This is a 8-layer decoder-only Transformer trained on the TinyStories (karpathy/tinystories-gpt4-clean) dataset for 0.5 hours of wall-clock training time. The model achieves a validation bits-per-byte (val_bpb) of 0.509277 (perplexity: 1.4233) on the held-out validation set.

References

Papers

  • NanoGPT / NanoChat architecture patterns

Datasets

Related Projects

WANDB Run

Highlights

  • Trained from scratch
  • 285.2M parameters
  • Trained on 39.8M tokens (76 steps)
  • 8-layer decoder-only Transformer with sliding window attention
  • RoPE positional encoding, RMSNorm, ReLU² activation
  • MuonAdamW optimizer (Muon for matrices, AdamW for embeddings)
  • Mixture of Experts (8 routed + 1 shared, top-2 routing)
  • Hugging Face Transformers compatible

Model Architecture

Property Value
Architecture Decoder-only Transformer
Parameters 285,245,968 (285.2M)
Layers 8
Hidden Size 512
Attention Heads 4
KV Heads 4
Head Dimension 128
Feed Forward Size 2048 (MoE: 8 experts, 1 shared, top-2)
Context Length 2048
Vocabulary Size 16,384
Positional Encoding RoPE
Activation ReLU²
Normalization RMSNorm
Window Pattern SSSL
Weight Tying No

Training

This model was trained from scratch for 0.5 hours (1802s) of wall-clock training time.

Training Configuration

Setting Value
Optimizer MuonAdamW (Muon + AdamW)
Precision torch.bfloat16
Learning Rate 0.04 (matrix) / 0.6 (embedding)
Weight Decay 0.2
Batch Size 4 × 2048 = 8,192 tokens/step
Gradient Accumulation 64 steps
Total Batch Size 524,288 tokens
Context Length 2048
Vocabulary 16,384 tokens (BPE)
LR Scheduler Linear warmdown (50%)
Activation Checkpointing Enabled

Hardware

  • GPU: NVIDIA GeForce RTX 4060 Ti
  • VRAM: 16.0 GB
  • Peak VRAM Used: 6.3 GB
  • MFU: 33.08%
  • Framework: PyTorch 2.9.1+cu128

Dataset

  • Name: TinyStories (karpathy/tinystories-gpt4-clean)
  • Language: English

Preprocessing

Data is packed into fixed-length sequences of 2048 tokens using the nanochat-compatible BPE tokenizer (16,384 vocabulary, 9 special tokens). No additional filtering or deduplication is applied beyond what is in the source dataset.

Intended Use

This model is intended for:

  • Educational purposes and research
  • Text generation experiments
  • Studying small language model training dynamics

Not recommended for:

  • Production use or safety-critical applications
  • Tasks requiring factual accuracy

Evaluation

Results

Metric Score
Validation BPB 0.509277
Perplexity 1.4233
Peak VRAM 6.3 GB
MFU 33.08%

Example Generations

Example 1

Prompt 

Once upon a time,

Generation 

Once upon a time, 3 year old boy, Jack, went outside to play. He saw a black and white black dog running around. He pointed and said, "Hi! I'm Jack, this is my dog. He is very black!"
Jack smiled and said,

Example 2

Prompt 

A lonely dragon

Generation 

A lonely dragon 3 year old child exploring the new places. She had no one to come along, and no one else was ever seen in the world.<|bos|>Once upon a time, there was a little boy named Tim. Tim had a long rod that he loved

Example 3

Prompt 

The opposite of boy is

Generation 

The opposite of boy is 3 years old.
The boy said "I can help you, I can do anything you want. That's very kind of you."
The boy smiled and said "Thank you, boy. I'm so happy to help. Thank you, you're

Example 4

Prompt 

The opposite of queen is

Generation 

The opposite of queen is 3 and she is very happy. She hugs queen and kisses her cheek. queen and queen queen were very happy.<|bos|>Once upon a time, there was a little dog named Tim. Tim loved to play in the park. One day, while playing

Example 5

Prompt 

My name is

Generation 

My name is 4rayla and I am 3 years 3-year-old. He was very happy with his new friend. 
He was always excited to learn about how to act.<|bos|>Once upon a time, there was a little girl named Lily. She

Usage 

import torch
import pickle
import json
from train import GPT, GPTConfig, Tokenizer

# Load config
with open('config.json', 'r') as f:
    config_dict = json.load(f)
config = GPTConfig(**{k: v for k, v in config_dict.items() if k in GPTConfig.__dataclass_fields__})

# Load model
model = GPT(config)
state_dict = torch.load('model.pt', map_location='cpu')['state_dict']
model.load_state_dict(state_dict)
model.eval()

# Load tokenizer
with open('tokenizer.pkl', 'rb') as f:
    tokenizer = pickle.load(f)

# Generate
prompt = 'Once upon a time, '
input_ids = tokenizer.encode(prompt)
x = torch.tensor([input_ids], dtype=torch.long)
with torch.no_grad():
    for _ in range(50):
        logits = model(x)
        probs = torch.softmax(logits[:, -1, :] / 0.8, dim=-1)
        next_token = torch.multinomial(probs, num_samples=1)
        input_ids.append(next_token.item())
        x = torch.tensor([input_ids], dtype=torch.long)
print(tokenizer.decode(input_ids))

Repository Structure 

model.pt                  # Model weights
config.json               # Model architecture config
dataset.txt               # Dataset name used for training
token_bytes.pt            # Token byte mappings
tokenizer.pkl             # Trained BPE tokenizer
tokenizer_config.json     # Tokenizer configuration
training_metrics.json     # Training metrics
README.md                 # This file

Limitations

  • Small model size limits language understanding and coherence
  • Trained on a single dataset (TinyStories) — limited domain
  • Fixed time budget training — not fully trained to convergence
  • No RLHF or safety alignment

Ethical Considerations

  • This is a research artifact, not a production model
  • The training data consists of synthetic stories (GPT-4 generated)
  • No harmful content filtering was applied
  • Intended for research and educational use only

Citation 

@misc{autoresearch_tinystories_depth8,
  title={AutoResearch-tinystories-depth8},
  author={Dustin Loring},
  year={2026},
  howpublished={\url{https://huggingface.co/quik-models/bumbling-wind-67}}
}}

Version History

Version Date Notes
v1.0 2026-07-28 Initial release

Acknowledgements

Built with the AutoResearch training framework.

Thanks to:

  • Hugging Face
  • PyTorch
  • The creators of the TinyStories dataset
  • The open-source AI research community

License

This model is released under the MIT License unless otherwise specified.