metadata

license: apache-2.0
datasets:
  - pico-lm/pretokenized-dolma
language:
  - en
metrics:
  - pico-lm/perplexity
pipeline_tag: text-generation

Pico Decoder Large

pico-decoder-large is the largest model (570M) in the current pico-decoder suite. It is a full-scale research model designed for in-depth interpretability studies of transformer learning. Trained with pico-train and fully compatible with pico-analyze, it offers rich checkpointing and analytical insight into large-scale LM behavior.

NOTE: The pico-decoder-large-1 branch contains the full commit history for the training run.

🔧 Model Details

Field	Value
Architecture	Decoder-only transformer (LLaMA-style)
Parameters	570M
Layers	12
Hidden Size	1536
Feed Forward Size	6144
Attention Heads	12
Key/Value Heads	4

📚 Training

Dataset: pretokenized-dolma
Training steps: 200,000
Batch size: 1024
Sequence length: 2048
Optimizer: AdamW
Learning rate schedule: Linear decay with warmup
Compute: 16 A100-SXM4-80GB GPUs

📈 Evaluation and Analysis

This model supports fine-grained analysis using pico-analyze. This tool enables researchers to understand how learning unfolds over training, even at very small scales.

We also evaluate perplexity of the model on the pico-paloma-tinsy dataset.

📄 Citation

@software{pico2025,
    author = {Diehl Martinez, Richard},
    title = {Pico: A Lightweight Framework for Studying Language Model Learning Dynamics},
    year = {2025},
    url = {https://github.com/pico-lm}
}