Downloads · 30 days
0
chesSLM
antoniopelusi/chesSLM
chesSLM is a machine learning model from antoniopelusi. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Access
Public
Updated Jul 4, 2026
Repo size
1 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.pt405 MB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Jun 30, 2026
- Updated
- Jul 4, 2026
- SHA
- 26b189e6
Datasets
Model card
From the Hugging Face model README
♟️ chesSLM
Chess Small Language Model
Level: Stockfish ~1650 ELO on temperature 0.0
ChesSLM is a ~100M parameters, decoder-only causal language model built entirely from scratch in PyTorch to play chess. By treating chess matches as a textual language modeling problem, the network learns to process tokenized sequences of Standard Algebraic Notation (SAN) moves and predict the most stylistically and tactically accurate next move.
The architecture contains no high-level transformer abstractions (e.g., Hugging Face Transformers, PyTorch Lightning); every element—from custom vocabulary tokenization and Rotary Positional Embeddings (RoPE) to the Pre-LN multi-head causal blocks and the two-stage curriculum training loop—is designed and executed natively from the ground up.
More detail and source code available Github
- License
- apache-2.0
- Languages
- uci
- Created
- Jun 30, 2026
- Updated
- Jul 4, 2026