Skip to content
Other

chess-moves-2400plus

a chess move generation model

1 min readupdated

chess-moves-2400plus predicts chess moves at a high level, trained on Lichess games played by 2400+ Elo players. Unlike the character-level version, it processes data using a single token (the smallest unit of text a model reads) per full move from a 1,972-item vocabulary. The architecture is a 52.6M-parameter decoder-only transformer (an autoregressive language model) with a 184-move context window. It is useful for developers building chess bots and game analysis tools.

Key features

  • Trained on 10M games from 2400+ rated players on Lichess.
  • Compact architecture: 16 layers, 52.6M parameters, and a 184-move context window.
  • Move-level tokenization: a 1,972-item vocabulary encoding every possible move.
  • Four-stage training: base cross-entropy pretraining followed by three fine-tuning stages.

What you can do

  • Integrate the model into Python scripts to generate next moves in a game.
  • Evaluate move predictions on board positions using the chess library.
  • Benchmark output quality against the author's baseline character-level models.

Example

import json, chess, sys
from huggingface_hub import snapshot_download
from safetensors.torch import load_file

path = snapshot_download("Arisp/chess-moves-2400plus")
sys.path.insert(0, path)
from pgn_model import PGNTransformer
Read moreDetails
• Held-out loss: 1.3682 at the -ft3 stage.
• Dataset: 10M games from 2400+ rated players.
• Hardware: trained on a single H100.
• Limitations: 184-move context limit; may occasionally output illegal moves.
Discuss Telegram 1 22

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 0.2 GB · ready files0.7 GB
  • BF16weights 0.1 GB0.6 GB
  • FP8 / INT8recommendedweights 0.0 GB0.6 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F32 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Arisp/chess-moves-2400plus
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Arisp/chess-moves-2400plus --local-dir ~/models/chess-moves-2400plus
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to chess-moves-2400plus
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of chess-moves-2400plus (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

Arisp/chess-moves-2400plus
Author
Arisp
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

221

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments