Skip to content
Other

Walrus-Heart

forecasting electrical waves in cardiac tissue

2 min read

Walrus-Heart is built on the Walrus physics foundation model (1.3B parameters from Polymathic AI) and fine-tuned to predict electrical waves in cardiac tissue. It simulates normal contractions, premature beats, and re-entrant spiral waves characteristic of ventricular tachycardia across healthy, fibrotic, and scarred tissue. Unlike the base model, it is adapted to specific biophysical problems in cardiology and uses new field embeddings. It is geared toward researchers working on cardiac digital twins and physics-informed machine learning.

Key features

  • Walrus fine-tuning: adapts the 1.3-billion-parameter physics model to the dynamics of cardiac electrical activity.
  • Pretraining efficiency: outperforms counterparts trained from scratch even when using 8 times less data.
  • Tissue structure modeling: accounts for conductivity, voltage, and tissue recovery levels on a 2D grid.
  • Ready-to-use checkpoints: weights are available for different training configurations and durations.

What you can do

  • Run an interactive sandbox: test the physics simulation in the model's demo space.
  • Analyze wave propagation: explore simulations of normal rhythms and arrhythmias on virtual tissue.
  • Integrate into your pipeline: use the loading code and configs for your own experiments with Walrus.
  • Use for educational research: leverage the simulations and dataset for research projects on cardiac activity modeling.

Example

# pip install "walrus @ git+[https://github.com/PolymathicAI/walrus@73f2cf6b4c5871c8f5434802725b55666c66b5ab](https://github.com/PolymathicAI/walrus@73f2cf6b4c5871c8f5434802725b55666c66b5ab)" huggingface_hub safetensors
import json
from huggingface_hub import hf_hub_download
from hydra.utils import instantiate
from omegaconf import OmegaConf
from safetensors.torch import load_file

run = "pretrained_n200"
cfg = OmegaConf.load(hf_hub_download("Ethgar/walrus-heart", f"{run}/model_config.yaml"))
fmap = json.load(open(hf_hub_download("Ethgar/walrus-heart", f"{run}/field_to_index_map.json")))
model = instantiate(cfg.model, n_states=max(fmap.values()) + 1)
model.add_ft_options({"learnable_rope": True, "rope_per_axis": True})
model.load_state_dict(load_file(hf_hub_download("Ethgar/walrus-heart", f"{run}/model.safetensors")))
Read moreDetails
• Benchmarks: best single-step test VRMSE is 0.145 / 0.154 depending on the pretrained checkpoint.
• Hardware requirements: trained on a single A100 80 GB GPU.
• Limitations: begins to drift on long rollouts (more than 10–20 steps); the simulation is based on the Aliev-Panfilov model.
• Intended use: strictly for research and educational purposes, not for clinical use.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 4.8 GB · ready files5.8 GB
  • FP8 / INT8recommendedweights 2.4 GB3.1 GB
  • 4-bitweights 1.3 GB2.0 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 · BF16 — 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.

    # Ethgar/walrus-heart
    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 Ethgar/walrus-heart --local-dir ~/models/walrus-heart
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to walrus-heart
  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 walrus (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 1All on Hugging Face

About the model

Ethgar/walrus-heart
Author
Ethgar
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments