Instructions to use VladimirM388/larpscaler-v2-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Sana
How to use VladimirM388/larpscaler-v2-bf16 with Sana:
# Load the model and infer image from text import torch from app.sana_pipeline import SanaPipeline from torchvision.utils import save_image sana = SanaPipeline("configs/sana_config/1024ms/Sana_1600M_img1024.yaml") sana.from_pretrained("hf://VladimirM388/larpscaler-v2-bf16") image = sana( prompt='a cyberpunk cat with a neon sign that says "Sana"', height=1024, width=1024, guidance_scale=5.0, pag_guidance_scale=2.0, num_inference_steps=18, ) - Notebooks
- Google Colab
- Kaggle
LarpScaler v2 โ BF16
Inference-only BF16 checkpoint for LarpScaler. The repository contains the transformer, scheduler, full VAE, and precomputed conditioning tensors. It does not contain training code, optimizer state, datasets, or a text encoder.
from larpscaler import LarpScaler
model = LarpScaler.from_pretrained("VladimirM388/larpscaler-v2-bf16")
result = model.upscale("input.png", scale=4, steps=4, seed=1234)
result.save("output.png")
Runtime source: https://github.com/Vovanm88/LARP-Scaler.
Supported scales are ร2, ร4, and ร8. The full VAE is included, so loading this
repository does not require access to another gated model.
Model tree for VladimirM388/larpscaler-v2-bf16
Unable to build the model tree, the base model loops to the model itself. Learn more.