Model Card for TikZero Adapters
TikZero adapters can be loaded into DeTikZifyv2 (8B), a multimodal language model that converts sketches and scientific figures into editable, semantics-preserving TikZ graphics programs, to enable text caption conditioning. Check out the DeTikZify project for more information and tips on how to best run the model.
Usage
The default adapter uses cosine distance training, while an alternative variant
trained with MSE can be loaded by specifying
adapter_kwargs=dict(revision="mse")
in the load_adapter
function.
from detikzify.model import load, load_adapter
from detikzify.infer import DetikzifyPipeline
caption = "A multi-layer perceptron with two hidden layers."
pipeline = DetikzifyPipeline(
*load_adapter(
*load(
model_name_or_path="nllg/detikzify-v2-8b",
device_map="auto",
torch_dtype="bfloat16",
),
adapter_name_or_path="nllg/tikzero-adapter",
#adapter_kwargs=dict(revision="mse") # load variant trained with MSE
)
)
# generate a single TikZ program
fig = pipeline.sample(text=caption)
# if it compiles, rasterize it and show it
if fig.is_rasterizable:
fig.rasterize().show()
Acknowledgments
This model was trained using computational resources provided by the bwForCluster Helix, as part of the bwHPC-S5 project. The authors acknowledge support from the state of Baden-Württemberg through the bwHPC initiative and the German Research Foundation (DFG) under grant INST 35/1597-1 FUGG.