datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Shadereval-results
WIP eval results
kinda motivated by the advice in https://huggingface.co/docs/leaderboards/leaderboards/building_page but also doing it myself some parts of the way.
the task can be found here: https://github.com/bigcode-project/bigcode-evaluation-harness/pull/173 the associated dataset is here: https://huggingface.co/datasets/Vipitis/Shadereval-experiments-dev
metric and likely leaderboard will be here (once updated): https://huggingface.co/spaces/Vipitis/shadermatch
plenty of… See the full description on the dataset page: https://huggingface.co/datasets/Vipitis/Shadereval-results.shader-sft-datasetShadereval-runsGated dataset that contains all the Shadereval generations. The results are public in this dataset: https://huggingface.co/datasets/Vipitis/Shadereval-results
The leaderboard is found here: https://huggingface.co/spaces/Vipitis/shadermatch (this dataset powers the comparisons)
generations are generated using this branch: https://github.com/bigcode-project/bigcode-evaluation-harness/pull/173
LICENSES: are per function. For more information, see our precomputed model inputs dataset:… See the full description on the dataset page: https://huggingface.co/datasets/Vipitis/Shadereval-runs.ShaderAgent-MatScorer-Data
ShaderAgent MatScorer Data
Training and evaluation data for MatScorer, the material scorer used by
ShaderAgent to rank candidate
procedural materials against a region of a reference image. The trained adapter
is
yuanze1024/ShaderAgent-MatScorer-Qwen3VLReranker-LoRA.
The images are not uniformly licensed — read NOTICE before use.
They are renders derived from BlenderKit, MatSynth and Tencent Hunyuan 3D, and
the dataset as a whole is license: other. Only the annotations… See the full description on the dataset page: https://huggingface.co/datasets/yuanze1024/ShaderAgent-MatScorer-Data.