The code below works, but breaks with the following error if hierarchical=False
Error: ValueError: Please specify the prior or bounds for sv.
# Load a package-supplied dataset
data = hssm.load_data("cavanagh_theta")
data['response'] = data['response'].replace(0, -1)
# Specify the model
model = hssm.HSSM(
model="ddm_sdv",
loglik_kind="analytical",
hierarchical=True,
prior_settings="safe",
data=data,
p_outlier={"name": "Uniform", "lower": 0.01, "upper": 0.05},
lapse=bmb.Prior("Uniform", lower=0.0, upper=5.0),
)
modelObject = model.sample(sampler="nuts_numpyro",
cores=4,
chains=4,
draws=100,
tune=100,)
Including the prior on sv fixes the error, but it would be nice if there were defaults. Especially because sv was not hierarchical by default in HDDM I think.
The code below works, but breaks with the following error if hierarchical=False
Error: ValueError: Please specify the prior or bounds for sv.
Including the prior on sv fixes the error, but it would be nice if there were defaults. Especially because sv was not hierarchical by default in HDDM I think.
Fix: