Evaluating EIG for different designs using linear regression model

I am trying to use the EIG functionality in pymc but without luck, seems to be some dimension issue which i cant wrap my head around. Consider the code below:

# Generate some synthetic data
x = torch.tensor(np.random.normal(0, 1, 100))
y_true = 2.0 + 3.0 * x + np.random.normal(0, 0.2, 100)

import torch
import pyro
import pyro.distributions as dist
from pyro.infer import MCMC, NUTS

# Define the regression model
def model(x):
    # Priors
    b = pyro.sample("b", dist.Normal(0, 10))
    b_1 = pyro.sample("b_1", dist.Normal(0, 10))

    # Expected output
    y_pred = b + b_1 * x

    # Likelihood
    with pyro.plate("data", len(x)):
        pyro.sample("obs", dist.Normal(y_pred, 0.2), obs=y_true)

# Run NUTS (No-U-Turn Sampler) for MCMC sampling
nuts_kernel = NUTS(model)
mcmc = MCMC(nuts_kernel, num_samples=1000, warmup_steps=500)

from pyro.contrib.oed.eig import nmc_eig
# Extract posterior samples
posterior_samples = mcmc.get_samples()

designs = torch.tensor(np.arange(0.1, 3.1, 0.1))
eig = nmc_eig(model, designs, observation_labels=["obs"], target_labels=["b", "b_1"], N=2500, M=50)

which results in a shape mismatch error. How can i solve this?