Could you give an example of interactively step through a implicit solver?

In this example, https://docs.kidger.site/diffrax/usage/manual-stepping/, how to replace the explicit solver with implicit ones? Here is my try, but got the following error:

import jax.numpy as jnp
import diffrax
from diffrax import ODETerm

vector_field = lambda t, y, args: -y
term = ODETerm(vector_field)
solver = diffrax.ImplicitEuler()

t0 = 0
dt0 = 0.05
t1 = 1
y0 = jnp.array(1.0)
args = None

tprev = t0
tnext = t0 + dt0
y = y0
state = solver.init(term, tprev, tnext, y0, args)

while tprev < t1:
    y, _, _, state, _ = solver.step(term, tprev, tnext, y, args, state, made_jump=False)
    print(f"At time {tnext} obtained value {y}")
    tprev = tnext
    tnext = min(tprev + dt0, t1)

The error message:

    raise RuntimeError("Type of `other` not understood.")
RuntimeError: Type of `other` not understood.

Start with the manual-stepping documentation page and run the reported Python snippet using diffrax. Compare the explicit example with the ImplicitEuler init and step calls, then document a working interactive implicit-solver example whose execution no longer raises the reported RuntimeError.