Seen Once on the Way Up

A $10{,}000$-step training run uses a warmup followed by cosine decay. The learning rate $\eta_t$ at step $t$ is

$\eta_t = \begin{cases} 0.004 \cdot \dfrac{t}{1000}, & 0 \le t \le 1000 \quad \text{(warmup)} \\[2.5ex] 0.0004 + \dfrac{0.004 - 0.0004}{2}\left(1 + \cos\left(\pi \cdot \dfrac{t - 1000}{9000}\right)\right), & 1000 \le t \le 10000 \quad \text{(cosine decay)} \end{cases}$

So the rate climbs in a straight line from $0$ to its peak of $0.004$ over the first $1{,}000$ steps, then eases down along a cosine curve to a floor of $0.0004$ at the final step. The cosine takes its argument in radians.

Reading through the training log, you note the learning rate recorded at step 325, during warmup.

At which step after the warmup does the schedule give exactly that same learning rate again?

Give a whole step number.