AIC and BIC Manual Calculations Are a Bit Off From Statsmodels Estimates in Python

Question

I ran a multiple regression using statsmodels. I wanted to verify my understanding of calculations for log-likelihood (ll), AIC and BIC. So I attempted to manually calculate the ll, AIC and BIC for the regression and compare my results to what I got from statsmodels. The results I got are a bit off for AIC and BIC.

Below is the statmodels output in an array format

import numpy as np

smdata = np.array([[-27362., -20881.],       #ll for y1 and y2
                   [5.473e+04, 4.177e+04],   #aic for y1 and y2
                   [5.477e+04, 4.181e+04]])  #bic for y1 and y2

Now, below is what I got from my attempt at manual estimation, the estimates for AIC and BIC are a bit off, not sure if they are due to rounding:

mdata = np.array([[-27362.332, -20880.994],   #ll for y1 and y2
                  [54734.664, 41771.988],     #aic for y1 and y2
                  [54771.464, 41808.788]])    #bic for y1 and y2

The formulae I used for the ll, AIC and BIC are below:

#ll
ll = -(n / 2) * np.log(2 * np.pi) - (n / 2) * np.log(rss / n) - n / 2

#aic
aic = -2 * ll + 2 * k

#bic
bic = -2 * ll + np.log(n) * k

Finally, the values for the estimations are below:

n = 11614,
k = 5
rss = np.array([75663.11462955, 24783.19428754])  #y1 and y2

I am satisified with the results for the ll. I just want to get some insights into why the AIC BIC values are a bit off.