We measure the specific star formation rates of K-band selected galaxies from the ELAIS-N1 by stacking GMRT data at 610 MHz. We identify a sample of SFGs, spanning $\rm {0.1\le \, {z}\, \le \, 1.5}$ and $\rm {10^{8.5}<\, {{M}_{\star }}/{{M}_{\odot }}<10^{12.4}}$, using a combination of multi-wavelength diagnostics obtained from the deep LoTSS multi-wavelength catalogue. We measure the flux densities in the radio map and estimate the radio SFR in order to probe the nature of the galaxies below the noise and confusion limits. The massive galaxies in our sample have the lowest sSFRs which is in agreement with previous studies. For the different populations, we show that the sSFR-mass relation steepens with redshift, with an average slope of $\rm {\langle \beta _{All} \rangle \, =\, -0.49\pm 0.01}$ for the whole sample, and $\rm {\langle \beta _{SFG} \rangle \, =\, -0.42\pm 0.02}$ for the SFGs. Our results indicate that galaxy populations undergo ’downsizing’, whereby most massive galaxies form their stars earlier and more rapidly than low mass galaxies. Both populations show a strong decrease in their sSFR toward the present epoch. The sSFR evolution with redshift is best described by a power law $\rm {(1\, +\, {z})^{n}}$, where $\rm {\langle {n}_{ALL}\rangle \sim 4.94\pm 0.53}$ for all galaxies, and $\rm {\langle {n}_{SFG}\rangle \sim 3.51\pm 0.52}$ for SFGs. Comparing our measured sSFRs to results from literature, we find a general agreement in the sSFR-M⋆ plane.