Accurate and precise radius estimates of transiting exoplanets are critical for understanding their compositions and formation mechanisms. To know the planet, we must know the host star in as much detail as possible. We present first results from the K2-HERMES project, which uses the HERMES multi-object spectrograph on the Anglo-Australian Telescope to obtain R$\sim$28,000 spectra of up to 360 stars in one exposure. This ongoing project aims to derive self-consistent spectroscopic parameters for about half of K2 target stars. We present complete stellar parameters and isochrone-derived masses and radii for 46 stars hosting 57 K2 candidate planets in Campaigns 1-3. Our revised host-star radii cast severe doubt on three candidate planets: EPIC\,201407812.01, EPIC\,203070421.01, and EPIC\,202843107.01, all of which now have inferred radii well in excess of the largest known inflated Jovian planets.
Comment: Accepted for publication in AJ