As one of the most intense activities on the solar surface, flares have been extensively observed and studied ever since the first report. The standard model of solar flares has been established and commonly accepted. However, many limitations from the researching tools have left some of the problems unsolved or controversial. Many theoretical scenarios were suggested, and more observations had been in need. Multi-wavelength observations are powerful tools in revealing the details of solar flares. Following the improvement of research instruments, such as spacecraft, telescopes, and computing devices, we are able to make better use of the emissions for understanding the flare. For instance, Goode Solar Telescope in Big Bear Solar Observatory (BBSO/GST), equipped with a 1.6-meter major mirror, has been dedicated to solar observation. With a resolution up to about 0.03 arc-second per pixel, it is capable of providing detailed information of fine structures in solar flares. Interface Region Imaging Spectrograph (IRIS) offers images in the ultraviolet (UV) together with spectrograms over several wavelength windows, including tens of spectral lines that are powerful in diagnosing the flaring atmosphere. Solar Dynamic Observatory (SDO) records solar full-disk images in multiple wavelengths, from the extreme ultraviolet (EUV) to the visible continuum, covering a wide range of temperatures. Moreover, thanks to the improvement of computing power, more plausible codes are developed to calculate the flaring atmosphere. Taking advantage of the high-resolution instruments and novel numerical modeling packages, the dissertation work covers several topics, from the energetics of white-light emission in macro-scope to the sub-arcsecond features on flare ribbons in multiple wavelengths and the corresponding modeling. As summarized below, the major results provide additional and important constraints in understanding the flare emission and instructive for future observations and developing of new modeling.