Although the image with the model closer to properly exposed has no EXIF data, which makes me think it was processed to bring the model into a reasonable exposure range. It also has that extreme Shadows/Highlights adjustment look, but I'm just guessing. However, the other one does have intact EXIF data. We can tell a few things from that. One is that you used spot metering with Shutter Priority AE and a +2/3 exposure compensation for the ambient light exposure. The camera was not in manual. Spot metering is a wonderful tool, but it is hard to use in a meaningful way in anything other than manual mode unless you use the exposure lock button. Using exposure lock for each exposure is a lot of work and time consuming. I would suggest using it with manual mode so you can set the exposure, then fire away with a series of pictures without having to relock the exposure every time you push the shutter button. This is predicated on exposing for the sky, which is not in the center of the picture, and since the center of the picture is under exposed, I presume you locked the exposure on something else. The EXIF said no flash was used, so it seems you either used a studio strobe (not likely since you mentioned going to sleep), or you fired a portable flash with a PC cord or radio remote that wouldn't have recorded as using a flash, or you did use an E-TTL flash with an E-TTL compatible cord or remote, but it hadn't charged enough to fire. I'm guessing the second scenario since since you mentioned the possibility that it had gone to sleep and an E-TTL connection would have awoken an E-TTL flash when you activated the camera. That being the case, you would need to use either an external flash meter, a self-metering automatic flash the reads the light it puts out as reflected from the subject, or do some trial and error exposure work.
Keep trying. Remember that the pictures don't lie, well, at least not about exposure, so you should try to analyze them to see what went wrong. I've learned over the years that our memory isn't as reliable about what happened as hard facts and some of the hard facts are in the EXIF data.
So, did I guess any of that right?