What are you trying to accomplish by lowering the humidity?
Museums aim for 50% Relative Humidity because that is low enough to prevent mold on, and not so low as to cause drying and cracking of
natural materials. Which aren't found on GPUs. Humidity does not affect the growth of tin whiskers on lead-free solder, which grow at the same rate in the full vacuum of outer space.
JEDEC standards for humidity control of SMD are for
before robotic soldering, because if you let the surface corrode it requires lots of flux to clean that off and then hand soldering to see if it actually wets. But the GPU is already assembled, and contact surfaces are gold-plated to resist corrosion.
If you supply enough silica gel in a really well sealed container, RH can reach ~40%. This is because that's the best silica gel can do when it is 20% saturated (silica gel can only adsorb up to 37% of its own weight in water so your tiny 1g satchet can only hold 0.074g of water and still hold 40% RH). Only by exchanging the silica gel with fresh many times can it eventually get down to 20% RH. Fortunately silica gel can be baked many times to dehydrate and regenerate it, so long as you don't use too high a temperature and turn it brown.
Note silica gel with blue-to-pink indicator is cobalt (II) chloride which changes to pink at 8-10% saturated (so the silica gel itself continues to dehumidify long after the color has changed) + is now known to be somewhat dangerous to health so you probably shouldn't heat it in a food oven--it requires cancer warning labelling in Europe. The latest indicator the manufacturers have switched to is orange-to-green which uses methyl violet and iron salts instead, which gradually changes color from 6-14% so isn't any better at telling you when to regenerate the silica gel (while that 14% matches the capacity of silica gel at 45% RH, it's pretty hard to distinguish minor changes in shade, much less when the shade stops changing...). Brown-to-light blue copper (II) chloride is available too (and is supposed to change closest to 20%) but besides lower cost, cobalt was originally selected because the orange and brown dyes require more heat than the blue to dehydrate and reset the indicator--much more than the silica gel requires. And of course brown indicator means you can't know if you overheated the silica gel and damaged it.
Other compounds are way more effective in reducing humidity but tend to be incredibly corrosive salts like calcium chloride which turn into splashable liquids as they work. If the object is to reduce corrosion, an oxygen absorber may be more effective. For example 1g of iron powder in one packet can remove 300 cm^3 of oxygen from your sealed container. Putting dry ice in the container and letting it sublime a bit before sealing can displace atmospheric gases. Or if you have a welder or soda machine, just filling the container with carbon dioxide or argon would do the same.