Has anyone here tried running a quantized neural language model (like a really tiny GPT-2 or RWKV) directly on a jailbroken Kindle Paperwhite (512 MB RAM, ARM Linux)? Is it actually possible to get any real LLM working locally on the hardware, even just as a proof of concept?
In theory, it seems doable with the smallest modelssomething like TinyLlama, DistilGPT2, or a worlds smallest RWKV build. Obviously, it would be painfully slow and not very smart, but technically it would be a real AI running natively on a Kindle, which would be kind of amazing.
Would love to hear if anyones managed this, or even just tried!
In theory, it seems doable with the smallest modelssomething like TinyLlama, DistilGPT2, or a worlds smallest RWKV build. Obviously, it would be painfully slow and not very smart, but technically it would be a real AI running natively on a Kindle, which would be kind of amazing.
Would love to hear if anyones managed this, or even just tried!









