fasttext quantization documentation is confusing/misleading

[yrro]

https://docs.rspamd.com/modules/neural/ says

# Download and extract a full model
wget https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.en.300.bin.gz
gunzip cc.en.300.bin.gz

# Quantize to reduce size and dimensions
./fasttext quantize -input cc.en.300 -output cc.en.300.q

but:

$ fasttext quantize -input cc.en.300 -output cc.en.300.q
terminate called after throwing an instance of 'std::invalid_argument'
  what():  cc.en.300.q.bin cannot be opened for loading!
Aborted (core dumped)

This is with

ii  fasttext       0.9.2+ds-7build1 arm64        Efficient learning of word representations and sentence classification library

My assumption is that the CLI changed since the docs were written. But then this version of fasttext is 6 years old. So maybe the docs are just totally wrong?

https://www.facebook.com/groups/1174547215919768/posts/1668216449886173/ lead me to try:

$ fasttext quantize -input cc.en.300 -output cc.en.300
terminate called after throwing an instance of 'std::invalid_argument'
  what():  For now we only support quantization of supervised models
Aborted (core dumped)

... research this leads to Gemini saying stuff like "Because cc.en.300.bin is a set of raw word vectors (unsupervised), it lacks the classification headers required by the quantize utility." and at this point I have to throw my hands up and admit I have no earthly idea what any of this means!