nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqky223zcc4q69d8t0me4vg5uw8mw0yxeukjgvz6h92laqnenr0ajsgze5g0 (nprofile…e5g0) Actually I'm not even sure it's the right tool for the job!
Pipelines are uni-directional so you'd lose the backward part
The types involved in neural network layers are not very informative so I think you'll end up with a pipeline [Matrix, Matrix, Matrix …]
You could implement a pipeline of bidirectional transformations in the category of lenses, but abstracting over categories is the topic of the next blog post :)
Andre Videla /
npub12p…ggft9
2025-01-14 18:02:44
in reply to nevent1q…ulst
Author Public Key
npub12ph3zehsckda4lwm2ek3leuzu20j8fjejk9eg32ptj9yrsqsce6qdggft9Seen on
wss://relay.nostr.bandPublished at
2025-01-14 18:02:44Event JSON
{
"id": "748cdd5556ddf8a1949a11815e3222916848a0c5ebe5d38bc5792137d5c9b613",
"pubkey": "506f1166f0c59bdafddb566d1fe782e29f23a659958b9445415c8a41c010c674",
"created_at": 1736877764,
"kind": 1,
"tags": [
[
"p",
"b114a88b18a834569d6fde6ac4538e3edcf21b3cb490c16ae557fa09e6637f65",
"wss://relay.mostr.pub"
],
[
"p",
"a60a88374d8e1cf092c7ea93662aa784fb33b3e75be7725017032e6929ebc5d5",
"wss://relay.mostr.pub"
],
[
"e",
"7b10b51e769017f5ca485be044565ece1bafa2cd5bfadb68957606eef7a5b900",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://types.pl/users/Andrev/statuses/113828021187492215",
"activitypub"
]
],
"content": "nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqky223zcc4q69d8t0me4vg5uw8mw0yxeukjgvz6h92laqnenr0ajsgze5g0 Actually I'm not even sure it's the right tool for the job!\n\nPipelines are uni-directional so you'd lose the backward part\nThe types involved in neural network layers are not very informative so I think you'll end up with a pipeline [Matrix, Matrix, Matrix …]\n\nYou could implement a pipeline of bidirectional transformations in the category of lenses, but abstracting over categories is the topic of the next blog post :)",
"sig": "8541dc8079f0972cacf325956c1c6c8defcbaa0475a19cf4167d232c307971e6ad612ebb1f742c2df4928cbd0618c1938342c7a83bea98e58141de5c1799d163"
}