ETHZ and EPFL announced the release of a Large Language Model (LLM) developed on public infrastructure: Trained on the “Alps” supercomputer at the Swiss National Supercomputing Centre (CSCS) in 8B and 70B parameters configurations, using open-source training data, respecting web crawling opt-outs during data acquisition, and natively fluent in over 1000 languages. Quoting: "The model will be fully open: source code and weights will be publicly available, and the training data will be transparent and reproducible".
I don't know how good it's going to be, but if true for me this is the real definition of "open-source" in AI (not the ridiculous, corporate-promiscuous definition by the Open Source Initiative).
https://ethz.ch/en/news-and-events/eth-news/news/2025/07/a-language-model-built-for-the-public-good.html
#AI #LLM #ETH #EPFL #switzerland #eu #sovereignty #digitalsovereignty #opensource #osi #openwashing #privacy #gdpr
Tommaso Gagliardoni
npub1qz…m5mlw
2025-07-11 21:07:35 UTC
Author Public Key
npub1qz00d7632xesdupfjrl9l7ztkj6q7eunl33cdhnuhguvghg6lyys2m5mlwSeen on
wss://relay.momostr.pinkPublished at
2025-07-11 21:07:35 UTCEvent JSON
{
"id": "c7869ad4caa56e0d983a31da72242152ccb847dc17906f45512c1c85305fc68f",
"pubkey": "009ef6fb5151b306f02990fe5ff84bb4b40f6793fc6386de7cba38c45d1af909",
"created_at": 1752268055,
"kind": 1,
"tags": [
[
"t",
"opensource"
],
[
"t",
"switzerland"
],
[
"t",
"eth"
],
[
"t",
"ai"
],
[
"t",
"epfl"
],
[
"t",
"osi"
],
[
"t",
"sovereignty"
],
[
"t",
"eu"
],
[
"t",
"digitalsovereignty"
],
[
"t",
"llm"
],
[
"t",
"openwashing"
],
[
"t",
"gdpr"
],
[
"proxy",
"https://infosec.exchange/@tomgag/114836639270348675",
"web"
],
[
"t",
"privacy"
],
[
"proxy",
"https://infosec.exchange/users/tomgag/statuses/114836639270348675",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://infosec.exchange/users/tomgag/statuses/114836639270348675",
"pink.momostr"
],
[
"-"
]
],
"content": "ETHZ and EPFL announced the release of a Large Language Model (LLM) developed on public infrastructure: Trained on the “Alps” supercomputer at the Swiss National Supercomputing Centre (CSCS) in 8B and 70B parameters configurations, using open-source training data, respecting web crawling opt-outs during data acquisition, and natively fluent in over 1000 languages. Quoting: \"The model will be fully open: source code and weights will be publicly available, and the training data will be transparent and reproducible\".\n\nI don't know how good it's going to be, but if true for me this is the real definition of \"open-source\" in AI (not the ridiculous, corporate-promiscuous definition by the Open Source Initiative).\n\nhttps://ethz.ch/en/news-and-events/eth-news/news/2025/07/a-language-model-built-for-the-public-good.html\n\n#AI #LLM #ETH #EPFL #switzerland #eu #sovereignty #digitalsovereignty #opensource #osi #openwashing #privacy #gdpr",
"sig": "474657bfaceaeb9e76ed9e7ed92651a92c309c57d6c524428d1a09dd57dcc051eabaae7605030c523223a49abd4ab314b32a9d19676b09c3991408015eb10694"
}