うちのGemma 4 31B(unsloth/IQ3_XXS, MTP付き)は70 tok/sくらい出るのである程度快適
(なおVRAM不足でコンテキストが伸ばせない模様)
sou7
npub1xy…m4txe
2026-11-02 15:01:17 UTC
in reply to nevent1q…93fm
Author Public Key
npub1xy3wwdfp4t58xnsq72zssajt3zn2pvfadjh85apyxdw0y2rvq6qq8m4txeSeen on
wss://relay.momostr.pinkPublished at
2026-11-02 15:01:17 UTCEvent JSON
{
"id": "8a49b22910b715a260536535509c1e4ca6b54e745c4153e5fc2ba0be6e671ccd",
"pubkey": "3122e73521aae8734e00f28508764b88a6a0b13d6cae7a7424335cf2286c0680",
"created_at": 1793631677,
"kind": 1,
"tags": [
[
"p",
"3122e73521aae8734e00f28508764b88a6a0b13d6cae7a7424335cf2286c0680"
],
[
"e",
"170a7faef3dfe92d7f0c29d913510a126e20e53be72e0f7b4836e3f639bac7e0",
"",
"root",
"3122e73521aae8734e00f28508764b88a6a0b13d6cae7a7424335cf2286c0680"
],
[
"proxy",
"https://mi.sou7.io/notes/at2yawzzqgcs05df",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://mi.sou7.io/notes/at2yawzzqgcs05df",
"pink.momostr"
],
[
"-"
]
],
"content": "うちのGemma 4 31B(unsloth/IQ3_XXS, MTP付き)は70 tok/sくらい出るのである程度快適\n(なおVRAM不足でコンテキストが伸ばせない模様)",
"sig": "822eb19cde54baffed36d64b31cab7448eeb78a854296c7316115d602e131f104162c985cfe4493b621d3cdb65ab2f7c45c4f501864d3cf1a81a3384ecf4ac9a"
}