queen-v6 / version.json
Crawlnet's picture
Queen v6: 896.5M params, trained from scratch on the Crawlnet crawl
eb40ba6 verified
Raw History Blame Contribute Delete
1.04 kB
{
"version": 6,
"datasetTokens": 622758683,
"datasetPages": 433197,
"params": 896533746,
"trainedAt": "2026-10-14T18:24:28+00:00",
"weightsUrl": null,
"name": "queen-v6",
"paramsNonEmbedding": 435160240,
"tokensTrained": 3309305856,
"epochs": 5.31,
"valBpb": 0.7836986502315648,
"sftMethod": "synthetic+openbook",
"datasetManifestSha256": "41ffc99ab827eeb79c5c45ae005e0114f7a787b4443e22f2a0626b50d5095199",
"gpuHours": 7.393,
"gpu": "8x H100 SXM (rented)",
"nanochatCommit": "92d63d4e8bb4df75c3b71618f31ddde2378b2bcd",
"toy": false,
"tokenizer": "trained on this dataset only (crawler pages and record pages)",
"license": "apache-2.0",
"datasetSources": {
"crawl": {
"pages": 387568
},
"records": {
"pages": 45629,
"tokensApprox": 10514046,
"what": "public market, funding and on-chain records (projects and their funding rounds, investors, token unlocks and token facts, DeFi protocols, chains and stablecoins, DEX and on-chain token statistics)"
}
}
}