berty-phaseD / config.json
electroglyph's picture
squashed: phaseD final step 11964
cd3d68a
Raw History Blame Contribute Delete
2.16 kB
{
"phase": "D",
"init_repo": "electroglyph/berty-phaseC",
"sources": [
{
"name": "msmarco-bm25",
"repo": "sentence-transformers/msmarco-bm25",
"pats": [
"triplet-50/*.parquet"
],
"weight": 0.26,
"w_end": 0.14,
"fmt": "m50"
},
{
"name": "msmarco-dense",
"repo": "sentence-transformers/msmarco-co-condenser-margin-mse-cls-v1",
"pats": [
"triplet-50/*.parquet"
],
"weight": 0.26,
"w_end": 0.38,
"fmt": "m50"
},
{
"name": "hotpotqa",
"repo": "sentence-transformers/hotpotqa",
"pats": [
"triplet-20/*.parquet"
],
"weight": 0.12,
"w_end": 0.12,
"fmt": "h20"
},
{
"name": "nq",
"repo": "Tevatron/wikipedia-nq",
"pats": [
"nq-train.jsonl.gz"
],
"weight": 0.1,
"w_end": 0.1,
"fmt": "dpr"
},
{
"name": "quora",
"repo": "sentence-transformers/quora-duplicates",
"pats": [
"triplet-all/*.parquet",
"pair-class/*.parquet"
],
"weight": 0.09,
"w_end": 0.09,
"fmt": "quora"
},
{
"name": "nli",
"repo": "sentence-transformers/all-nli",
"pats": [
"triplet/train-*.parquet"
],
"weight": 0.07,
"w_end": 0.07,
"fmt": "nli"
},
{
"name": "pairs",
"repo": "sentence-transformers/eli5",
"pats": [
"pair/*.parquet"
],
"weight": 0.04,
"w_end": 0.04,
"fmt": "qa"
},
{
"name": "squad",
"repo": "sentence-transformers/squad",
"pats": [
"pair/*.parquet"
],
"weight": 0.02,
"w_end": 0.02,
"fmt": "qa"
},
{
"name": "stackex",
"repo": "sentence-transformers/stackexchange-duplicates",
"pats": [
"post-post-pair/*.parquet"
],
"weight": 0.04,
"w_end": 0.04,
"fmt": "t12",
"cols": [
"post1",
"post2"
]
}
],
"base_lr": 5e-05,
"seq_q": 64,
"seq_d": 512,
"h_frozen": [
7,
4
],
"n_ance": [
2,
6
],
"curr_anneal_steps": 15000,
"h_hard_eval": 15,
"n_pos": 2,
"nce_lambda": 1.0,
"bwd_lambda": 0.5,
"kl_lambda": 0.4,
"kl_temp": 1.0,
"distill_lambda": 0.05,
"reservoir": 262144,
"scifact_probe": true,
"batch_cfg": {
"micro_q": 64,
"accum": 2,
"q_target": "128 queries/step",
"vram": 95.0
}
}