Commit ·
cd3d68a
0
Parent(s):
squashed: phaseD final step 11964
Browse files- .gitattributes +35 -0
- body.pt +3 -0
- body_best.pt +3 -0
- body_step0011428.pt +3 -0
- body_step0011729.pt +3 -0
- body_step0011964.pt +3 -0
- config.json +128 -0
- dense_opt.pt +3 -0
- heldD.json +0 -0
- logtemp.pt +3 -0
- logtemp_best.pt +3 -0
- metrics.jsonl +75 -0
- pooler.pt +3 -0
- pooler_best.pt +3 -0
- sparse_opt.pt +3 -0
- training_state.json +56 -0
.gitattributes
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
| 2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
| 3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
| 4 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
| 5 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
| 6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
| 7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
| 8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
| 9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
| 10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
| 11 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
| 12 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
| 13 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
| 14 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
| 15 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
| 16 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
| 17 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
| 18 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
| 19 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
| 20 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
| 21 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
| 22 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
| 23 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
| 24 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
| 25 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
| 26 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
| 27 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
| 28 |
+
*.tar filter=lfs diff=lfs merge=lfs -text
|
| 29 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
| 30 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
| 31 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
| 32 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
| 33 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
body.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc81789a395c7c3c0ef94beaf87c8084c1aa9053236a0f35d330e6c2211c2888
|
| 3 |
+
size 8113043905
|
body_best.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da5f1a22cecc8de5f908742feb2cc1f8289f19188f216820aab4b821f5d32714
|
| 3 |
+
size 8113056307
|
body_step0011428.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ca6819e3cf12f416e8c9f87ef566a1c0aa65f6c41f02772a2a772d1af1d5a1e
|
| 3 |
+
size 8113043905
|
body_step0011729.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc75b160fc7794ab21b49dbfe4a59c7fd7d385936cc2748b8ff9251f3c9e42c7
|
| 3 |
+
size 8113043905
|
body_step0011964.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc81789a395c7c3c0ef94beaf87c8084c1aa9053236a0f35d330e6c2211c2888
|
| 3 |
+
size 8113043905
|
config.json
ADDED
|
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"phase": "D",
|
| 3 |
+
"init_repo": "electroglyph/berty-phaseC",
|
| 4 |
+
"sources": [
|
| 5 |
+
{
|
| 6 |
+
"name": "msmarco-bm25",
|
| 7 |
+
"repo": "sentence-transformers/msmarco-bm25",
|
| 8 |
+
"pats": [
|
| 9 |
+
"triplet-50/*.parquet"
|
| 10 |
+
],
|
| 11 |
+
"weight": 0.26,
|
| 12 |
+
"w_end": 0.14,
|
| 13 |
+
"fmt": "m50"
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"name": "msmarco-dense",
|
| 17 |
+
"repo": "sentence-transformers/msmarco-co-condenser-margin-mse-cls-v1",
|
| 18 |
+
"pats": [
|
| 19 |
+
"triplet-50/*.parquet"
|
| 20 |
+
],
|
| 21 |
+
"weight": 0.26,
|
| 22 |
+
"w_end": 0.38,
|
| 23 |
+
"fmt": "m50"
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"name": "hotpotqa",
|
| 27 |
+
"repo": "sentence-transformers/hotpotqa",
|
| 28 |
+
"pats": [
|
| 29 |
+
"triplet-20/*.parquet"
|
| 30 |
+
],
|
| 31 |
+
"weight": 0.12,
|
| 32 |
+
"w_end": 0.12,
|
| 33 |
+
"fmt": "h20"
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"name": "nq",
|
| 37 |
+
"repo": "Tevatron/wikipedia-nq",
|
| 38 |
+
"pats": [
|
| 39 |
+
"nq-train.jsonl.gz"
|
| 40 |
+
],
|
| 41 |
+
"weight": 0.1,
|
| 42 |
+
"w_end": 0.1,
|
| 43 |
+
"fmt": "dpr"
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"name": "quora",
|
| 47 |
+
"repo": "sentence-transformers/quora-duplicates",
|
| 48 |
+
"pats": [
|
| 49 |
+
"triplet-all/*.parquet",
|
| 50 |
+
"pair-class/*.parquet"
|
| 51 |
+
],
|
| 52 |
+
"weight": 0.09,
|
| 53 |
+
"w_end": 0.09,
|
| 54 |
+
"fmt": "quora"
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"name": "nli",
|
| 58 |
+
"repo": "sentence-transformers/all-nli",
|
| 59 |
+
"pats": [
|
| 60 |
+
"triplet/train-*.parquet"
|
| 61 |
+
],
|
| 62 |
+
"weight": 0.07,
|
| 63 |
+
"w_end": 0.07,
|
| 64 |
+
"fmt": "nli"
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"name": "pairs",
|
| 68 |
+
"repo": "sentence-transformers/eli5",
|
| 69 |
+
"pats": [
|
| 70 |
+
"pair/*.parquet"
|
| 71 |
+
],
|
| 72 |
+
"weight": 0.04,
|
| 73 |
+
"w_end": 0.04,
|
| 74 |
+
"fmt": "qa"
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"name": "squad",
|
| 78 |
+
"repo": "sentence-transformers/squad",
|
| 79 |
+
"pats": [
|
| 80 |
+
"pair/*.parquet"
|
| 81 |
+
],
|
| 82 |
+
"weight": 0.02,
|
| 83 |
+
"w_end": 0.02,
|
| 84 |
+
"fmt": "qa"
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"name": "stackex",
|
| 88 |
+
"repo": "sentence-transformers/stackexchange-duplicates",
|
| 89 |
+
"pats": [
|
| 90 |
+
"post-post-pair/*.parquet"
|
| 91 |
+
],
|
| 92 |
+
"weight": 0.04,
|
| 93 |
+
"w_end": 0.04,
|
| 94 |
+
"fmt": "t12",
|
| 95 |
+
"cols": [
|
| 96 |
+
"post1",
|
| 97 |
+
"post2"
|
| 98 |
+
]
|
| 99 |
+
}
|
| 100 |
+
],
|
| 101 |
+
"base_lr": 5e-05,
|
| 102 |
+
"seq_q": 64,
|
| 103 |
+
"seq_d": 512,
|
| 104 |
+
"h_frozen": [
|
| 105 |
+
7,
|
| 106 |
+
4
|
| 107 |
+
],
|
| 108 |
+
"n_ance": [
|
| 109 |
+
2,
|
| 110 |
+
6
|
| 111 |
+
],
|
| 112 |
+
"curr_anneal_steps": 15000,
|
| 113 |
+
"h_hard_eval": 15,
|
| 114 |
+
"n_pos": 2,
|
| 115 |
+
"nce_lambda": 1.0,
|
| 116 |
+
"bwd_lambda": 0.5,
|
| 117 |
+
"kl_lambda": 0.4,
|
| 118 |
+
"kl_temp": 1.0,
|
| 119 |
+
"distill_lambda": 0.05,
|
| 120 |
+
"reservoir": 262144,
|
| 121 |
+
"scifact_probe": true,
|
| 122 |
+
"batch_cfg": {
|
| 123 |
+
"micro_q": 64,
|
| 124 |
+
"accum": 2,
|
| 125 |
+
"q_target": "128 queries/step",
|
| 126 |
+
"vram": 95.0
|
| 127 |
+
}
|
| 128 |
+
}
|
dense_opt.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ce43d48e48ef6b6482d3d3684f6919c4964602fda1ec19e8d45019dbc3447c6
|
| 3 |
+
size 1202222667
|
heldD.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
logtemp.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f775ccf6fa37b0327b204c07f4f6089644268a505c1eba0320b368544911963
|
| 3 |
+
size 1577
|
logtemp_best.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b480237e7d86104931ba9c2784bb7e61a6b505bd75db7125bda481393bfa21b
|
| 3 |
+
size 1612
|
metrics.jsonl
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 128, "loss_ema": 14.1454, "held_nce_hard": 3.0302, "held_acc_hard": 0.3125, "held_margin": 0.0473, "held_margin_hard": -0.0233, "held_cos_q": 0.4675, "scifact_ndcg": 0.5157, "held_n": 32, "held_Hn": 15, "temp": 0.02198, "lr": 1.2800000000000001e-05, "sched": "warmup", "bank_filled": 32768, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.2117, "mrl_64": 0.2578, "mrl_128": 0.3283, "mrl_256": 0.3998, "mrl_512": 0.4658, "mrl_768": 0.5018, "mrl_1024": 0.5447, "held_srcs": {"stackex": 2, "msmarco-dense": 5, "nli": 3, "msmarco-bm25": 8, "quora": 2, "hotpotqa": 5, "nq": 5, "pairs": 2}}
|
| 2 |
+
{"step": 262, "loss_ema": 6.2918, "held_nce_hard": 2.105, "held_acc_hard": 0.4531, "held_margin": 0.0483, "held_margin_hard": 0.006, "held_cos_q": 0.3619, "scifact_ndcg": 0.5476, "held_n": 67, "held_Hn": 15, "temp": 0.022, "lr": 2.62e-05, "sched": "warmup", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.64, "mrl_32": 0.3265, "mrl_64": 0.3689, "mrl_128": 0.4477, "mrl_256": 0.5267, "mrl_512": 0.6106, "mrl_768": 0.6495, "mrl_1024": 0.6899, "held_srcs": {"stackex": 3, "msmarco-dense": 10, "nli": 5, "msmarco-bm25": 18, "quora": 8, "hotpotqa": 11, "nq": 6, "pairs": 3}}
|
| 3 |
+
{"step": 404, "loss_ema": 4.9224, "held_nce_hard": 1.8885, "held_acc_hard": 0.5, "held_margin": 0.0499, "held_margin_hard": 0.0068, "held_cos_q": 0.3311, "scifact_ndcg": 0.5887, "held_n": 103, "held_Hn": 15, "temp": 0.0219, "lr": 4.0400000000000006e-05, "sched": "warmup", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.51, "mrl_32": 0.3226, "mrl_64": 0.3864, "mrl_128": 0.4808, "mrl_256": 0.5692, "mrl_512": 0.6577, "mrl_768": 0.7025, "mrl_1024": 0.7343, "held_srcs": {"stackex": 3, "msmarco-dense": 10, "nli": 5, "msmarco-bm25": 18, "quora": 8, "hotpotqa": 11, "nq": 6, "pairs": 3}}
|
| 4 |
+
{"step": 536, "loss_ema": 4.6934, "held_nce_hard": 1.6477, "held_acc_hard": 0.5469, "held_margin": 0.0614, "held_margin_hard": 0.0118, "held_cos_q": 0.2893, "scifact_ndcg": 0.6197, "held_n": 137, "held_Hn": 15, "temp": 0.02177, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.63, "mrl_32": 0.348, "mrl_64": 0.42, "mrl_128": 0.5334, "mrl_256": 0.6137, "mrl_512": 0.7081, "mrl_768": 0.7465, "mrl_1024": 0.7749, "held_srcs": {"stackex": 4, "nli": 5, "msmarco-bm25": 16, "hotpotqa": 12, "msmarco-dense": 10, "pairs": 1, "nq": 7, "quora": 8, "squad": 1}}
|
| 5 |
+
{"step": 683, "loss_ema": 4.6014, "held_nce_hard": 1.4499, "held_acc_hard": 0.6406, "held_margin": 0.0538, "held_margin_hard": 0.0082, "held_cos_q": 0.2578, "scifact_ndcg": 0.5756, "held_n": 175, "held_Hn": 15, "temp": 0.02163, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.4, "mrl_32": 0.2705, "mrl_64": 0.3772, "mrl_128": 0.4943, "mrl_256": 0.597, "mrl_512": 0.6984, "mrl_768": 0.7392, "mrl_1024": 0.7693, "held_srcs": {"stackex": 4, "nli": 5, "msmarco-bm25": 16, "hotpotqa": 12, "msmarco-dense": 10, "pairs": 1, "nq": 7, "quora": 8, "squad": 1}}
|
| 6 |
+
{"step": 830, "loss_ema": 4.5461, "held_nce_hard": 1.9509, "held_acc_hard": 0.4844, "held_margin": 0.0523, "held_margin_hard": -0.0014, "held_cos_q": 0.1459, "scifact_ndcg": 0.6004, "held_n": 212, "held_Hn": 15, "temp": 0.02148, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.48, "mrl_32": 0.7424, "mrl_64": 0.743, "mrl_128": 0.7935, "mrl_256": 0.8205, "mrl_512": 0.8631, "mrl_768": 0.8752, "mrl_1024": 0.8882, "held_srcs": {"stackex": 2, "msmarco-bm25": 22, "quora": 4, "msmarco-dense": 18, "nq": 8, "pairs": 2, "hotpotqa": 3, "nli": 4, "squad": 1}}
|
| 7 |
+
{"step": 963, "loss_ema": 4.5408, "held_nce_hard": 1.957, "held_acc_hard": 0.4844, "held_margin": 0.0508, "held_margin_hard": -0.0002, "held_cos_q": 0.2112, "scifact_ndcg": 0.6085, "held_n": 247, "held_Hn": 15, "temp": 0.02135, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.62, "mrl_32": 0.4797, "mrl_64": 0.5309, "mrl_128": 0.666, "mrl_256": 0.7324, "mrl_512": 0.7966, "mrl_768": 0.8232, "mrl_1024": 0.8422, "held_srcs": {"stackex": 2, "msmarco-bm25": 22, "quora": 4, "msmarco-dense": 18, "nq": 8, "pairs": 2, "hotpotqa": 3, "nli": 4, "squad": 1}}
|
| 8 |
+
{"step": 1107, "loss_ema": 4.55, "held_nce_hard": 1.8913, "held_acc_hard": 0.4375, "held_margin": 0.0567, "held_margin_hard": 0.0055, "held_cos_q": 0.1854, "scifact_ndcg": 0.6042, "held_n": 283, "held_Hn": 15, "temp": 0.02121, "lr": 4.9910544977990376e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.39, "mrl_32": 0.5227, "mrl_64": 0.569, "mrl_128": 0.6933, "mrl_256": 0.7513, "mrl_512": 0.8147, "mrl_768": 0.8391, "mrl_1024": 0.8591, "held_srcs": {"stackex": 2, "msmarco-bm25": 21, "pairs": 2, "nq": 5, "hotpotqa": 11, "nli": 6, "quora": 5, "msmarco-dense": 10, "squad": 2}}
|
| 9 |
+
{"step": 1243, "loss_ema": 4.4777, "held_nce_hard": 1.8958, "held_acc_hard": 0.4844, "held_margin": 0.0511, "held_margin_hard": 0.0032, "held_cos_q": 0.1433, "scifact_ndcg": 0.5836, "held_n": 318, "held_Hn": 15, "temp": 0.02108, "lr": 4.982531986851665e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.38, "mrl_32": 0.6695, "mrl_64": 0.6917, "mrl_128": 0.7882, "mrl_256": 0.8221, "mrl_512": 0.8539, "mrl_768": 0.8758, "mrl_1024": 0.8877, "held_srcs": {"stackex": 2, "msmarco-bm25": 21, "pairs": 2, "nq": 5, "hotpotqa": 11, "nli": 6, "quora": 5, "msmarco-dense": 10, "squad": 2}}
|
| 10 |
+
{"step": 1380, "loss_ema": 4.4267, "held_nce_hard": 1.7678, "held_acc_hard": 0.5312, "held_margin": 0.05, "held_margin_hard": 0.0001, "held_cos_q": 0.2272, "scifact_ndcg": 0.5476, "held_n": 353, "held_Hn": 15, "temp": 0.02095, "lr": 4.973932046178355e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.66, "mrl_32": 0.3577, "mrl_64": 0.4378, "mrl_128": 0.5758, "mrl_256": 0.671, "mrl_512": 0.7548, "mrl_768": 0.7923, "mrl_1024": 0.8205, "held_srcs": {"stackex": 1, "nli": 8, "hotpotqa": 6, "msmarco-dense": 16, "nq": 9, "msmarco-bm25": 15, "quora": 5, "pairs": 2, "squad": 2}}
|
| 11 |
+
{"step": 1528, "loss_ema": 4.4308, "held_nce_hard": 1.9644, "held_acc_hard": 0.4844, "held_margin": 0.0528, "held_margin_hard": -0.0046, "held_cos_q": 0.2125, "scifact_ndcg": 0.563, "held_n": 391, "held_Hn": 15, "temp": 0.02081, "lr": 4.964624859946621e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.47, "mrl_32": 0.408, "mrl_64": 0.5151, "mrl_128": 0.6198, "mrl_256": 0.7095, "mrl_512": 0.7769, "mrl_768": 0.8141, "mrl_1024": 0.8363, "held_srcs": {"stackex": 3, "msmarco-bm25": 21, "nq": 9, "msmarco-dense": 20, "hotpotqa": 4, "quora": 2, "nli": 2, "pairs": 1, "squad": 2}}
|
| 12 |
+
{"step": 1661, "loss_ema": 4.4027, "held_nce_hard": 1.7158, "held_acc_hard": 0.5469, "held_margin": 0.0561, "held_margin_hard": -0.0027, "held_cos_q": 0.1801, "scifact_ndcg": 0.5789, "held_n": 426, "held_Hn": 15, "temp": 0.02069, "lr": 4.956246059267034e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.47, "mrl_32": 0.5107, "mrl_64": 0.5475, "mrl_128": 0.6749, "mrl_256": 0.7542, "mrl_512": 0.8231, "mrl_768": 0.848, "mrl_1024": 0.8677, "held_srcs": {"stackex": 3, "msmarco-bm25": 21, "nq": 9, "msmarco-dense": 20, "hotpotqa": 4, "quora": 2, "nli": 2, "pairs": 1, "squad": 2}}
|
| 13 |
+
{"step": 1814, "loss_ema": 4.4153, "held_nce_hard": 1.6669, "held_acc_hard": 0.6094, "held_margin": 0.0484, "held_margin_hard": 0.0015, "held_cos_q": 0.2116, "scifact_ndcg": 0.561, "held_n": 465, "held_Hn": 15, "temp": 0.02056, "lr": 4.946589734352345e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.4553, "mrl_64": 0.5153, "mrl_128": 0.6253, "mrl_256": 0.709, "mrl_512": 0.7868, "mrl_768": 0.821, "mrl_1024": 0.8471, "held_srcs": {"stackex": 3, "nli": 5, "msmarco-dense": 15, "nq": 8, "hotpotqa": 7, "quora": 4, "msmarco-bm25": 19, "squad": 3}}
|
| 14 |
+
{"step": 1932, "loss_ema": 4.4587, "held_nce_hard": 2.1305, "held_acc_hard": 0.5312, "held_margin": 0.0287, "held_margin_hard": -0.0065, "held_cos_q": 0.1763, "scifact_ndcg": 0.5533, "held_n": 495, "held_Hn": 15, "temp": 0.02046, "lr": 5e-05, "sched": "stable", "bank_filled": 30208, "micro_q": 64, "accum": 2, "gb": 25.53, "mrl_32": 0.4594, "mrl_64": 0.5825, "mrl_128": 0.6979, "mrl_256": 0.7708, "mrl_512": 0.8239, "mrl_768": 0.8512, "mrl_1024": 0.8684, "held_srcs": {"stackex": 7, "msmarco-dense": 63, "nli": 24, "msmarco-bm25": 71, "quora": 21, "hotpotqa": 26, "nq": 31, "pairs": 6, "squad": 7}}
|
| 15 |
+
{"step": 2071, "loss_ema": 4.403, "held_nce_hard": 2.0368, "held_acc_hard": 0.5234, "held_margin": 0.037, "held_margin_hard": 0.0001, "held_cos_q": 0.1851, "scifact_ndcg": 0.5678, "held_n": 512, "held_Hn": 15, "temp": 0.02033, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.43, "mrl_32": 0.4444, "mrl_64": 0.4935, "mrl_128": 0.6352, "mrl_256": 0.7259, "mrl_512": 0.7961, "mrl_768": 0.826, "mrl_1024": 0.8478, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 16 |
+
{"step": 2221, "loss_ema": 4.4601, "held_nce_hard": 2.0605, "held_acc_hard": 0.5039, "held_margin": 0.0345, "held_margin_hard": -0.0031, "held_cos_q": 0.1286, "scifact_ndcg": 0.5058, "held_n": 512, "held_Hn": 15, "temp": 0.0202, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.48, "mrl_32": 0.7226, "mrl_64": 0.7476, "mrl_128": 0.8151, "mrl_256": 0.8457, "mrl_512": 0.87, "mrl_768": 0.892, "mrl_1024": 0.9016, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 17 |
+
{"step": 2359, "loss_ema": 4.5165, "held_nce_hard": 2.1596, "held_acc_hard": 0.4961, "held_margin": 0.0319, "held_margin_hard": -0.0027, "held_cos_q": 0.1456, "scifact_ndcg": 0.5519, "held_n": 512, "held_Hn": 15, "temp": 0.02008, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.53, "mrl_32": 0.7177, "mrl_64": 0.7284, "mrl_128": 0.7852, "mrl_256": 0.8298, "mrl_512": 0.8539, "mrl_768": 0.8773, "mrl_1024": 0.8916, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 18 |
+
{"step": 2510, "loss_ema": 4.5088, "held_nce_hard": 2.1554, "held_acc_hard": 0.4766, "held_margin": 0.0365, "held_margin_hard": -0.0023, "held_cos_q": 0.1279, "scifact_ndcg": 0.5597, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.52, "mrl_32": 0.7754, "mrl_64": 0.7951, "mrl_128": 0.8247, "mrl_256": 0.8634, "mrl_512": 0.8848, "mrl_768": 0.8975, "mrl_1024": 0.9052, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 19 |
+
{"step": 2679, "loss_ema": 4.5402, "held_nce_hard": 2.1751, "held_acc_hard": 0.4219, "held_margin": 0.037, "held_margin_hard": -0.0011, "held_cos_q": 0.1593, "scifact_ndcg": 0.5742, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.63, "mrl_32": 0.5312, "mrl_64": 0.5898, "mrl_128": 0.7049, "mrl_256": 0.7704, "mrl_512": 0.8272, "mrl_768": 0.8566, "mrl_1024": 0.874, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 20 |
+
{"step": 2828, "loss_ema": 4.3861, "held_nce_hard": 1.9393, "held_acc_hard": 0.5586, "held_margin": 0.033, "held_margin_hard": -0.0014, "held_cos_q": 0.232, "scifact_ndcg": 0.5441, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.45, "mrl_32": 0.3772, "mrl_64": 0.4392, "mrl_128": 0.5621, "mrl_256": 0.6621, "mrl_512": 0.7387, "mrl_768": 0.7788, "mrl_1024": 0.807, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 21 |
+
{"step": 2994, "loss_ema": 4.491, "held_nce_hard": 2.1207, "held_acc_hard": 0.4805, "held_margin": 0.0345, "held_margin_hard": -0.0015, "held_cos_q": 0.1902, "scifact_ndcg": 0.5311, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 5e-05, "sched": "stable", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.51, "mrl_32": 0.4631, "mrl_64": 0.5487, "mrl_128": 0.6354, "mrl_256": 0.7273, "mrl_512": 0.7845, "mrl_768": 0.8191, "mrl_1024": 0.8419, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 22 |
+
{"step": 3146, "loss_ema": 4.3859, "held_nce_hard": 2.0425, "held_acc_hard": 0.5039, "held_margin": 0.0345, "held_margin_hard": -0.0018, "held_cos_q": 0.1832, "scifact_ndcg": 0.5425, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.9905535765083215e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.5, "mrl_32": 0.4757, "mrl_64": 0.5226, "mrl_128": 0.6462, "mrl_256": 0.7368, "mrl_512": 0.7958, "mrl_768": 0.8333, "mrl_1024": 0.8564, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 23 |
+
{"step": 3297, "loss_ema": 4.409, "held_nce_hard": 2.1438, "held_acc_hard": 0.4336, "held_margin": 0.038, "held_margin_hard": -0.0022, "held_cos_q": 0.1504, "scifact_ndcg": 0.5466, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.981089238309228e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.42, "mrl_32": 0.6265, "mrl_64": 0.7224, "mrl_128": 0.773, "mrl_256": 0.8126, "mrl_512": 0.8441, "mrl_768": 0.8673, "mrl_1024": 0.8827, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 24 |
+
{"step": 3459, "loss_ema": 4.303, "held_nce_hard": 2.0114, "held_acc_hard": 0.5859, "held_margin": 0.0296, "held_margin_hard": -0.0015, "held_cos_q": 0.1547, "scifact_ndcg": 0.5191, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.970915408654627e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.55, "mrl_32": 0.4062, "mrl_64": 0.4941, "mrl_128": 0.6231, "mrl_256": 0.7283, "mrl_512": 0.8056, "mrl_768": 0.8423, "mrl_1024": 0.863, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 25 |
+
{"step": 3611, "loss_ema": 4.3829, "held_nce_hard": 2.2549, "held_acc_hard": 0.4336, "held_margin": 0.0299, "held_margin_hard": -0.0042, "held_cos_q": 0.1355, "scifact_ndcg": 0.4971, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.961350622562368e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.45, "mrl_32": 0.5873, "mrl_64": 0.6635, "mrl_128": 0.7566, "mrl_256": 0.8353, "mrl_512": 0.8744, "mrl_768": 0.8906, "mrl_1024": 0.9045, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 26 |
+
{"step": 3776, "loss_ema": 4.3646, "held_nce_hard": 2.2276, "held_acc_hard": 0.3477, "held_margin": 0.0381, "held_margin_hard": -0.0038, "held_cos_q": 0.164, "scifact_ndcg": 0.5293, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.950946879133324e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.5172, "mrl_64": 0.5786, "mrl_128": 0.6764, "mrl_256": 0.764, "mrl_512": 0.8211, "mrl_768": 0.8511, "mrl_1024": 0.8702, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 27 |
+
{"step": 3928, "loss_ema": 4.3309, "held_nce_hard": 1.986, "held_acc_hard": 0.5234, "held_margin": 0.0346, "held_margin_hard": -0.0022, "held_cos_q": 0.1838, "scifact_ndcg": 0.5478, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.941343440806357e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.5, "mrl_32": 0.4298, "mrl_64": 0.5083, "mrl_128": 0.6241, "mrl_256": 0.7087, "mrl_512": 0.7939, "mrl_768": 0.8259, "mrl_1024": 0.848, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 28 |
+
{"step": 4080, "loss_ema": 4.3158, "held_nce_hard": 1.9799, "held_acc_hard": 0.4688, "held_margin": 0.0389, "held_margin_hard": -0.0023, "held_cos_q": 0.1808, "scifact_ndcg": 0.5149, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.931721301939111e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.47, "mrl_32": 0.4576, "mrl_64": 0.5245, "mrl_128": 0.6216, "mrl_256": 0.7092, "mrl_512": 0.7785, "mrl_768": 0.813, "mrl_1024": 0.8375, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 29 |
+
{"step": 4242, "loss_ema": 4.2976, "held_nce_hard": 2.0513, "held_acc_hard": 0.5078, "held_margin": 0.0341, "held_margin_hard": -0.0026, "held_cos_q": 0.1343, "scifact_ndcg": 0.5487, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.9214454177609245e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.5875, "mrl_64": 0.6282, "mrl_128": 0.731, "mrl_256": 0.7868, "mrl_512": 0.8479, "mrl_768": 0.8706, "mrl_1024": 0.886, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 30 |
+
{"step": 4394, "loss_ema": 4.3762, "held_nce_hard": 2.5529, "held_acc_hard": 0.3086, "held_margin": 0.03, "held_margin_hard": -0.0162, "held_cos_q": 0.1949, "scifact_ndcg": 0.4295, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.9117842990098825e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.48, "mrl_32": 0.4054, "mrl_64": 0.4759, "mrl_128": 0.6075, "mrl_256": 0.7055, "mrl_512": 0.7858, "mrl_768": 0.8199, "mrl_1024": 0.8433, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 31 |
+
{"step": 4561, "loss_ema": 4.2822, "held_nce_hard": 2.3676, "held_acc_hard": 0.3164, "held_margin": 0.0331, "held_margin_hard": -0.0101, "held_cos_q": 0.1855, "scifact_ndcg": 0.3979, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.90114782474473e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.54, "mrl_32": 0.4527, "mrl_64": 0.5213, "mrl_128": 0.6314, "mrl_256": 0.733, "mrl_512": 0.8058, "mrl_768": 0.8391, "mrl_1024": 0.8648, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 32 |
+
{"step": 4714, "loss_ema": 4.3292, "held_nce_hard": 2.045, "held_acc_hard": 0.4062, "held_margin": 0.0436, "held_margin_hard": -0.0013, "held_cos_q": 0.1608, "scifact_ndcg": 0.5508, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.891382728840588e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.53, "mrl_32": 0.4957, "mrl_64": 0.5461, "mrl_128": 0.6758, "mrl_256": 0.7665, "mrl_512": 0.8093, "mrl_768": 0.836, "mrl_1024": 0.8586, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 33 |
+
{"step": 4867, "loss_ema": 4.4119, "held_nce_hard": 2.0373, "held_acc_hard": 0.5312, "held_margin": 0.0353, "held_margin_hard": -0.0018, "held_cos_q": 0.1593, "scifact_ndcg": 0.5494, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.8815980989835696e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.61, "mrl_32": 0.4399, "mrl_64": 0.5284, "mrl_128": 0.6493, "mrl_256": 0.741, "mrl_512": 0.8132, "mrl_768": 0.8461, "mrl_1024": 0.8664, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 34 |
+
{"step": 5033, "loss_ema": 4.2901, "held_nce_hard": 2.1311, "held_acc_hard": 0.4141, "held_margin": 0.04, "held_margin_hard": -0.0035, "held_cos_q": 0.1967, "scifact_ndcg": 0.5072, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.8709598643388554e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.4078, "mrl_64": 0.4894, "mrl_128": 0.6078, "mrl_256": 0.7023, "mrl_512": 0.7788, "mrl_768": 0.8198, "mrl_1024": 0.8427, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 35 |
+
{"step": 5183, "loss_ema": 4.2714, "held_nce_hard": 1.9515, "held_acc_hard": 0.5898, "held_margin": 0.0325, "held_margin_hard": -0.0025, "held_cos_q": 0.1202, "scifact_ndcg": 0.5245, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.861326979333935e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.59, "mrl_32": 0.6315, "mrl_64": 0.6706, "mrl_128": 0.7491, "mrl_256": 0.8086, "mrl_512": 0.8586, "mrl_768": 0.89, "mrl_1024": 0.8987, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 36 |
+
{"step": 5348, "loss_ema": 4.3124, "held_nce_hard": 2.0715, "held_acc_hard": 0.4219, "held_margin": 0.0405, "held_margin_hard": -0.0029, "held_cos_q": 0.1789, "scifact_ndcg": 0.5345, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.8507087111060385e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.5, "mrl_32": 0.4181, "mrl_64": 0.5263, "mrl_128": 0.6494, "mrl_256": 0.7432, "mrl_512": 0.8098, "mrl_768": 0.8434, "mrl_1024": 0.8645, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 37 |
+
{"step": 5501, "loss_ema": 4.2943, "held_nce_hard": 2.3333, "held_acc_hard": 0.3359, "held_margin": 0.0409, "held_margin_hard": -0.0098, "held_cos_q": 0.1965, "scifact_ndcg": 0.5058, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.840841868931478e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.4787, "mrl_64": 0.5452, "mrl_128": 0.654, "mrl_256": 0.7484, "mrl_512": 0.8034, "mrl_768": 0.8333, "mrl_1024": 0.8509, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 38 |
+
{"step": 5652, "loss_ema": 4.3043, "held_nce_hard": 1.9855, "held_acc_hard": 0.5234, "held_margin": 0.0351, "held_margin_hard": -0.0017, "held_cos_q": 0.1586, "scifact_ndcg": 0.5428, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.83108424683321e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.41, "mrl_32": 0.5269, "mrl_64": 0.5833, "mrl_128": 0.6736, "mrl_256": 0.7522, "mrl_512": 0.8235, "mrl_768": 0.857, "mrl_1024": 0.8749, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 39 |
+
{"step": 5819, "loss_ema": 4.3874, "held_nce_hard": 2.1857, "held_acc_hard": 0.457, "held_margin": 0.0324, "held_margin_hard": -0.003, "held_cos_q": 0.1728, "scifact_ndcg": 0.5437, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.820269701998012e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.51, "mrl_32": 0.4111, "mrl_64": 0.5037, "mrl_128": 0.5995, "mrl_256": 0.7173, "mrl_512": 0.7951, "mrl_768": 0.8368, "mrl_1024": 0.8593, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 40 |
+
{"step": 5968, "loss_ema": 4.2392, "held_nce_hard": 2.0579, "held_acc_hard": 0.5469, "held_margin": 0.0332, "held_margin_hard": -0.0019, "held_cos_q": 0.1315, "scifact_ndcg": 0.5487, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.8106002743940386e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.63, "mrl_32": 0.54, "mrl_64": 0.5985, "mrl_128": 0.6984, "mrl_256": 0.765, "mrl_512": 0.8288, "mrl_768": 0.8565, "mrl_1024": 0.8753, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 41 |
+
{"step": 6132, "loss_ema": 4.2543, "held_nce_hard": 2.031, "held_acc_hard": 0.4766, "held_margin": 0.0371, "held_margin_hard": -0.0024, "held_cos_q": 0.1759, "scifact_ndcg": 0.5305, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7999348953918115e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.63, "mrl_32": 0.4537, "mrl_64": 0.5028, "mrl_128": 0.617, "mrl_256": 0.6993, "mrl_512": 0.7819, "mrl_768": 0.8258, "mrl_1024": 0.8477, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 42 |
+
{"step": 6283, "loss_ema": 4.2854, "held_nce_hard": 1.9881, "held_acc_hard": 0.5703, "held_margin": 0.0354, "held_margin_hard": -0.0016, "held_cos_q": 0.1534, "scifact_ndcg": 0.564, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.790093944798995e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.51, "mrl_32": 0.556, "mrl_64": 0.6108, "mrl_128": 0.6957, "mrl_256": 0.7584, "mrl_512": 0.819, "mrl_768": 0.844, "mrl_1024": 0.8624, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 43 |
+
{"step": 6434, "loss_ema": 4.3034, "held_nce_hard": 1.9811, "held_acc_hard": 0.5234, "held_margin": 0.0374, "held_margin_hard": -0.0021, "held_cos_q": 0.1094, "scifact_ndcg": 0.542, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.780232734919922e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.46, "mrl_32": 0.6527, "mrl_64": 0.7191, "mrl_128": 0.792, "mrl_256": 0.8403, "mrl_512": 0.8808, "mrl_768": 0.8995, "mrl_1024": 0.9082, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 44 |
+
{"step": 6601, "loss_ema": 4.2365, "held_nce_hard": 2.0406, "held_acc_hard": 0.4453, "held_margin": 0.0364, "held_margin_hard": -0.0028, "held_cos_q": 0.1679, "scifact_ndcg": 0.5234, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7693028840701656e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.5, "mrl_32": 0.5002, "mrl_64": 0.552, "mrl_128": 0.6451, "mrl_256": 0.727, "mrl_512": 0.8014, "mrl_768": 0.8382, "mrl_1024": 0.8619, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 45 |
+
{"step": 6753, "loss_ema": 4.1879, "held_nce_hard": 1.9168, "held_acc_hard": 0.6641, "held_margin": 0.0306, "held_margin_hard": -0.0012, "held_cos_q": 0.133, "scifact_ndcg": 0.5494, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.759332936452335e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.47, "mrl_32": 0.6339, "mrl_64": 0.689, "mrl_128": 0.7481, "mrl_256": 0.7966, "mrl_512": 0.8532, "mrl_768": 0.8792, "mrl_1024": 0.8956, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 46 |
+
{"step": 6908, "loss_ema": 4.2706, "held_nce_hard": 2.1369, "held_acc_hard": 0.4922, "held_margin": 0.0341, "held_margin_hard": -0.0037, "held_cos_q": 0.1529, "scifact_ndcg": 0.5462, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7491446598308624e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.5856, "mrl_64": 0.6246, "mrl_128": 0.718, "mrl_256": 0.7833, "mrl_512": 0.8441, "mrl_768": 0.8708, "mrl_1024": 0.8867, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 47 |
+
{"step": 7055, "loss_ema": 4.311, "held_nce_hard": 2.0453, "held_acc_hard": 0.4766, "held_margin": 0.0367, "held_margin_hard": -0.0019, "held_cos_q": 0.1654, "scifact_ndcg": 0.5266, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7394619947837966e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.5127, "mrl_64": 0.55, "mrl_128": 0.641, "mrl_256": 0.7295, "mrl_512": 0.8057, "mrl_768": 0.8454, "mrl_1024": 0.8606, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 48 |
+
{"step": 7204, "loss_ema": 4.2348, "held_nce_hard": 1.9775, "held_acc_hard": 0.5312, "held_margin": 0.0369, "held_margin_hard": -0.0015, "held_cos_q": 0.1422, "scifact_ndcg": 0.5297, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.729627363757107e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.5479, "mrl_64": 0.6027, "mrl_128": 0.6929, "mrl_256": 0.7785, "mrl_512": 0.8405, "mrl_768": 0.8684, "mrl_1024": 0.8818, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 49 |
+
{"step": 7362, "loss_ema": 4.2316, "held_nce_hard": 1.8087, "held_acc_hard": 0.6055, "held_margin": 0.0367, "held_margin_hard": -0.0011, "held_cos_q": 0.1593, "scifact_ndcg": 0.5329, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7191763052464994e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.53, "mrl_32": 0.4803, "mrl_64": 0.565, "mrl_128": 0.6518, "mrl_256": 0.7476, "mrl_512": 0.8056, "mrl_768": 0.8381, "mrl_1024": 0.856, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 50 |
+
{"step": 7514, "loss_ema": 4.2471, "held_nce_hard": 2.081, "held_acc_hard": 0.5156, "held_margin": 0.0352, "held_margin_hard": -0.0017, "held_cos_q": 0.1916, "scifact_ndcg": 0.5327, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.7091002325285025e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.43, "mrl_32": 0.4001, "mrl_64": 0.4893, "mrl_128": 0.6095, "mrl_256": 0.7011, "mrl_512": 0.7855, "mrl_768": 0.8236, "mrl_1024": 0.8461, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 51 |
+
{"step": 7701, "loss_ema": 4.1317, "held_nce_hard": 1.8842, "held_acc_hard": 0.5469, "held_margin": 0.0366, "held_margin_hard": -0.0027, "held_cos_q": 0.1766, "scifact_ndcg": 0.5014, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.69667435532846e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.42, "mrl_32": 0.4361, "mrl_64": 0.5184, "mrl_128": 0.6091, "mrl_256": 0.7052, "mrl_512": 0.7856, "mrl_768": 0.8276, "mrl_1024": 0.852, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 52 |
+
{"step": 7867, "loss_ema": 4.1287, "held_nce_hard": 1.8449, "held_acc_hard": 0.5586, "held_margin": 0.0377, "held_margin_hard": -0.0021, "held_cos_q": 0.1882, "scifact_ndcg": 0.5085, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.685616288173841e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.41, "mrl_32": 0.4256, "mrl_64": 0.489, "mrl_128": 0.595, "mrl_256": 0.6867, "mrl_512": 0.7794, "mrl_768": 0.8178, "mrl_1024": 0.8444, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 53 |
+
{"step": 8033, "loss_ema": 4.1996, "held_nce_hard": 2.0445, "held_acc_hard": 0.457, "held_margin": 0.0379, "held_margin_hard": -0.0037, "held_cos_q": 0.1453, "scifact_ndcg": 0.4737, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.674532062142691e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.39, "mrl_32": 0.5062, "mrl_64": 0.5855, "mrl_128": 0.6925, "mrl_256": 0.7573, "mrl_512": 0.8309, "mrl_768": 0.8584, "mrl_1024": 0.8768, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 54 |
+
{"step": 8210, "loss_ema": 4.1497, "held_nce_hard": 2.2682, "held_acc_hard": 0.3359, "held_margin": 0.0391, "held_margin_hard": -0.0064, "held_cos_q": 0.1854, "scifact_ndcg": 0.523, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.662684312710866e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.45, "mrl_32": 0.4912, "mrl_64": 0.5322, "mrl_128": 0.607, "mrl_256": 0.7019, "mrl_512": 0.7895, "mrl_768": 0.8229, "mrl_1024": 0.846, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 55 |
+
{"step": 8373, "loss_ema": 4.1265, "held_nce_hard": 1.9926, "held_acc_hard": 0.4844, "held_margin": 0.0406, "held_margin_hard": -0.0028, "held_cos_q": 0.1659, "scifact_ndcg": 0.5214, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.651746983661085e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.39, "mrl_32": 0.4563, "mrl_64": 0.5343, "mrl_128": 0.6324, "mrl_256": 0.7218, "mrl_512": 0.8105, "mrl_768": 0.8442, "mrl_1024": 0.8643, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 56 |
+
{"step": 8554, "loss_ema": 4.1047, "held_nce_hard": 1.9631, "held_acc_hard": 0.5391, "held_margin": 0.0374, "held_margin_hard": -0.0021, "held_cos_q": 0.1683, "scifact_ndcg": 0.5111, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.639571639709856e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.4505, "mrl_64": 0.5313, "mrl_128": 0.6261, "mrl_256": 0.7135, "mrl_512": 0.7929, "mrl_768": 0.8261, "mrl_1024": 0.8491, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 57 |
+
{"step": 8720, "loss_ema": 4.224, "held_nce_hard": 2.163, "held_acc_hard": 0.3711, "held_margin": 0.0461, "held_margin_hard": -0.006, "held_cos_q": 0.2036, "scifact_ndcg": 0.4332, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.628377145393405e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.41, "mrl_32": 0.4408, "mrl_64": 0.5085, "mrl_128": 0.5965, "mrl_256": 0.6967, "mrl_512": 0.7787, "mrl_768": 0.8171, "mrl_1024": 0.8374, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 58 |
+
{"step": 8946, "loss_ema": 4.1759, "held_nce_hard": 2.1155, "held_acc_hard": 0.4492, "held_margin": 0.0386, "held_margin_hard": -0.0041, "held_cos_q": 0.158, "scifact_ndcg": 0.5006, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.613092780337287e-05, "sched": "decay", "bank_filled": 38400, "micro_q": 64, "accum": 2, "gb": 25.56, "mrl_32": 0.4937, "mrl_64": 0.5758, "mrl_128": 0.6646, "mrl_256": 0.752, "mrl_512": 0.8229, "mrl_768": 0.8554, "mrl_1024": 0.872, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 59 |
+
{"step": 9112, "loss_ema": 4.1013, "held_nce_hard": 1.9292, "held_acc_hard": 0.4492, "held_margin": 0.0457, "held_margin_hard": -0.0022, "held_cos_q": 0.2136, "scifact_ndcg": 0.4971, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.6018338735769244e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.42, "mrl_32": 0.4169, "mrl_64": 0.4697, "mrl_128": 0.5673, "mrl_256": 0.6621, "mrl_512": 0.7434, "mrl_768": 0.7868, "mrl_1024": 0.8129, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 60 |
+
{"step": 9293, "loss_ema": 4.1216, "held_nce_hard": 2.0107, "held_acc_hard": 0.4062, "held_margin": 0.0486, "held_margin_hard": -0.0027, "held_cos_q": 0.1879, "scifact_ndcg": 0.4668, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.589526119328662e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.4107, "mrl_64": 0.4959, "mrl_128": 0.5819, "mrl_256": 0.6982, "mrl_512": 0.7901, "mrl_768": 0.8199, "mrl_1024": 0.8458, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 61 |
+
{"step": 9468, "loss_ema": 4.1278, "held_nce_hard": 2.2729, "held_acc_hard": 0.3906, "held_margin": 0.0302, "held_margin_hard": -0.0034, "held_cos_q": 0.2152, "scifact_ndcg": 0.4959, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.577594892517248e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.4, "mrl_32": 0.3831, "mrl_64": 0.4446, "mrl_128": 0.5514, "mrl_256": 0.6545, "mrl_512": 0.7558, "mrl_768": 0.8011, "mrl_1024": 0.8294, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 62 |
+
{"step": 9652, "loss_ema": 4.0437, "held_nce_hard": 2.0809, "held_acc_hard": 0.4922, "held_margin": 0.033, "held_margin_hard": -0.0009, "held_cos_q": 0.1684, "scifact_ndcg": 0.5498, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.565016429324214e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.4589, "mrl_64": 0.5217, "mrl_128": 0.617, "mrl_256": 0.716, "mrl_512": 0.7928, "mrl_768": 0.8333, "mrl_1024": 0.8581, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 63 |
+
{"step": 9819, "loss_ema": 4.1213, "held_nce_hard": 2.1585, "held_acc_hard": 0.418, "held_margin": 0.0342, "held_margin_hard": -0.0054, "held_cos_q": 0.1966, "scifact_ndcg": 0.5045, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.553570028010989e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.4044, "mrl_64": 0.4704, "mrl_128": 0.5621, "mrl_256": 0.6677, "mrl_512": 0.7603, "mrl_768": 0.8036, "mrl_1024": 0.8312, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 64 |
+
{"step": 9989, "loss_ema": 4.2315, "held_nce_hard": 1.9398, "held_acc_hard": 0.5781, "held_margin": 0.0341, "held_margin_hard": -0.0019, "held_cos_q": 0.1343, "scifact_ndcg": 0.5312, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.541888373793438e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.45, "mrl_32": 0.5066, "mrl_64": 0.5982, "mrl_128": 0.7078, "mrl_256": 0.7707, "mrl_512": 0.8328, "mrl_768": 0.8704, "mrl_1024": 0.8904, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 65 |
+
{"step": 10178, "loss_ema": 4.0309, "held_nce_hard": 2.2194, "held_acc_hard": 0.3281, "held_margin": 0.0365, "held_margin_hard": -0.007, "held_cos_q": 0.1935, "scifact_ndcg": 0.4052, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.528865752039908e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.38, "mrl_32": 0.4604, "mrl_64": 0.5291, "mrl_128": 0.6125, "mrl_256": 0.7116, "mrl_512": 0.7961, "mrl_768": 0.8344, "mrl_1024": 0.8561, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 66 |
+
{"step": 10349, "loss_ema": 4.1496, "held_nce_hard": 1.8957, "held_acc_hard": 0.5039, "held_margin": 0.0391, "held_margin_hard": -0.0007, "held_cos_q": 0.1776, "scifact_ndcg": 0.5444, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.51705102915608e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.56, "mrl_32": 0.4757, "mrl_64": 0.5455, "mrl_128": 0.6357, "mrl_256": 0.73, "mrl_512": 0.7992, "mrl_768": 0.8375, "mrl_1024": 0.8621, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 67 |
+
{"step": 10534, "loss_ema": 4.0474, "held_nce_hard": 1.8761, "held_acc_hard": 0.6484, "held_margin": 0.0321, "held_margin_hard": -0.0018, "held_cos_q": 0.1347, "scifact_ndcg": 0.53, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.504234119137237e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.47, "mrl_32": 0.5686, "mrl_64": 0.6621, "mrl_128": 0.7305, "mrl_256": 0.802, "mrl_512": 0.8467, "mrl_768": 0.8752, "mrl_1024": 0.892, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 68 |
+
{"step": 10709, "loss_ema": 4.181, "held_nce_hard": 1.8906, "held_acc_hard": 0.5234, "held_margin": 0.0376, "held_margin_hard": -0.002, "held_cos_q": 0.1791, "scifact_ndcg": 0.5369, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.4920763573207433e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.54, "mrl_32": 0.416, "mrl_64": 0.5069, "mrl_128": 0.5819, "mrl_256": 0.6987, "mrl_512": 0.7745, "mrl_768": 0.8184, "mrl_1024": 0.8398, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 69 |
+
{"step": 10880, "loss_ema": 4.1126, "held_nce_hard": 2.0006, "held_acc_hard": 0.4844, "held_margin": 0.04, "held_margin_hard": -0.0022, "held_cos_q": 0.1428, "scifact_ndcg": 0.5431, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.480164617511281e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.46, "mrl_32": 0.6866, "mrl_64": 0.7354, "mrl_128": 0.7491, "mrl_256": 0.8065, "mrl_512": 0.8456, "mrl_768": 0.8734, "mrl_1024": 0.8906, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 70 |
+
{"step": 11065, "loss_ema": 4.0634, "held_nce_hard": 1.9167, "held_acc_hard": 0.5117, "held_margin": 0.0372, "held_margin_hard": -0.0013, "held_cos_q": 0.1827, "scifact_ndcg": 0.5226, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.4672418783853646e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.57, "mrl_32": 0.465, "mrl_64": 0.5153, "mrl_128": 0.6112, "mrl_256": 0.7156, "mrl_512": 0.7965, "mrl_768": 0.8315, "mrl_1024": 0.8538, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 71 |
+
{"step": 11237, "loss_ema": 4.0541, "held_nce_hard": 1.9562, "held_acc_hard": 0.4766, "held_margin": 0.0416, "held_margin_hard": -0.0016, "held_cos_q": 0.1692, "scifact_ndcg": 0.5221, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.4551935984870516e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.5, "mrl_32": 0.4294, "mrl_64": 0.532, "mrl_128": 0.625, "mrl_256": 0.7209, "mrl_512": 0.7908, "mrl_768": 0.832, "mrl_1024": 0.856, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 72 |
+
{"step": 11428, "loss_ema": 4.0626, "held_nce_hard": 2.0919, "held_acc_hard": 0.4805, "held_margin": 0.0334, "held_margin_hard": -0.0014, "held_cos_q": 0.1819, "scifact_ndcg": 0.539, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.4417761087204746e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.52, "mrl_32": 0.4021, "mrl_64": 0.4782, "mrl_128": 0.5964, "mrl_256": 0.7072, "mrl_512": 0.79, "mrl_768": 0.8322, "mrl_1024": 0.853, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 73 |
+
{"step": 11601, "loss_ema": 4.1584, "held_nce_hard": 2.0189, "held_acc_hard": 0.4375, "held_margin": 0.0408, "held_margin_hard": -0.0019, "held_cos_q": 0.1994, "scifact_ndcg": 0.5227, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.429588016960494e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.39, "mrl_32": 0.4174, "mrl_64": 0.4802, "mrl_128": 0.5802, "mrl_256": 0.6863, "mrl_512": 0.7695, "mrl_768": 0.813, "mrl_1024": 0.8412, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 74 |
+
{"step": 11776, "loss_ema": 4.1935, "held_nce_hard": 2.0724, "held_acc_hard": 0.4414, "held_margin": 0.0429, "held_margin_hard": -0.0058, "held_cos_q": 0.1507, "scifact_ndcg": 0.4714, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.417224807500746e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.44, "mrl_32": 0.5025, "mrl_64": 0.6107, "mrl_128": 0.7086, "mrl_256": 0.7764, "mrl_512": 0.8292, "mrl_768": 0.8561, "mrl_1024": 0.8728, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
| 75 |
+
{"step": 11964, "loss_ema": 4.1703, "held_nce_hard": 2.0075, "held_acc_hard": 0.4219, "held_margin": 0.0397, "held_margin_hard": -0.0029, "held_cos_q": 0.1971, "scifact_ndcg": 0.5103, "held_n": 512, "held_Hn": 15, "temp": 0.01999, "lr": 4.403904517584367e-05, "sched": "decay", "bank_filled": 65536, "micro_q": 64, "accum": 2, "gb": 25.49, "mrl_32": 0.4821, "mrl_64": 0.5646, "mrl_128": 0.6372, "mrl_256": 0.7172, "mrl_512": 0.7916, "mrl_768": 0.8281, "mrl_1024": 0.849, "held_srcs": {"stackex": 14, "nli": 20, "msmarco-bm25": 76, "hotpotqa": 27, "msmarco-dense": 63, "pairs": 5, "nq": 29, "quora": 14, "squad": 8}}
|
pooler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c98b4c71302ebe46941a490e4c1c5dc91b340b122efbd95b4fbf770f6c8ecaee
|
| 3 |
+
size 4233993
|
pooler_best.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca962b60fa85bd07e33163c3296c3b48258342bed0e18e43a294fa689779408a
|
| 3 |
+
size 4234437
|
sparse_opt.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7767ec0c56651097d3200ac2ffe6aec9d85cb23e166633e02c2fd5bb874ccc84
|
| 3 |
+
size 15032388235
|
training_state.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"global_step": 11964,
|
| 3 |
+
"rows_seen": 1531904,
|
| 4 |
+
"d_pos": {
|
| 5 |
+
"msmarco-bm25": [
|
| 6 |
+
11,
|
| 7 |
+
17096
|
| 8 |
+
],
|
| 9 |
+
"msmarco-dense": [
|
| 10 |
+
17,
|
| 11 |
+
21812
|
| 12 |
+
],
|
| 13 |
+
"nli": [
|
| 14 |
+
0,
|
| 15 |
+
107212
|
| 16 |
+
],
|
| 17 |
+
"hotpotqa": [
|
| 18 |
+
0,
|
| 19 |
+
14617
|
| 20 |
+
],
|
| 21 |
+
"quora": [
|
| 22 |
+
0,
|
| 23 |
+
373484
|
| 24 |
+
],
|
| 25 |
+
"squad": [
|
| 26 |
+
0,
|
| 27 |
+
30660
|
| 28 |
+
],
|
| 29 |
+
"pairs": [
|
| 30 |
+
0,
|
| 31 |
+
61365
|
| 32 |
+
],
|
| 33 |
+
"nq": [
|
| 34 |
+
0,
|
| 35 |
+
36226
|
| 36 |
+
],
|
| 37 |
+
"stackex": [
|
| 38 |
+
0,
|
| 39 |
+
61449
|
| 40 |
+
]
|
| 41 |
+
},
|
| 42 |
+
"elapsed_cum": 16247.754494428635,
|
| 43 |
+
"loss_ema": 4.170292145188064,
|
| 44 |
+
"total_tok": 3433251469,
|
| 45 |
+
"done": true,
|
| 46 |
+
"bank_filled": 65536,
|
| 47 |
+
"schedule_version": 1,
|
| 48 |
+
"decay_start": 2994,
|
| 49 |
+
"schedule_phase": "decay",
|
| 50 |
+
"best_margin": 0.0118215037509799,
|
| 51 |
+
"best_acc": 0.6640625,
|
| 52 |
+
"best_step": 536,
|
| 53 |
+
"stagnant_evals": 30,
|
| 54 |
+
"auto_finalized": true,
|
| 55 |
+
"stopped_early": true
|
| 56 |
+
}
|