Upload g2_delta_s1.json with huggingface_hub
Browse files- g2_delta_s1.json +94 -0
g2_delta_s1.json
ADDED
|
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"screen": "G2-MEM",
|
| 3 |
+
"arm": "delta",
|
| 4 |
+
"seeds": [
|
| 5 |
+
1
|
| 6 |
+
],
|
| 7 |
+
"law": {
|
| 8 |
+
"pairs": 4,
|
| 9 |
+
"key_len": 8,
|
| 10 |
+
"val_len": 8,
|
| 11 |
+
"delays": [
|
| 12 |
+
32,
|
| 13 |
+
128,
|
| 14 |
+
512,
|
| 15 |
+
1024
|
| 16 |
+
],
|
| 17 |
+
"d_model": 128,
|
| 18 |
+
"layers": 4,
|
| 19 |
+
"slots": 64,
|
| 20 |
+
"batch": 8,
|
| 21 |
+
"lr": 0.0003,
|
| 22 |
+
"clip": 1.0,
|
| 23 |
+
"warmup": 100,
|
| 24 |
+
"eval_events": 24,
|
| 25 |
+
"law_rev": "candm-v1",
|
| 26 |
+
"kappa": 8.0,
|
| 27 |
+
"echo_phase": 1500,
|
| 28 |
+
"echo_delay": 32
|
| 29 |
+
},
|
| 30 |
+
"off_law": {},
|
| 31 |
+
"device": "cuda",
|
| 32 |
+
"runs": [
|
| 33 |
+
{
|
| 34 |
+
"params": 536835,
|
| 35 |
+
"state_elems": 10176,
|
| 36 |
+
"history": [
|
| 37 |
+
{
|
| 38 |
+
"law_rev": "candm-v1",
|
| 39 |
+
"value_acc_d32": 0.0026,
|
| 40 |
+
"recall_exact_d32": 0.0,
|
| 41 |
+
"key_acc_d32": 0.0039,
|
| 42 |
+
"value_acc_d128": 0.0039,
|
| 43 |
+
"recall_exact_d128": 0.0,
|
| 44 |
+
"key_acc_d128": 0.0013,
|
| 45 |
+
"value_acc_d512": 0.0026,
|
| 46 |
+
"recall_exact_d512": 0.0,
|
| 47 |
+
"key_acc_d512": 0.0039,
|
| 48 |
+
"value_acc_d1024": 0.0013,
|
| 49 |
+
"recall_exact_d1024": 0.0,
|
| 50 |
+
"key_acc_d1024": 0.0026,
|
| 51 |
+
"alpha_mean": 0.9991,
|
| 52 |
+
"beta_mean": 0.0457,
|
| 53 |
+
"m_fro": 3.551,
|
| 54 |
+
"p_fro": 4.899,
|
| 55 |
+
"step": 0,
|
| 56 |
+
"train_bpb": 15.6033,
|
| 57 |
+
"grad_norm": 16.33498192,
|
| 58 |
+
"pairs": 4,
|
| 59 |
+
"lr": 0.0003,
|
| 60 |
+
"eta": 1.0,
|
| 61 |
+
"s_per_step": 4.1729
|
| 62 |
+
}
|
| 63 |
+
],
|
| 64 |
+
"final": {
|
| 65 |
+
"law_rev": "candm-v1",
|
| 66 |
+
"value_acc_d32": 0.0026,
|
| 67 |
+
"recall_exact_d32": 0.0,
|
| 68 |
+
"key_acc_d32": 0.0039,
|
| 69 |
+
"value_acc_d128": 0.0039,
|
| 70 |
+
"recall_exact_d128": 0.0,
|
| 71 |
+
"key_acc_d128": 0.0013,
|
| 72 |
+
"value_acc_d512": 0.0026,
|
| 73 |
+
"recall_exact_d512": 0.0,
|
| 74 |
+
"key_acc_d512": 0.0039,
|
| 75 |
+
"value_acc_d1024": 0.0013,
|
| 76 |
+
"recall_exact_d1024": 0.0,
|
| 77 |
+
"key_acc_d1024": 0.0026,
|
| 78 |
+
"alpha_mean": 0.9991,
|
| 79 |
+
"beta_mean": 0.0457,
|
| 80 |
+
"m_fro": 3.551,
|
| 81 |
+
"p_fro": 4.899,
|
| 82 |
+
"step": 0,
|
| 83 |
+
"train_bpb": 15.6033,
|
| 84 |
+
"grad_norm": 16.33498192,
|
| 85 |
+
"pairs": 4,
|
| 86 |
+
"lr": 0.0003,
|
| 87 |
+
"eta": 1.0,
|
| 88 |
+
"s_per_step": 4.1729
|
| 89 |
+
}
|
| 90 |
+
}
|
| 91 |
+
],
|
| 92 |
+
"partial": true,
|
| 93 |
+
"result_file": "g2_delta_s1.json"
|
| 94 |
+
}
|