Upload checkpoint-500/trainer_state.json with huggingface_hub
Browse files
checkpoint-500/trainer_state.json
ADDED
|
@@ -0,0 +1,815 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7111111111111111,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.014222222222222223,
|
| 13 |
+
"grad_norm": 37.622314591135336,
|
| 14 |
+
"learning_rate": 2.8169014084507043e-07,
|
| 15 |
+
"logits/chosen": 0.22460369765758514,
|
| 16 |
+
"logits/rejected": 0.22931505739688873,
|
| 17 |
+
"logps/chosen": -169.9998321533203,
|
| 18 |
+
"logps/rejected": -165.81625366210938,
|
| 19 |
+
"loss": 0.692,
|
| 20 |
+
"rewards/accuracies": 0.4257812798023224,
|
| 21 |
+
"rewards/chosen": -0.00582541897892952,
|
| 22 |
+
"rewards/margins": 0.0030124844051897526,
|
| 23 |
+
"rewards/rejected": -0.008837902918457985,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.028444444444444446,
|
| 28 |
+
"grad_norm": 34.645567235965046,
|
| 29 |
+
"learning_rate": 5.633802816901409e-07,
|
| 30 |
+
"logits/chosen": 0.21702095866203308,
|
| 31 |
+
"logits/rejected": 0.22135533392429352,
|
| 32 |
+
"logps/chosen": -158.21507263183594,
|
| 33 |
+
"logps/rejected": -159.20578002929688,
|
| 34 |
+
"loss": 0.6871,
|
| 35 |
+
"rewards/accuracies": 0.5625,
|
| 36 |
+
"rewards/chosen": -0.10022798180580139,
|
| 37 |
+
"rewards/margins": 0.014327095821499825,
|
| 38 |
+
"rewards/rejected": -0.11455506831407547,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.042666666666666665,
|
| 43 |
+
"grad_norm": 36.11162335188131,
|
| 44 |
+
"learning_rate": 8.450704225352112e-07,
|
| 45 |
+
"logits/chosen": 0.22448155283927917,
|
| 46 |
+
"logits/rejected": 0.22727997601032257,
|
| 47 |
+
"logps/chosen": -176.5776824951172,
|
| 48 |
+
"logps/rejected": -172.39144897460938,
|
| 49 |
+
"loss": 0.6743,
|
| 50 |
+
"rewards/accuracies": 0.5765625238418579,
|
| 51 |
+
"rewards/chosen": -0.4727466404438019,
|
| 52 |
+
"rewards/margins": 0.06264183670282364,
|
| 53 |
+
"rewards/rejected": -0.5353884696960449,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.05688888888888889,
|
| 58 |
+
"grad_norm": 31.16390845270907,
|
| 59 |
+
"learning_rate": 1.1267605633802817e-06,
|
| 60 |
+
"logits/chosen": 0.23577681183815002,
|
| 61 |
+
"logits/rejected": 0.23861564695835114,
|
| 62 |
+
"logps/chosen": -170.12828063964844,
|
| 63 |
+
"logps/rejected": -173.585205078125,
|
| 64 |
+
"loss": 0.6274,
|
| 65 |
+
"rewards/accuracies": 0.649218738079071,
|
| 66 |
+
"rewards/chosen": -0.5824733376502991,
|
| 67 |
+
"rewards/margins": 0.20194238424301147,
|
| 68 |
+
"rewards/rejected": -0.7844158411026001,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.07111111111111111,
|
| 73 |
+
"grad_norm": 31.43717823219279,
|
| 74 |
+
"learning_rate": 1.408450704225352e-06,
|
| 75 |
+
"logits/chosen": 0.2460279017686844,
|
| 76 |
+
"logits/rejected": 0.24856708943843842,
|
| 77 |
+
"logps/chosen": -164.169189453125,
|
| 78 |
+
"logps/rejected": -168.471435546875,
|
| 79 |
+
"loss": 0.5882,
|
| 80 |
+
"rewards/accuracies": 0.7109375,
|
| 81 |
+
"rewards/chosen": -0.5867726802825928,
|
| 82 |
+
"rewards/margins": 0.354371041059494,
|
| 83 |
+
"rewards/rejected": -0.9411437511444092,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.08533333333333333,
|
| 88 |
+
"grad_norm": 29.466665588707553,
|
| 89 |
+
"learning_rate": 1.6901408450704225e-06,
|
| 90 |
+
"logits/chosen": 0.26712918281555176,
|
| 91 |
+
"logits/rejected": 0.27090156078338623,
|
| 92 |
+
"logps/chosen": -168.39569091796875,
|
| 93 |
+
"logps/rejected": -174.5313262939453,
|
| 94 |
+
"loss": 0.5673,
|
| 95 |
+
"rewards/accuracies": 0.6929687857627869,
|
| 96 |
+
"rewards/chosen": -0.44874832034111023,
|
| 97 |
+
"rewards/margins": 0.5710285305976868,
|
| 98 |
+
"rewards/rejected": -1.0197768211364746,
|
| 99 |
+
"step": 60
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.09955555555555555,
|
| 103 |
+
"grad_norm": 29.29672362284883,
|
| 104 |
+
"learning_rate": 1.971830985915493e-06,
|
| 105 |
+
"logits/chosen": 0.26626870036125183,
|
| 106 |
+
"logits/rejected": 0.27328580617904663,
|
| 107 |
+
"logps/chosen": -170.83888244628906,
|
| 108 |
+
"logps/rejected": -175.48245239257812,
|
| 109 |
+
"loss": 0.5323,
|
| 110 |
+
"rewards/accuracies": 0.7390625476837158,
|
| 111 |
+
"rewards/chosen": -0.6967486143112183,
|
| 112 |
+
"rewards/margins": 0.8345546126365662,
|
| 113 |
+
"rewards/rejected": -1.5313031673431396,
|
| 114 |
+
"step": 70
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.11377777777777778,
|
| 118 |
+
"grad_norm": 32.075198095164076,
|
| 119 |
+
"learning_rate": 1.9989994283927284e-06,
|
| 120 |
+
"logits/chosen": 0.28503507375717163,
|
| 121 |
+
"logits/rejected": 0.2870316207408905,
|
| 122 |
+
"logps/chosen": -164.9480438232422,
|
| 123 |
+
"logps/rejected": -172.48333740234375,
|
| 124 |
+
"loss": 0.5101,
|
| 125 |
+
"rewards/accuracies": 0.7515625357627869,
|
| 126 |
+
"rewards/chosen": -0.805090606212616,
|
| 127 |
+
"rewards/margins": 0.9741694331169128,
|
| 128 |
+
"rewards/rejected": -1.7792601585388184,
|
| 129 |
+
"step": 80
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.128,
|
| 133 |
+
"grad_norm": 28.794301270291456,
|
| 134 |
+
"learning_rate": 1.9955432328988433e-06,
|
| 135 |
+
"logits/chosen": 0.2928016185760498,
|
| 136 |
+
"logits/rejected": 0.2973910868167877,
|
| 137 |
+
"logps/chosen": -172.90721130371094,
|
| 138 |
+
"logps/rejected": -185.6675262451172,
|
| 139 |
+
"loss": 0.4841,
|
| 140 |
+
"rewards/accuracies": 0.7625000476837158,
|
| 141 |
+
"rewards/chosen": -1.0631386041641235,
|
| 142 |
+
"rewards/margins": 1.2136086225509644,
|
| 143 |
+
"rewards/rejected": -2.276747226715088,
|
| 144 |
+
"step": 90
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.14222222222222222,
|
| 148 |
+
"grad_norm": 30.037256928052916,
|
| 149 |
+
"learning_rate": 1.9896275971073322e-06,
|
| 150 |
+
"logits/chosen": 0.3249911367893219,
|
| 151 |
+
"logits/rejected": 0.3287336230278015,
|
| 152 |
+
"logps/chosen": -170.0114288330078,
|
| 153 |
+
"logps/rejected": -185.2743377685547,
|
| 154 |
+
"loss": 0.4993,
|
| 155 |
+
"rewards/accuracies": 0.7625000476837158,
|
| 156 |
+
"rewards/chosen": -0.91056889295578,
|
| 157 |
+
"rewards/margins": 1.2012264728546143,
|
| 158 |
+
"rewards/rejected": -2.111795425415039,
|
| 159 |
+
"step": 100
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.15644444444444444,
|
| 163 |
+
"grad_norm": 23.930285809273254,
|
| 164 |
+
"learning_rate": 1.9812671353037137e-06,
|
| 165 |
+
"logits/chosen": 0.3139130473136902,
|
| 166 |
+
"logits/rejected": 0.319477915763855,
|
| 167 |
+
"logps/chosen": -169.1543426513672,
|
| 168 |
+
"logps/rejected": -184.03173828125,
|
| 169 |
+
"loss": 0.4571,
|
| 170 |
+
"rewards/accuracies": 0.80078125,
|
| 171 |
+
"rewards/chosen": -0.965703547000885,
|
| 172 |
+
"rewards/margins": 1.4910372495651245,
|
| 173 |
+
"rewards/rejected": -2.4567408561706543,
|
| 174 |
+
"step": 110
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.17066666666666666,
|
| 178 |
+
"grad_norm": 28.491945703626907,
|
| 179 |
+
"learning_rate": 1.9704825015952003e-06,
|
| 180 |
+
"logits/chosen": 0.3043367862701416,
|
| 181 |
+
"logits/rejected": 0.31123000383377075,
|
| 182 |
+
"logps/chosen": -171.40054321289062,
|
| 183 |
+
"logps/rejected": -187.75230407714844,
|
| 184 |
+
"loss": 0.4555,
|
| 185 |
+
"rewards/accuracies": 0.803906261920929,
|
| 186 |
+
"rewards/chosen": -1.192291259765625,
|
| 187 |
+
"rewards/margins": 1.632858395576477,
|
| 188 |
+
"rewards/rejected": -2.8251495361328125,
|
| 189 |
+
"step": 120
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.18488888888888888,
|
| 193 |
+
"grad_norm": 30.70726336010305,
|
| 194 |
+
"learning_rate": 1.9573003388857475e-06,
|
| 195 |
+
"logits/chosen": 0.3287965655326843,
|
| 196 |
+
"logits/rejected": 0.3384350836277008,
|
| 197 |
+
"logps/chosen": -178.2093505859375,
|
| 198 |
+
"logps/rejected": -200.10389709472656,
|
| 199 |
+
"loss": 0.44,
|
| 200 |
+
"rewards/accuracies": 0.8039062023162842,
|
| 201 |
+
"rewards/chosen": -1.2602012157440186,
|
| 202 |
+
"rewards/margins": 1.8096774816513062,
|
| 203 |
+
"rewards/rejected": -3.069878578186035,
|
| 204 |
+
"step": 130
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.1991111111111111,
|
| 208 |
+
"grad_norm": 30.33953791143264,
|
| 209 |
+
"learning_rate": 1.941753213056078e-06,
|
| 210 |
+
"logits/chosen": 0.3240712583065033,
|
| 211 |
+
"logits/rejected": 0.3240274488925934,
|
| 212 |
+
"logps/chosen": -178.25926208496094,
|
| 213 |
+
"logps/rejected": -204.4981689453125,
|
| 214 |
+
"loss": 0.4325,
|
| 215 |
+
"rewards/accuracies": 0.8109375238418579,
|
| 216 |
+
"rewards/chosen": -1.5995012521743774,
|
| 217 |
+
"rewards/margins": 2.0143957138061523,
|
| 218 |
+
"rewards/rejected": -3.6138970851898193,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.21333333333333335,
|
| 223 |
+
"grad_norm": 26.883589478358978,
|
| 224 |
+
"learning_rate": 1.9238795325112867e-06,
|
| 225 |
+
"logits/chosen": 0.32417288422584534,
|
| 226 |
+
"logits/rejected": 0.324298620223999,
|
| 227 |
+
"logps/chosen": -170.05540466308594,
|
| 228 |
+
"logps/rejected": -193.8038330078125,
|
| 229 |
+
"loss": 0.4386,
|
| 230 |
+
"rewards/accuracies": 0.8093750476837158,
|
| 231 |
+
"rewards/chosen": -1.4086307287216187,
|
| 232 |
+
"rewards/margins": 2.022059679031372,
|
| 233 |
+
"rewards/rejected": -3.4306905269622803,
|
| 234 |
+
"step": 150
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.22755555555555557,
|
| 238 |
+
"grad_norm": 23.96004108690596,
|
| 239 |
+
"learning_rate": 1.9037234532947735e-06,
|
| 240 |
+
"logits/chosen": 0.3645060062408447,
|
| 241 |
+
"logits/rejected": 0.3710418939590454,
|
| 242 |
+
"logps/chosen": -173.90704345703125,
|
| 243 |
+
"logps/rejected": -191.5853729248047,
|
| 244 |
+
"loss": 0.4164,
|
| 245 |
+
"rewards/accuracies": 0.805468738079071,
|
| 246 |
+
"rewards/chosen": -1.3054120540618896,
|
| 247 |
+
"rewards/margins": 2.069995164871216,
|
| 248 |
+
"rewards/rejected": -3.3754072189331055,
|
| 249 |
+
"step": 160
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.24177777777777779,
|
| 253 |
+
"grad_norm": 23.17426643588145,
|
| 254 |
+
"learning_rate": 1.8813347700029242e-06,
|
| 255 |
+
"logits/chosen": 0.39987513422966003,
|
| 256 |
+
"logits/rejected": 0.40979668498039246,
|
| 257 |
+
"logps/chosen": -178.6532440185547,
|
| 258 |
+
"logps/rejected": -206.7698974609375,
|
| 259 |
+
"loss": 0.4369,
|
| 260 |
+
"rewards/accuracies": 0.8195312023162842,
|
| 261 |
+
"rewards/chosen": -1.8244224786758423,
|
| 262 |
+
"rewards/margins": 2.2108898162841797,
|
| 263 |
+
"rewards/rejected": -4.035312175750732,
|
| 264 |
+
"step": 170
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.256,
|
| 268 |
+
"grad_norm": 24.91179023868515,
|
| 269 |
+
"learning_rate": 1.8567687927700252e-06,
|
| 270 |
+
"logits/chosen": 0.4278528094291687,
|
| 271 |
+
"logits/rejected": 0.4357163906097412,
|
| 272 |
+
"logps/chosen": -167.036376953125,
|
| 273 |
+
"logps/rejected": -198.56353759765625,
|
| 274 |
+
"loss": 0.4067,
|
| 275 |
+
"rewards/accuracies": 0.813281238079071,
|
| 276 |
+
"rewards/chosen": -1.3153953552246094,
|
| 277 |
+
"rewards/margins": 2.0069518089294434,
|
| 278 |
+
"rewards/rejected": -3.3223471641540527,
|
| 279 |
+
"step": 180
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.2702222222222222,
|
| 283 |
+
"grad_norm": 25.261537622265273,
|
| 284 |
+
"learning_rate": 1.8300862106273111e-06,
|
| 285 |
+
"logits/chosen": 0.3883928060531616,
|
| 286 |
+
"logits/rejected": 0.39530688524246216,
|
| 287 |
+
"logps/chosen": -176.3590087890625,
|
| 288 |
+
"logps/rejected": -208.9456024169922,
|
| 289 |
+
"loss": 0.3584,
|
| 290 |
+
"rewards/accuracies": 0.8414062261581421,
|
| 291 |
+
"rewards/chosen": -2.198540687561035,
|
| 292 |
+
"rewards/margins": 2.693372964859009,
|
| 293 |
+
"rewards/rejected": -4.891913890838623,
|
| 294 |
+
"step": 190
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.28444444444444444,
|
| 298 |
+
"grad_norm": 25.275402327404514,
|
| 299 |
+
"learning_rate": 1.8013529415737175e-06,
|
| 300 |
+
"logits/chosen": 0.3748525083065033,
|
| 301 |
+
"logits/rejected": 0.3877173960208893,
|
| 302 |
+
"logps/chosen": -173.32003784179688,
|
| 303 |
+
"logps/rejected": -214.57608032226562,
|
| 304 |
+
"loss": 0.4065,
|
| 305 |
+
"rewards/accuracies": 0.835156261920929,
|
| 306 |
+
"rewards/chosen": -1.989816665649414,
|
| 307 |
+
"rewards/margins": 2.6171391010284424,
|
| 308 |
+
"rewards/rejected": -4.606955528259277,
|
| 309 |
+
"step": 200
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.28444444444444444,
|
| 313 |
+
"eval_logits/chosen": 0.3876859247684479,
|
| 314 |
+
"eval_logits/rejected": 0.3957110643386841,
|
| 315 |
+
"eval_logps/chosen": -171.17259216308594,
|
| 316 |
+
"eval_logps/rejected": -202.42398071289062,
|
| 317 |
+
"eval_loss": 0.3801611363887787,
|
| 318 |
+
"eval_rewards/accuracies": 0.8434999585151672,
|
| 319 |
+
"eval_rewards/chosen": -1.2062910795211792,
|
| 320 |
+
"eval_rewards/margins": 2.4256691932678223,
|
| 321 |
+
"eval_rewards/rejected": -3.631960153579712,
|
| 322 |
+
"eval_runtime": 557.8889,
|
| 323 |
+
"eval_samples_per_second": 17.925,
|
| 324 |
+
"eval_steps_per_second": 2.241,
|
| 325 |
+
"step": 200
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.2986666666666667,
|
| 329 |
+
"grad_norm": 24.072532739733493,
|
| 330 |
+
"learning_rate": 1.7706399697287258e-06,
|
| 331 |
+
"logits/chosen": 0.4026853144168854,
|
| 332 |
+
"logits/rejected": 0.412026047706604,
|
| 333 |
+
"logps/chosen": -167.03302001953125,
|
| 334 |
+
"logps/rejected": -200.06622314453125,
|
| 335 |
+
"loss": 0.3903,
|
| 336 |
+
"rewards/accuracies": 0.83203125,
|
| 337 |
+
"rewards/chosen": -1.3341785669326782,
|
| 338 |
+
"rewards/margins": 2.3634612560272217,
|
| 339 |
+
"rewards/rejected": -3.6976399421691895,
|
| 340 |
+
"step": 210
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.3128888888888889,
|
| 344 |
+
"grad_norm": 24.07534365113526,
|
| 345 |
+
"learning_rate": 1.7380231699696077e-06,
|
| 346 |
+
"logits/chosen": 0.4304381012916565,
|
| 347 |
+
"logits/rejected": 0.44892990589141846,
|
| 348 |
+
"logps/chosen": -181.48057556152344,
|
| 349 |
+
"logps/rejected": -214.39857482910156,
|
| 350 |
+
"loss": 0.3542,
|
| 351 |
+
"rewards/accuracies": 0.854687511920929,
|
| 352 |
+
"rewards/chosen": -2.030439853668213,
|
| 353 |
+
"rewards/margins": 2.744889736175537,
|
| 354 |
+
"rewards/rejected": -4.775330066680908,
|
| 355 |
+
"step": 220
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.32711111111111113,
|
| 359 |
+
"grad_norm": 22.108293550882866,
|
| 360 |
+
"learning_rate": 1.703583120486297e-06,
|
| 361 |
+
"logits/chosen": 0.4489140510559082,
|
| 362 |
+
"logits/rejected": 0.45542919635772705,
|
| 363 |
+
"logps/chosen": -184.29531860351562,
|
| 364 |
+
"logps/rejected": -207.33291625976562,
|
| 365 |
+
"loss": 0.3715,
|
| 366 |
+
"rewards/accuracies": 0.8335937857627869,
|
| 367 |
+
"rewards/chosen": -1.929568886756897,
|
| 368 |
+
"rewards/margins": 2.5038599967956543,
|
| 369 |
+
"rewards/rejected": -4.433429718017578,
|
| 370 |
+
"step": 230
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.3413333333333333,
|
| 374 |
+
"grad_norm": 23.925638393745633,
|
| 375 |
+
"learning_rate": 1.6674049037169562e-06,
|
| 376 |
+
"logits/chosen": 0.478533536195755,
|
| 377 |
+
"logits/rejected": 0.4951012134552002,
|
| 378 |
+
"logps/chosen": -171.228515625,
|
| 379 |
+
"logps/rejected": -202.62612915039062,
|
| 380 |
+
"loss": 0.3597,
|
| 381 |
+
"rewards/accuracies": 0.859375,
|
| 382 |
+
"rewards/chosen": -1.6307088136672974,
|
| 383 |
+
"rewards/margins": 2.4791078567504883,
|
| 384 |
+
"rewards/rejected": -4.109816551208496,
|
| 385 |
+
"step": 240
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.35555555555555557,
|
| 389 |
+
"grad_norm": 30.54556364244028,
|
| 390 |
+
"learning_rate": 1.629577896156024e-06,
|
| 391 |
+
"logits/chosen": 0.48382341861724854,
|
| 392 |
+
"logits/rejected": 0.49492910504341125,
|
| 393 |
+
"logps/chosen": -177.4665069580078,
|
| 394 |
+
"logps/rejected": -221.9366912841797,
|
| 395 |
+
"loss": 0.3629,
|
| 396 |
+
"rewards/accuracies": 0.8453124761581421,
|
| 397 |
+
"rewards/chosen": -2.1776082515716553,
|
| 398 |
+
"rewards/margins": 2.9196786880493164,
|
| 399 |
+
"rewards/rejected": -5.097287178039551,
|
| 400 |
+
"step": 250
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.36977777777777776,
|
| 404 |
+
"grad_norm": 25.498807701982457,
|
| 405 |
+
"learning_rate": 1.5901955475540083e-06,
|
| 406 |
+
"logits/chosen": 0.43249523639678955,
|
| 407 |
+
"logits/rejected": 0.4461674690246582,
|
| 408 |
+
"logps/chosen": -175.78306579589844,
|
| 409 |
+
"logps/rejected": -220.20843505859375,
|
| 410 |
+
"loss": 0.3714,
|
| 411 |
+
"rewards/accuracies": 0.8617187738418579,
|
| 412 |
+
"rewards/chosen": -2.1237916946411133,
|
| 413 |
+
"rewards/margins": 3.081571340560913,
|
| 414 |
+
"rewards/rejected": -5.2053632736206055,
|
| 415 |
+
"step": 260
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.384,
|
| 419 |
+
"grad_norm": 21.98152324484991,
|
| 420 |
+
"learning_rate": 1.5493551500545005e-06,
|
| 421 |
+
"logits/chosen": 0.44403743743896484,
|
| 422 |
+
"logits/rejected": 0.45100969076156616,
|
| 423 |
+
"logps/chosen": -171.3573455810547,
|
| 424 |
+
"logps/rejected": -205.21868896484375,
|
| 425 |
+
"loss": 0.3192,
|
| 426 |
+
"rewards/accuracies": 0.867968738079071,
|
| 427 |
+
"rewards/chosen": -1.330743670463562,
|
| 428 |
+
"rewards/margins": 3.0044314861297607,
|
| 429 |
+
"rewards/rejected": -4.335175514221191,
|
| 430 |
+
"step": 270
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.3982222222222222,
|
| 434 |
+
"grad_norm": 26.597345332666173,
|
| 435 |
+
"learning_rate": 1.5071575978387502e-06,
|
| 436 |
+
"logits/chosen": 0.38800927996635437,
|
| 437 |
+
"logits/rejected": 0.4022008776664734,
|
| 438 |
+
"logps/chosen": -184.14996337890625,
|
| 439 |
+
"logps/rejected": -209.67626953125,
|
| 440 |
+
"loss": 0.3398,
|
| 441 |
+
"rewards/accuracies": 0.8523437976837158,
|
| 442 |
+
"rewards/chosen": -1.988921046257019,
|
| 443 |
+
"rewards/margins": 2.9822380542755127,
|
| 444 |
+
"rewards/rejected": -4.9711594581604,
|
| 445 |
+
"step": 280
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.41244444444444445,
|
| 449 |
+
"grad_norm": 21.215110311422276,
|
| 450 |
+
"learning_rate": 1.4637071378715804e-06,
|
| 451 |
+
"logits/chosen": 0.3857256770133972,
|
| 452 |
+
"logits/rejected": 0.39900335669517517,
|
| 453 |
+
"logps/chosen": -172.09994506835938,
|
| 454 |
+
"logps/rejected": -217.18148803710938,
|
| 455 |
+
"loss": 0.3144,
|
| 456 |
+
"rewards/accuracies": 0.8742187023162842,
|
| 457 |
+
"rewards/chosen": -2.315093755722046,
|
| 458 |
+
"rewards/margins": 3.1719839572906494,
|
| 459 |
+
"rewards/rejected": -5.487077236175537,
|
| 460 |
+
"step": 290
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.4266666666666667,
|
| 464 |
+
"grad_norm": 19.597986581568264,
|
| 465 |
+
"learning_rate": 1.419111112364422e-06,
|
| 466 |
+
"logits/chosen": 0.44452688097953796,
|
| 467 |
+
"logits/rejected": 0.45345011353492737,
|
| 468 |
+
"logps/chosen": -184.4666290283203,
|
| 469 |
+
"logps/rejected": -210.34649658203125,
|
| 470 |
+
"loss": 0.2989,
|
| 471 |
+
"rewards/accuracies": 0.8828125,
|
| 472 |
+
"rewards/chosen": -1.6965656280517578,
|
| 473 |
+
"rewards/margins": 3.0673210620880127,
|
| 474 |
+
"rewards/rejected": -4.763886451721191,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.4408888888888889,
|
| 479 |
+
"grad_norm": 22.36768683602415,
|
| 480 |
+
"learning_rate": 1.3734796935916885e-06,
|
| 481 |
+
"logits/chosen": 0.4570707380771637,
|
| 482 |
+
"logits/rejected": 0.45680785179138184,
|
| 483 |
+
"logps/chosen": -173.09716796875,
|
| 484 |
+
"logps/rejected": -211.5676727294922,
|
| 485 |
+
"loss": 0.3123,
|
| 486 |
+
"rewards/accuracies": 0.8765625357627869,
|
| 487 |
+
"rewards/chosen": -2.017775297164917,
|
| 488 |
+
"rewards/margins": 3.1919538974761963,
|
| 489 |
+
"rewards/rejected": -5.2097296714782715,
|
| 490 |
+
"step": 310
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.45511111111111113,
|
| 494 |
+
"grad_norm": 26.62400911934529,
|
| 495 |
+
"learning_rate": 1.3269256117156266e-06,
|
| 496 |
+
"logits/chosen": 0.4452747702598572,
|
| 497 |
+
"logits/rejected": 0.46178725361824036,
|
| 498 |
+
"logps/chosen": -168.95912170410156,
|
| 499 |
+
"logps/rejected": -219.78575134277344,
|
| 500 |
+
"loss": 0.2951,
|
| 501 |
+
"rewards/accuracies": 0.889843761920929,
|
| 502 |
+
"rewards/chosen": -1.6834113597869873,
|
| 503 |
+
"rewards/margins": 3.441744804382324,
|
| 504 |
+
"rewards/rejected": -5.125155925750732,
|
| 505 |
+
"step": 320
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.4693333333333333,
|
| 509 |
+
"grad_norm": 22.453155219367574,
|
| 510 |
+
"learning_rate": 1.279563876292025e-06,
|
| 511 |
+
"logits/chosen": 0.5002372860908508,
|
| 512 |
+
"logits/rejected": 0.5076306462287903,
|
| 513 |
+
"logps/chosen": -175.60882568359375,
|
| 514 |
+
"logps/rejected": -206.75450134277344,
|
| 515 |
+
"loss": 0.3062,
|
| 516 |
+
"rewards/accuracies": 0.8726562857627869,
|
| 517 |
+
"rewards/chosen": -1.6064902544021606,
|
| 518 |
+
"rewards/margins": 3.4437012672424316,
|
| 519 |
+
"rewards/rejected": -5.050191402435303,
|
| 520 |
+
"step": 330
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.48355555555555557,
|
| 524 |
+
"grad_norm": 23.874159638189937,
|
| 525 |
+
"learning_rate": 1.231511492144801e-06,
|
| 526 |
+
"logits/chosen": 0.5309115052223206,
|
| 527 |
+
"logits/rejected": 0.5370917320251465,
|
| 528 |
+
"logps/chosen": -169.13853454589844,
|
| 529 |
+
"logps/rejected": -210.06500244140625,
|
| 530 |
+
"loss": 0.3079,
|
| 531 |
+
"rewards/accuracies": 0.87890625,
|
| 532 |
+
"rewards/chosen": -1.7956911325454712,
|
| 533 |
+
"rewards/margins": 3.3945703506469727,
|
| 534 |
+
"rewards/rejected": -5.1902618408203125,
|
| 535 |
+
"step": 340
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.49777777777777776,
|
| 539 |
+
"grad_norm": 25.16314769602878,
|
| 540 |
+
"learning_rate": 1.1828871703113684e-06,
|
| 541 |
+
"logits/chosen": 0.546948254108429,
|
| 542 |
+
"logits/rejected": 0.5572729110717773,
|
| 543 |
+
"logps/chosen": -176.99685668945312,
|
| 544 |
+
"logps/rejected": -211.9476318359375,
|
| 545 |
+
"loss": 0.2934,
|
| 546 |
+
"rewards/accuracies": 0.8859375715255737,
|
| 547 |
+
"rewards/chosen": -1.864843487739563,
|
| 548 |
+
"rewards/margins": 3.38896107673645,
|
| 549 |
+
"rewards/rejected": -5.253804683685303,
|
| 550 |
+
"step": 350
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.512,
|
| 554 |
+
"grad_norm": 21.59285813836473,
|
| 555 |
+
"learning_rate": 1.133811034772897e-06,
|
| 556 |
+
"logits/chosen": 0.5784353613853455,
|
| 557 |
+
"logits/rejected": 0.5889633893966675,
|
| 558 |
+
"logps/chosen": -174.65452575683594,
|
| 559 |
+
"logps/rejected": -209.51783752441406,
|
| 560 |
+
"loss": 0.2762,
|
| 561 |
+
"rewards/accuracies": 0.8890625834465027,
|
| 562 |
+
"rewards/chosen": -1.494131088256836,
|
| 563 |
+
"rewards/margins": 3.530411958694458,
|
| 564 |
+
"rewards/rejected": -5.024543285369873,
|
| 565 |
+
"step": 360
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.5262222222222223,
|
| 569 |
+
"grad_norm": 25.616067480178305,
|
| 570 |
+
"learning_rate": 1.0844043256939583e-06,
|
| 571 |
+
"logits/chosen": 0.5757034420967102,
|
| 572 |
+
"logits/rejected": 0.5903257131576538,
|
| 573 |
+
"logps/chosen": -176.61965942382812,
|
| 574 |
+
"logps/rejected": -213.92478942871094,
|
| 575 |
+
"loss": 0.286,
|
| 576 |
+
"rewards/accuracies": 0.9007812142372131,
|
| 577 |
+
"rewards/chosen": -1.30091392993927,
|
| 578 |
+
"rewards/margins": 3.603193998336792,
|
| 579 |
+
"rewards/rejected": -4.904108047485352,
|
| 580 |
+
"step": 370
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.5404444444444444,
|
| 584 |
+
"grad_norm": 22.42460837898434,
|
| 585 |
+
"learning_rate": 1.0347890999046998e-06,
|
| 586 |
+
"logits/chosen": 0.5664804577827454,
|
| 587 |
+
"logits/rejected": 0.5777785778045654,
|
| 588 |
+
"logps/chosen": -170.44606018066406,
|
| 589 |
+
"logps/rejected": -217.41619873046875,
|
| 590 |
+
"loss": 0.2698,
|
| 591 |
+
"rewards/accuracies": 0.8921874761581421,
|
| 592 |
+
"rewards/chosen": -1.511139988899231,
|
| 593 |
+
"rewards/margins": 3.564502239227295,
|
| 594 |
+
"rewards/rejected": -5.075642108917236,
|
| 595 |
+
"step": 380
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.5546666666666666,
|
| 599 |
+
"grad_norm": 22.62116915500314,
|
| 600 |
+
"learning_rate": 9.850879293654827e-07,
|
| 601 |
+
"logits/chosen": 0.5146647691726685,
|
| 602 |
+
"logits/rejected": 0.519727349281311,
|
| 603 |
+
"logps/chosen": -184.5218505859375,
|
| 604 |
+
"logps/rejected": -216.04702758789062,
|
| 605 |
+
"loss": 0.2716,
|
| 606 |
+
"rewards/accuracies": 0.899218738079071,
|
| 607 |
+
"rewards/chosen": -1.9234856367111206,
|
| 608 |
+
"rewards/margins": 3.6874711513519287,
|
| 609 |
+
"rewards/rejected": -5.61095666885376,
|
| 610 |
+
"step": 390
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.5688888888888889,
|
| 614 |
+
"grad_norm": 19.796696127701463,
|
| 615 |
+
"learning_rate": 9.354235983589227e-07,
|
| 616 |
+
"logits/chosen": 0.5061088800430298,
|
| 617 |
+
"logits/rejected": 0.514001727104187,
|
| 618 |
+
"logps/chosen": -173.64303588867188,
|
| 619 |
+
"logps/rejected": -217.069580078125,
|
| 620 |
+
"loss": 0.2492,
|
| 621 |
+
"rewards/accuracies": 0.8960937857627869,
|
| 622 |
+
"rewards/chosen": -1.4420337677001953,
|
| 623 |
+
"rewards/margins": 3.665221691131592,
|
| 624 |
+
"rewards/rejected": -5.107255458831787,
|
| 625 |
+
"step": 400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.5688888888888889,
|
| 629 |
+
"eval_logits/chosen": 0.5132250785827637,
|
| 630 |
+
"eval_logits/rejected": 0.521979808807373,
|
| 631 |
+
"eval_logps/chosen": -172.3598175048828,
|
| 632 |
+
"eval_logps/rejected": -214.32054138183594,
|
| 633 |
+
"eval_loss": 0.264188677072525,
|
| 634 |
+
"eval_rewards/accuracies": 0.8984999656677246,
|
| 635 |
+
"eval_rewards/chosen": -1.3250131607055664,
|
| 636 |
+
"eval_rewards/margins": 3.49660587310791,
|
| 637 |
+
"eval_rewards/rejected": -4.821618556976318,
|
| 638 |
+
"eval_runtime": 557.5686,
|
| 639 |
+
"eval_samples_per_second": 17.935,
|
| 640 |
+
"eval_steps_per_second": 2.242,
|
| 641 |
+
"step": 400
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.5831111111111111,
|
| 645 |
+
"grad_norm": 19.738936537341292,
|
| 646 |
+
"learning_rate": 8.859188001573915e-07,
|
| 647 |
+
"logits/chosen": 0.5021483302116394,
|
| 648 |
+
"logits/rejected": 0.5080961585044861,
|
| 649 |
+
"logps/chosen": -177.53768920898438,
|
| 650 |
+
"logps/rejected": -220.1824188232422,
|
| 651 |
+
"loss": 0.2368,
|
| 652 |
+
"rewards/accuracies": 0.9117187261581421,
|
| 653 |
+
"rewards/chosen": -1.589120864868164,
|
| 654 |
+
"rewards/margins": 3.846306800842285,
|
| 655 |
+
"rewards/rejected": -5.435427665710449,
|
| 656 |
+
"step": 410
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.5973333333333334,
|
| 660 |
+
"grad_norm": 26.995300789664732,
|
| 661 |
+
"learning_rate": 8.366958339153598e-07,
|
| 662 |
+
"logits/chosen": 0.46122005581855774,
|
| 663 |
+
"logits/rejected": 0.47435277700424194,
|
| 664 |
+
"logps/chosen": -190.0319366455078,
|
| 665 |
+
"logps/rejected": -236.9222412109375,
|
| 666 |
+
"loss": 0.2476,
|
| 667 |
+
"rewards/accuracies": 0.8984375,
|
| 668 |
+
"rewards/chosen": -2.7638442516326904,
|
| 669 |
+
"rewards/margins": 4.4079084396362305,
|
| 670 |
+
"rewards/rejected": -7.171752452850342,
|
| 671 |
+
"step": 420
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.6115555555555555,
|
| 675 |
+
"grad_norm": 19.30988927869592,
|
| 676 |
+
"learning_rate": 7.878763025353874e-07,
|
| 677 |
+
"logits/chosen": 0.4701952338218689,
|
| 678 |
+
"logits/rejected": 0.4757159352302551,
|
| 679 |
+
"logps/chosen": -181.3089141845703,
|
| 680 |
+
"logps/rejected": -224.1280517578125,
|
| 681 |
+
"loss": 0.257,
|
| 682 |
+
"rewards/accuracies": 0.903124988079071,
|
| 683 |
+
"rewards/chosen": -2.4702322483062744,
|
| 684 |
+
"rewards/margins": 4.117379665374756,
|
| 685 |
+
"rewards/rejected": -6.587612152099609,
|
| 686 |
+
"step": 430
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.6257777777777778,
|
| 690 |
+
"grad_norm": 19.013100126792075,
|
| 691 |
+
"learning_rate": 7.395808122541695e-07,
|
| 692 |
+
"logits/chosen": 0.497684508562088,
|
| 693 |
+
"logits/rejected": 0.5002421140670776,
|
| 694 |
+
"logps/chosen": -183.77711486816406,
|
| 695 |
+
"logps/rejected": -220.78939819335938,
|
| 696 |
+
"loss": 0.2288,
|
| 697 |
+
"rewards/accuracies": 0.913281261920929,
|
| 698 |
+
"rewards/chosen": -1.9102922677993774,
|
| 699 |
+
"rewards/margins": 4.032773971557617,
|
| 700 |
+
"rewards/rejected": -5.943066120147705,
|
| 701 |
+
"step": 440
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 0.64,
|
| 705 |
+
"grad_norm": 21.305103958890808,
|
| 706 |
+
"learning_rate": 6.919286746907962e-07,
|
| 707 |
+
"logits/chosen": 0.5036830902099609,
|
| 708 |
+
"logits/rejected": 0.5137404203414917,
|
| 709 |
+
"logps/chosen": -171.51963806152344,
|
| 710 |
+
"logps/rejected": -215.0752716064453,
|
| 711 |
+
"loss": 0.225,
|
| 712 |
+
"rewards/accuracies": 0.9210937023162842,
|
| 713 |
+
"rewards/chosen": -1.7843152284622192,
|
| 714 |
+
"rewards/margins": 3.941601037979126,
|
| 715 |
+
"rewards/rejected": -5.725916862487793,
|
| 716 |
+
"step": 450
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.6542222222222223,
|
| 720 |
+
"grad_norm": 14.039456923478271,
|
| 721 |
+
"learning_rate": 6.450376120933008e-07,
|
| 722 |
+
"logits/chosen": 0.49933820962905884,
|
| 723 |
+
"logits/rejected": 0.5057050585746765,
|
| 724 |
+
"logps/chosen": -179.2303466796875,
|
| 725 |
+
"logps/rejected": -222.6201934814453,
|
| 726 |
+
"loss": 0.2072,
|
| 727 |
+
"rewards/accuracies": 0.9195312261581421,
|
| 728 |
+
"rewards/chosen": -1.802661418914795,
|
| 729 |
+
"rewards/margins": 4.118900299072266,
|
| 730 |
+
"rewards/rejected": -5.9215617179870605,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.6684444444444444,
|
| 735 |
+
"grad_norm": 21.70845956801297,
|
| 736 |
+
"learning_rate": 5.990234665116712e-07,
|
| 737 |
+
"logits/chosen": 0.5047470331192017,
|
| 738 |
+
"logits/rejected": 0.5078807473182678,
|
| 739 |
+
"logps/chosen": -177.2841339111328,
|
| 740 |
+
"logps/rejected": -217.75094604492188,
|
| 741 |
+
"loss": 0.2539,
|
| 742 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 743 |
+
"rewards/chosen": -1.7726950645446777,
|
| 744 |
+
"rewards/margins": 4.051829814910889,
|
| 745 |
+
"rewards/rejected": -5.824525833129883,
|
| 746 |
+
"step": 470
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.6826666666666666,
|
| 750 |
+
"grad_norm": 20.038733770110362,
|
| 751 |
+
"learning_rate": 5.539999136157976e-07,
|
| 752 |
+
"logits/chosen": 0.5074894428253174,
|
| 753 |
+
"logits/rejected": 0.5155391693115234,
|
| 754 |
+
"logps/chosen": -170.88722229003906,
|
| 755 |
+
"logps/rejected": -215.5301055908203,
|
| 756 |
+
"loss": 0.2568,
|
| 757 |
+
"rewards/accuracies": 0.8976563215255737,
|
| 758 |
+
"rewards/chosen": -1.3999384641647339,
|
| 759 |
+
"rewards/margins": 3.897874116897583,
|
| 760 |
+
"rewards/rejected": -5.297812461853027,
|
| 761 |
+
"step": 480
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.6968888888888889,
|
| 765 |
+
"grad_norm": 22.766458720803445,
|
| 766 |
+
"learning_rate": 5.100781818653547e-07,
|
| 767 |
+
"logits/chosen": 0.49489301443099976,
|
| 768 |
+
"logits/rejected": 0.5095649361610413,
|
| 769 |
+
"logps/chosen": -169.5738525390625,
|
| 770 |
+
"logps/rejected": -226.97323608398438,
|
| 771 |
+
"loss": 0.2281,
|
| 772 |
+
"rewards/accuracies": 0.9101563096046448,
|
| 773 |
+
"rewards/chosen": -1.710286259651184,
|
| 774 |
+
"rewards/margins": 4.080615043640137,
|
| 775 |
+
"rewards/rejected": -5.790901184082031,
|
| 776 |
+
"step": 490
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.7111111111111111,
|
| 780 |
+
"grad_norm": 18.468814904555817,
|
| 781 |
+
"learning_rate": 4.6736677772539435e-07,
|
| 782 |
+
"logits/chosen": 0.5003610849380493,
|
| 783 |
+
"logits/rejected": 0.5155267119407654,
|
| 784 |
+
"logps/chosen": -175.24559020996094,
|
| 785 |
+
"logps/rejected": -229.2140655517578,
|
| 786 |
+
"loss": 0.2293,
|
| 787 |
+
"rewards/accuracies": 0.91015625,
|
| 788 |
+
"rewards/chosen": -1.8444459438323975,
|
| 789 |
+
"rewards/margins": 4.331246852874756,
|
| 790 |
+
"rewards/rejected": -6.175692558288574,
|
| 791 |
+
"step": 500
|
| 792 |
+
}
|
| 793 |
+
],
|
| 794 |
+
"logging_steps": 10,
|
| 795 |
+
"max_steps": 703,
|
| 796 |
+
"num_input_tokens_seen": 0,
|
| 797 |
+
"num_train_epochs": 1,
|
| 798 |
+
"save_steps": 500,
|
| 799 |
+
"stateful_callbacks": {
|
| 800 |
+
"TrainerControl": {
|
| 801 |
+
"args": {
|
| 802 |
+
"should_epoch_stop": false,
|
| 803 |
+
"should_evaluate": false,
|
| 804 |
+
"should_log": false,
|
| 805 |
+
"should_save": true,
|
| 806 |
+
"should_training_stop": false
|
| 807 |
+
},
|
| 808 |
+
"attributes": {}
|
| 809 |
+
}
|
| 810 |
+
},
|
| 811 |
+
"total_flos": 237826049835008.0,
|
| 812 |
+
"train_batch_size": 1,
|
| 813 |
+
"trial_name": null,
|
| 814 |
+
"trial_params": null
|
| 815 |
+
}
|