Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +799 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,799 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 4.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 516,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.07782101167315175,
|
| 14 |
+
"grad_norm": 309.9635314941406,
|
| 15 |
+
"learning_rate": 3.461538461538461e-07,
|
| 16 |
+
"logits/chosen": 2.9046390056610107,
|
| 17 |
+
"logits/rejected": 2.998429775238037,
|
| 18 |
+
"logps/chosen": -250.68215942382812,
|
| 19 |
+
"logps/rejected": -271.45562744140625,
|
| 20 |
+
"loss": 0.8341,
|
| 21 |
+
"rewards/accuracies": 0.534375011920929,
|
| 22 |
+
"rewards/chosen": 0.07394347339868546,
|
| 23 |
+
"rewards/margins": 0.05555616691708565,
|
| 24 |
+
"rewards/rejected": 0.018387287855148315,
|
| 25 |
+
"step": 10
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.1556420233463035,
|
| 29 |
+
"grad_norm": 180.70631408691406,
|
| 30 |
+
"learning_rate": 7.307692307692307e-07,
|
| 31 |
+
"logits/chosen": 2.893648386001587,
|
| 32 |
+
"logits/rejected": 2.9738540649414062,
|
| 33 |
+
"logps/chosen": -255.9049835205078,
|
| 34 |
+
"logps/rejected": -271.83197021484375,
|
| 35 |
+
"loss": 0.6594,
|
| 36 |
+
"rewards/accuracies": 0.6656249761581421,
|
| 37 |
+
"rewards/chosen": 0.6080068349838257,
|
| 38 |
+
"rewards/margins": 0.7042483687400818,
|
| 39 |
+
"rewards/rejected": -0.09624147415161133,
|
| 40 |
+
"step": 20
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.23346303501945526,
|
| 44 |
+
"grad_norm": 152.26849365234375,
|
| 45 |
+
"learning_rate": 1.1153846153846154e-06,
|
| 46 |
+
"logits/chosen": 2.938114643096924,
|
| 47 |
+
"logits/rejected": 2.979564666748047,
|
| 48 |
+
"logps/chosen": -260.2184143066406,
|
| 49 |
+
"logps/rejected": -274.0084228515625,
|
| 50 |
+
"loss": 0.4358,
|
| 51 |
+
"rewards/accuracies": 0.78125,
|
| 52 |
+
"rewards/chosen": 2.1423935890197754,
|
| 53 |
+
"rewards/margins": 2.729647159576416,
|
| 54 |
+
"rewards/rejected": -0.5872532725334167,
|
| 55 |
+
"step": 30
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.311284046692607,
|
| 59 |
+
"grad_norm": 61.16145324707031,
|
| 60 |
+
"learning_rate": 1.5e-06,
|
| 61 |
+
"logits/chosen": 2.962891101837158,
|
| 62 |
+
"logits/rejected": 3.0372633934020996,
|
| 63 |
+
"logps/chosen": -264.4953918457031,
|
| 64 |
+
"logps/rejected": -278.98541259765625,
|
| 65 |
+
"loss": 0.2424,
|
| 66 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 67 |
+
"rewards/chosen": 3.3210906982421875,
|
| 68 |
+
"rewards/margins": 5.616915225982666,
|
| 69 |
+
"rewards/rejected": -2.2958242893218994,
|
| 70 |
+
"step": 40
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.38910505836575876,
|
| 74 |
+
"grad_norm": 72.79554748535156,
|
| 75 |
+
"learning_rate": 1.8846153846153845e-06,
|
| 76 |
+
"logits/chosen": 2.916743278503418,
|
| 77 |
+
"logits/rejected": 2.9901554584503174,
|
| 78 |
+
"logps/chosen": -242.43307495117188,
|
| 79 |
+
"logps/rejected": -276.2446594238281,
|
| 80 |
+
"loss": 0.0823,
|
| 81 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 82 |
+
"rewards/chosen": 4.046940803527832,
|
| 83 |
+
"rewards/margins": 9.120622634887695,
|
| 84 |
+
"rewards/rejected": -5.0736823081970215,
|
| 85 |
+
"step": 50
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.4669260700389105,
|
| 89 |
+
"grad_norm": 51.30080795288086,
|
| 90 |
+
"learning_rate": 1.998877080632712e-06,
|
| 91 |
+
"logits/chosen": 2.911020517349243,
|
| 92 |
+
"logits/rejected": 2.9875214099884033,
|
| 93 |
+
"logps/chosen": -252.1035614013672,
|
| 94 |
+
"logps/rejected": -278.92047119140625,
|
| 95 |
+
"loss": 0.0745,
|
| 96 |
+
"rewards/accuracies": 0.973437488079071,
|
| 97 |
+
"rewards/chosen": 3.5658841133117676,
|
| 98 |
+
"rewards/margins": 11.655838966369629,
|
| 99 |
+
"rewards/rejected": -8.089953422546387,
|
| 100 |
+
"step": 60
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.5447470817120622,
|
| 104 |
+
"grad_norm": 69.88111877441406,
|
| 105 |
+
"learning_rate": 1.9933831375786213e-06,
|
| 106 |
+
"logits/chosen": 2.9574735164642334,
|
| 107 |
+
"logits/rejected": 3.0135583877563477,
|
| 108 |
+
"logps/chosen": -250.85684204101562,
|
| 109 |
+
"logps/rejected": -278.03582763671875,
|
| 110 |
+
"loss": 0.0516,
|
| 111 |
+
"rewards/accuracies": 0.981249988079071,
|
| 112 |
+
"rewards/chosen": 2.9169883728027344,
|
| 113 |
+
"rewards/margins": 13.218663215637207,
|
| 114 |
+
"rewards/rejected": -10.301675796508789,
|
| 115 |
+
"step": 70
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.622568093385214,
|
| 119 |
+
"grad_norm": 16.288284301757812,
|
| 120 |
+
"learning_rate": 1.9833370645493046e-06,
|
| 121 |
+
"logits/chosen": 2.9377493858337402,
|
| 122 |
+
"logits/rejected": 3.007570743560791,
|
| 123 |
+
"logps/chosen": -253.7940216064453,
|
| 124 |
+
"logps/rejected": -284.8679504394531,
|
| 125 |
+
"loss": 0.0387,
|
| 126 |
+
"rewards/accuracies": 0.987500011920929,
|
| 127 |
+
"rewards/chosen": 3.284533977508545,
|
| 128 |
+
"rewards/margins": 16.070585250854492,
|
| 129 |
+
"rewards/rejected": -12.786050796508789,
|
| 130 |
+
"step": 80
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.7003891050583657,
|
| 134 |
+
"grad_norm": 8.968023300170898,
|
| 135 |
+
"learning_rate": 1.9687848971864386e-06,
|
| 136 |
+
"logits/chosen": 2.9306931495666504,
|
| 137 |
+
"logits/rejected": 2.964722156524658,
|
| 138 |
+
"logps/chosen": -252.9971160888672,
|
| 139 |
+
"logps/rejected": -275.87188720703125,
|
| 140 |
+
"loss": 0.0492,
|
| 141 |
+
"rewards/accuracies": 0.987500011920929,
|
| 142 |
+
"rewards/chosen": 3.5943214893341064,
|
| 143 |
+
"rewards/margins": 17.51387596130371,
|
| 144 |
+
"rewards/rejected": -13.9195556640625,
|
| 145 |
+
"step": 90
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.7782101167315175,
|
| 149 |
+
"grad_norm": 2.590059757232666,
|
| 150 |
+
"learning_rate": 1.9497933200901047e-06,
|
| 151 |
+
"logits/chosen": 2.9647717475891113,
|
| 152 |
+
"logits/rejected": 3.0007081031799316,
|
| 153 |
+
"logps/chosen": -267.41375732421875,
|
| 154 |
+
"logps/rejected": -281.07513427734375,
|
| 155 |
+
"loss": 0.0337,
|
| 156 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 157 |
+
"rewards/chosen": 4.1861348152160645,
|
| 158 |
+
"rewards/margins": 18.587297439575195,
|
| 159 |
+
"rewards/rejected": -14.401159286499023,
|
| 160 |
+
"step": 100
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.8560311284046692,
|
| 164 |
+
"grad_norm": 77.77639770507812,
|
| 165 |
+
"learning_rate": 1.926449361239848e-06,
|
| 166 |
+
"logits/chosen": 2.8810315132141113,
|
| 167 |
+
"logits/rejected": 2.9851112365722656,
|
| 168 |
+
"logps/chosen": -253.496337890625,
|
| 169 |
+
"logps/rejected": -284.06610107421875,
|
| 170 |
+
"loss": 0.0533,
|
| 171 |
+
"rewards/accuracies": 0.987500011920929,
|
| 172 |
+
"rewards/chosen": 4.9546074867248535,
|
| 173 |
+
"rewards/margins": 20.860849380493164,
|
| 174 |
+
"rewards/rejected": -15.906242370605469,
|
| 175 |
+
"step": 110
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.933852140077821,
|
| 179 |
+
"grad_norm": 127.29902648925781,
|
| 180 |
+
"learning_rate": 1.8988599931931866e-06,
|
| 181 |
+
"logits/chosen": 2.8698949813842773,
|
| 182 |
+
"logits/rejected": 2.943873405456543,
|
| 183 |
+
"logps/chosen": -246.2525177001953,
|
| 184 |
+
"logps/rejected": -279.6893310546875,
|
| 185 |
+
"loss": 0.0179,
|
| 186 |
+
"rewards/accuracies": 0.995312511920929,
|
| 187 |
+
"rewards/chosen": 3.602506160736084,
|
| 188 |
+
"rewards/margins": 21.305601119995117,
|
| 189 |
+
"rewards/rejected": -17.703094482421875,
|
| 190 |
+
"step": 120
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.0077821011673151,
|
| 194 |
+
"grad_norm": 0.10526127368211746,
|
| 195 |
+
"learning_rate": 1.8671516428890646e-06,
|
| 196 |
+
"logits/chosen": 2.960531711578369,
|
| 197 |
+
"logits/rejected": 3.0081727504730225,
|
| 198 |
+
"logps/chosen": -251.44546508789062,
|
| 199 |
+
"logps/rejected": -297.5830383300781,
|
| 200 |
+
"loss": 0.0043,
|
| 201 |
+
"rewards/accuracies": 0.9983552694320679,
|
| 202 |
+
"rewards/chosen": 5.055600166320801,
|
| 203 |
+
"rewards/margins": 22.011268615722656,
|
| 204 |
+
"rewards/rejected": -16.955669403076172,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.0856031128404668,
|
| 209 |
+
"grad_norm": 3.8229660987854004,
|
| 210 |
+
"learning_rate": 1.8314696123025452e-06,
|
| 211 |
+
"logits/chosen": 2.9409918785095215,
|
| 212 |
+
"logits/rejected": 2.994568109512329,
|
| 213 |
+
"logps/chosen": -251.5284423828125,
|
| 214 |
+
"logps/rejected": -283.5034484863281,
|
| 215 |
+
"loss": 0.002,
|
| 216 |
+
"rewards/accuracies": 0.9984375238418579,
|
| 217 |
+
"rewards/chosen": 5.045574188232422,
|
| 218 |
+
"rewards/margins": 24.76313591003418,
|
| 219 |
+
"rewards/rejected": -19.717559814453125,
|
| 220 |
+
"step": 140
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.1634241245136188,
|
| 224 |
+
"grad_norm": 0.1067415103316307,
|
| 225 |
+
"learning_rate": 1.7919774126055671e-06,
|
| 226 |
+
"logits/chosen": 2.9014945030212402,
|
| 227 |
+
"logits/rejected": 3.0171058177948,
|
| 228 |
+
"logps/chosen": -244.58154296875,
|
| 229 |
+
"logps/rejected": -300.16827392578125,
|
| 230 |
+
"loss": 0.0005,
|
| 231 |
+
"rewards/accuracies": 1.0,
|
| 232 |
+
"rewards/chosen": 5.922303199768066,
|
| 233 |
+
"rewards/margins": 27.960514068603516,
|
| 234 |
+
"rewards/rejected": -22.038206100463867,
|
| 235 |
+
"step": 150
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.2412451361867705,
|
| 239 |
+
"grad_norm": 0.11893969029188156,
|
| 240 |
+
"learning_rate": 1.7488560148849425e-06,
|
| 241 |
+
"logits/chosen": 2.930441379547119,
|
| 242 |
+
"logits/rejected": 2.9720261096954346,
|
| 243 |
+
"logps/chosen": -257.78741455078125,
|
| 244 |
+
"logps/rejected": -283.66986083984375,
|
| 245 |
+
"loss": 0.0044,
|
| 246 |
+
"rewards/accuracies": 0.9984375238418579,
|
| 247 |
+
"rewards/chosen": 4.894805908203125,
|
| 248 |
+
"rewards/margins": 27.222713470458984,
|
| 249 |
+
"rewards/rejected": -22.32790756225586,
|
| 250 |
+
"step": 160
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.3190661478599222,
|
| 254 |
+
"grad_norm": 0.019848737865686417,
|
| 255 |
+
"learning_rate": 1.7023030208511418e-06,
|
| 256 |
+
"logits/chosen": 2.9215731620788574,
|
| 257 |
+
"logits/rejected": 2.998058795928955,
|
| 258 |
+
"logps/chosen": -257.8537902832031,
|
| 259 |
+
"logps/rejected": -300.69781494140625,
|
| 260 |
+
"loss": 0.0005,
|
| 261 |
+
"rewards/accuracies": 1.0,
|
| 262 |
+
"rewards/chosen": 4.8431878089904785,
|
| 263 |
+
"rewards/margins": 27.765335083007812,
|
| 264 |
+
"rewards/rejected": -22.92214584350586,
|
| 265 |
+
"step": 170
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.3968871595330739,
|
| 269 |
+
"grad_norm": 0.013338716700673103,
|
| 270 |
+
"learning_rate": 1.6525317573380523e-06,
|
| 271 |
+
"logits/chosen": 2.905754566192627,
|
| 272 |
+
"logits/rejected": 2.9750256538391113,
|
| 273 |
+
"logps/chosen": -256.52801513671875,
|
| 274 |
+
"logps/rejected": -292.92559814453125,
|
| 275 |
+
"loss": 0.0002,
|
| 276 |
+
"rewards/accuracies": 1.0,
|
| 277 |
+
"rewards/chosen": 5.072903633117676,
|
| 278 |
+
"rewards/margins": 28.44639015197754,
|
| 279 |
+
"rewards/rejected": -23.373485565185547,
|
| 280 |
+
"step": 180
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.4747081712062258,
|
| 284 |
+
"grad_norm": 1.3543317317962646,
|
| 285 |
+
"learning_rate": 1.5997702987431278e-06,
|
| 286 |
+
"logits/chosen": 2.8545498847961426,
|
| 287 |
+
"logits/rejected": 2.93253755569458,
|
| 288 |
+
"logps/chosen": -249.04165649414062,
|
| 289 |
+
"logps/rejected": -290.2785949707031,
|
| 290 |
+
"loss": 0.0001,
|
| 291 |
+
"rewards/accuracies": 1.0,
|
| 292 |
+
"rewards/chosen": 4.263331413269043,
|
| 293 |
+
"rewards/margins": 29.8680362701416,
|
| 294 |
+
"rewards/rejected": -25.604705810546875,
|
| 295 |
+
"step": 190
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.5525291828793775,
|
| 299 |
+
"grad_norm": 0.0023172160144895315,
|
| 300 |
+
"learning_rate": 1.5442604218875369e-06,
|
| 301 |
+
"logits/chosen": 2.8726515769958496,
|
| 302 |
+
"logits/rejected": 2.96691632270813,
|
| 303 |
+
"logps/chosen": -235.4767608642578,
|
| 304 |
+
"logps/rejected": -299.6068420410156,
|
| 305 |
+
"loss": 0.0174,
|
| 306 |
+
"rewards/accuracies": 0.9984375238418579,
|
| 307 |
+
"rewards/chosen": 4.076581001281738,
|
| 308 |
+
"rewards/margins": 29.738256454467773,
|
| 309 |
+
"rewards/rejected": -25.661672592163086,
|
| 310 |
+
"step": 200
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.6303501945525292,
|
| 314 |
+
"grad_norm": 0.12606221437454224,
|
| 315 |
+
"learning_rate": 1.4862564980856257e-06,
|
| 316 |
+
"logits/chosen": 2.866564989089966,
|
| 317 |
+
"logits/rejected": 2.9289443492889404,
|
| 318 |
+
"logps/chosen": -251.0595245361328,
|
| 319 |
+
"logps/rejected": -291.28009033203125,
|
| 320 |
+
"loss": 0.0002,
|
| 321 |
+
"rewards/accuracies": 1.0,
|
| 322 |
+
"rewards/chosen": 2.828310012817383,
|
| 323 |
+
"rewards/margins": 29.65151023864746,
|
| 324 |
+
"rewards/rejected": -26.823205947875977,
|
| 325 |
+
"step": 210
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.708171206225681,
|
| 329 |
+
"grad_norm": 1.1696350574493408,
|
| 330 |
+
"learning_rate": 1.4260243275007265e-06,
|
| 331 |
+
"logits/chosen": 2.876115083694458,
|
| 332 |
+
"logits/rejected": 2.9629712104797363,
|
| 333 |
+
"logps/chosen": -252.755126953125,
|
| 334 |
+
"logps/rejected": -293.724609375,
|
| 335 |
+
"loss": 0.0001,
|
| 336 |
+
"rewards/accuracies": 1.0,
|
| 337 |
+
"rewards/chosen": 2.9194681644439697,
|
| 338 |
+
"rewards/margins": 30.779850006103516,
|
| 339 |
+
"rewards/rejected": -27.86037826538086,
|
| 340 |
+
"step": 220
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 1.7859922178988326,
|
| 344 |
+
"grad_norm": 0.7943425178527832,
|
| 345 |
+
"learning_rate": 1.3638399211288185e-06,
|
| 346 |
+
"logits/chosen": 2.853001356124878,
|
| 347 |
+
"logits/rejected": 2.91581392288208,
|
| 348 |
+
"logps/chosen": -248.50790405273438,
|
| 349 |
+
"logps/rejected": -303.24749755859375,
|
| 350 |
+
"loss": 0.0009,
|
| 351 |
+
"rewards/accuracies": 1.0,
|
| 352 |
+
"rewards/chosen": 3.9409661293029785,
|
| 353 |
+
"rewards/margins": 31.501480102539062,
|
| 354 |
+
"rewards/rejected": -27.560516357421875,
|
| 355 |
+
"step": 230
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 1.8638132295719845,
|
| 359 |
+
"grad_norm": 22.544157028198242,
|
| 360 |
+
"learning_rate": 1.2999882359915478e-06,
|
| 361 |
+
"logits/chosen": 2.838768482208252,
|
| 362 |
+
"logits/rejected": 2.8869760036468506,
|
| 363 |
+
"logps/chosen": -257.80755615234375,
|
| 364 |
+
"logps/rejected": -301.2661437988281,
|
| 365 |
+
"loss": 0.0009,
|
| 366 |
+
"rewards/accuracies": 1.0,
|
| 367 |
+
"rewards/chosen": 4.118237018585205,
|
| 368 |
+
"rewards/margins": 31.00071144104004,
|
| 369 |
+
"rewards/rejected": -26.882476806640625,
|
| 370 |
+
"step": 240
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 1.9416342412451362,
|
| 374 |
+
"grad_norm": 0.08505785465240479,
|
| 375 |
+
"learning_rate": 1.2347618693345148e-06,
|
| 376 |
+
"logits/chosen": 2.8469579219818115,
|
| 377 |
+
"logits/rejected": 2.91058349609375,
|
| 378 |
+
"logps/chosen": -274.753173828125,
|
| 379 |
+
"logps/rejected": -310.84820556640625,
|
| 380 |
+
"loss": 0.0097,
|
| 381 |
+
"rewards/accuracies": 0.996874988079071,
|
| 382 |
+
"rewards/chosen": 3.213207960128784,
|
| 383 |
+
"rewards/margins": 32.35531234741211,
|
| 384 |
+
"rewards/rejected": -29.142101287841797,
|
| 385 |
+
"step": 250
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 2.0155642023346303,
|
| 389 |
+
"grad_norm": 0.15910857915878296,
|
| 390 |
+
"learning_rate": 1.1684597178146115e-06,
|
| 391 |
+
"logits/chosen": 2.820605516433716,
|
| 392 |
+
"logits/rejected": 2.9221818447113037,
|
| 393 |
+
"logps/chosen": -240.77276611328125,
|
| 394 |
+
"logps/rejected": -284.52947998046875,
|
| 395 |
+
"loss": 0.0095,
|
| 396 |
+
"rewards/accuracies": 1.0,
|
| 397 |
+
"rewards/chosen": 4.895946979522705,
|
| 398 |
+
"rewards/margins": 31.758358001708984,
|
| 399 |
+
"rewards/rejected": -26.862411499023438,
|
| 400 |
+
"step": 260
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.093385214007782,
|
| 404 |
+
"grad_norm": 0.036736179143190384,
|
| 405 |
+
"learning_rate": 1.1013856078206078e-06,
|
| 406 |
+
"logits/chosen": 2.7885546684265137,
|
| 407 |
+
"logits/rejected": 2.9099419116973877,
|
| 408 |
+
"logps/chosen": -255.7447509765625,
|
| 409 |
+
"logps/rejected": -300.6316223144531,
|
| 410 |
+
"loss": 0.0,
|
| 411 |
+
"rewards/accuracies": 1.0,
|
| 412 |
+
"rewards/chosen": 4.1806111335754395,
|
| 413 |
+
"rewards/margins": 32.345706939697266,
|
| 414 |
+
"rewards/rejected": -28.16509437561035,
|
| 415 |
+
"step": 270
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.1712062256809337,
|
| 419 |
+
"grad_norm": 0.010115005075931549,
|
| 420 |
+
"learning_rate": 1.0338469032034844e-06,
|
| 421 |
+
"logits/chosen": 2.880408763885498,
|
| 422 |
+
"logits/rejected": 2.913715124130249,
|
| 423 |
+
"logps/chosen": -244.33837890625,
|
| 424 |
+
"logps/rejected": -284.74456787109375,
|
| 425 |
+
"loss": 0.0,
|
| 426 |
+
"rewards/accuracies": 1.0,
|
| 427 |
+
"rewards/chosen": 5.522616386413574,
|
| 428 |
+
"rewards/margins": 32.65098571777344,
|
| 429 |
+
"rewards/rejected": -27.128368377685547,
|
| 430 |
+
"step": 280
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 2.2490272373540856,
|
| 434 |
+
"grad_norm": 0.0006411768263205886,
|
| 435 |
+
"learning_rate": 9.661530967965155e-07,
|
| 436 |
+
"logits/chosen": 2.8637478351593018,
|
| 437 |
+
"logits/rejected": 2.8776774406433105,
|
| 438 |
+
"logps/chosen": -251.5993194580078,
|
| 439 |
+
"logps/rejected": -285.7321472167969,
|
| 440 |
+
"loss": 0.0,
|
| 441 |
+
"rewards/accuracies": 1.0,
|
| 442 |
+
"rewards/chosen": 5.7556328773498535,
|
| 443 |
+
"rewards/margins": 32.75992202758789,
|
| 444 |
+
"rewards/rejected": -27.004287719726562,
|
| 445 |
+
"step": 290
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 2.3268482490272375,
|
| 449 |
+
"grad_norm": 0.010583557188510895,
|
| 450 |
+
"learning_rate": 8.986143921793921e-07,
|
| 451 |
+
"logits/chosen": 2.8255324363708496,
|
| 452 |
+
"logits/rejected": 2.884343385696411,
|
| 453 |
+
"logps/chosen": -256.0443115234375,
|
| 454 |
+
"logps/rejected": -298.6083068847656,
|
| 455 |
+
"loss": 0.0,
|
| 456 |
+
"rewards/accuracies": 1.0,
|
| 457 |
+
"rewards/chosen": 4.289700984954834,
|
| 458 |
+
"rewards/margins": 32.55815887451172,
|
| 459 |
+
"rewards/rejected": -28.26845359802246,
|
| 460 |
+
"step": 300
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 2.404669260700389,
|
| 464 |
+
"grad_norm": 0.05514160171151161,
|
| 465 |
+
"learning_rate": 8.315402821853885e-07,
|
| 466 |
+
"logits/chosen": 2.8168251514434814,
|
| 467 |
+
"logits/rejected": 2.911100387573242,
|
| 468 |
+
"logps/chosen": -242.12930297851562,
|
| 469 |
+
"logps/rejected": -311.6902770996094,
|
| 470 |
+
"loss": 0.0,
|
| 471 |
+
"rewards/accuracies": 1.0,
|
| 472 |
+
"rewards/chosen": 4.304950714111328,
|
| 473 |
+
"rewards/margins": 33.0163459777832,
|
| 474 |
+
"rewards/rejected": -28.71139144897461,
|
| 475 |
+
"step": 310
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 2.482490272373541,
|
| 479 |
+
"grad_norm": 0.0014977785758674145,
|
| 480 |
+
"learning_rate": 7.65238130665485e-07,
|
| 481 |
+
"logits/chosen": 2.8739066123962402,
|
| 482 |
+
"logits/rejected": 2.9437661170959473,
|
| 483 |
+
"logps/chosen": -254.6911163330078,
|
| 484 |
+
"logps/rejected": -298.3764953613281,
|
| 485 |
+
"loss": 0.0,
|
| 486 |
+
"rewards/accuracies": 1.0,
|
| 487 |
+
"rewards/chosen": 4.761944770812988,
|
| 488 |
+
"rewards/margins": 32.24135208129883,
|
| 489 |
+
"rewards/rejected": -27.479406356811523,
|
| 490 |
+
"step": 320
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 2.5603112840466924,
|
| 494 |
+
"grad_norm": 0.003176012309268117,
|
| 495 |
+
"learning_rate": 7.000117640084525e-07,
|
| 496 |
+
"logits/chosen": 2.849517822265625,
|
| 497 |
+
"logits/rejected": 2.9144535064697266,
|
| 498 |
+
"logps/chosen": -253.174560546875,
|
| 499 |
+
"logps/rejected": -302.97296142578125,
|
| 500 |
+
"loss": 0.0,
|
| 501 |
+
"rewards/accuracies": 1.0,
|
| 502 |
+
"rewards/chosen": 5.072056770324707,
|
| 503 |
+
"rewards/margins": 33.25294876098633,
|
| 504 |
+
"rewards/rejected": -28.180889129638672,
|
| 505 |
+
"step": 330
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 2.6381322957198443,
|
| 509 |
+
"grad_norm": 0.007333577610552311,
|
| 510 |
+
"learning_rate": 6.361600788711815e-07,
|
| 511 |
+
"logits/chosen": 2.801302671432495,
|
| 512 |
+
"logits/rejected": 2.9121482372283936,
|
| 513 |
+
"logps/chosen": -259.02398681640625,
|
| 514 |
+
"logps/rejected": -303.2242431640625,
|
| 515 |
+
"loss": 0.0,
|
| 516 |
+
"rewards/accuracies": 1.0,
|
| 517 |
+
"rewards/chosen": 4.999932289123535,
|
| 518 |
+
"rewards/margins": 32.66350555419922,
|
| 519 |
+
"rewards/rejected": -27.66357421875,
|
| 520 |
+
"step": 340
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.7159533073929962,
|
| 524 |
+
"grad_norm": 0.0016177126672118902,
|
| 525 |
+
"learning_rate": 5.739756724992736e-07,
|
| 526 |
+
"logits/chosen": 2.85249662399292,
|
| 527 |
+
"logits/rejected": 2.9361824989318848,
|
| 528 |
+
"logps/chosen": -244.018310546875,
|
| 529 |
+
"logps/rejected": -286.9032897949219,
|
| 530 |
+
"loss": 0.0,
|
| 531 |
+
"rewards/accuracies": 1.0,
|
| 532 |
+
"rewards/chosen": 4.892648220062256,
|
| 533 |
+
"rewards/margins": 32.217613220214844,
|
| 534 |
+
"rewards/rejected": -27.324966430664062,
|
| 535 |
+
"step": 350
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 2.7937743190661477,
|
| 539 |
+
"grad_norm": 0.3721546530723572,
|
| 540 |
+
"learning_rate": 5.137435019143744e-07,
|
| 541 |
+
"logits/chosen": 2.801509141921997,
|
| 542 |
+
"logits/rejected": 2.8811545372009277,
|
| 543 |
+
"logps/chosen": -257.4078063964844,
|
| 544 |
+
"logps/rejected": -306.99542236328125,
|
| 545 |
+
"loss": 0.0001,
|
| 546 |
+
"rewards/accuracies": 1.0,
|
| 547 |
+
"rewards/chosen": 4.106388092041016,
|
| 548 |
+
"rewards/margins": 33.20668411254883,
|
| 549 |
+
"rewards/rejected": -29.100292205810547,
|
| 550 |
+
"step": 360
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 2.8715953307392996,
|
| 554 |
+
"grad_norm": 0.004263872280716896,
|
| 555 |
+
"learning_rate": 4.557395781124631e-07,
|
| 556 |
+
"logits/chosen": 2.853585958480835,
|
| 557 |
+
"logits/rejected": 2.9362740516662598,
|
| 558 |
+
"logps/chosen": -265.7266845703125,
|
| 559 |
+
"logps/rejected": -301.75091552734375,
|
| 560 |
+
"loss": 0.0,
|
| 561 |
+
"rewards/accuracies": 1.0,
|
| 562 |
+
"rewards/chosen": 4.696002960205078,
|
| 563 |
+
"rewards/margins": 32.666351318359375,
|
| 564 |
+
"rewards/rejected": -27.9703426361084,
|
| 565 |
+
"step": 370
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 2.9494163424124515,
|
| 569 |
+
"grad_norm": 6.243278503417969,
|
| 570 |
+
"learning_rate": 4.0022970125687217e-07,
|
| 571 |
+
"logits/chosen": 2.8308067321777344,
|
| 572 |
+
"logits/rejected": 2.924168586730957,
|
| 573 |
+
"logps/chosen": -253.6175537109375,
|
| 574 |
+
"logps/rejected": -304.56488037109375,
|
| 575 |
+
"loss": 0.0001,
|
| 576 |
+
"rewards/accuracies": 1.0,
|
| 577 |
+
"rewards/chosen": 4.487698554992676,
|
| 578 |
+
"rewards/margins": 33.75385284423828,
|
| 579 |
+
"rewards/rejected": -29.26615333557129,
|
| 580 |
+
"step": 380
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 3.0233463035019454,
|
| 584 |
+
"grad_norm": 0.010610430501401424,
|
| 585 |
+
"learning_rate": 3.474682426619474e-07,
|
| 586 |
+
"logits/chosen": 2.86039662361145,
|
| 587 |
+
"logits/rejected": 2.955193042755127,
|
| 588 |
+
"logps/chosen": -244.86038208007812,
|
| 589 |
+
"logps/rejected": -298.5942077636719,
|
| 590 |
+
"loss": 0.0,
|
| 591 |
+
"rewards/accuracies": 1.0,
|
| 592 |
+
"rewards/chosen": 5.364467144012451,
|
| 593 |
+
"rewards/margins": 33.471588134765625,
|
| 594 |
+
"rewards/rejected": -28.107120513916016,
|
| 595 |
+
"step": 390
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 3.1011673151750974,
|
| 599 |
+
"grad_norm": 0.0002077793178614229,
|
| 600 |
+
"learning_rate": 2.976969791488586e-07,
|
| 601 |
+
"logits/chosen": 2.794825792312622,
|
| 602 |
+
"logits/rejected": 2.913062572479248,
|
| 603 |
+
"logps/chosen": -249.8114471435547,
|
| 604 |
+
"logps/rejected": -304.4586181640625,
|
| 605 |
+
"loss": 0.0,
|
| 606 |
+
"rewards/accuracies": 1.0,
|
| 607 |
+
"rewards/chosen": 4.951796054840088,
|
| 608 |
+
"rewards/margins": 33.71596908569336,
|
| 609 |
+
"rewards/rejected": -28.764175415039062,
|
| 610 |
+
"step": 400
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.178988326848249,
|
| 614 |
+
"grad_norm": 0.0018190924311056733,
|
| 615 |
+
"learning_rate": 2.511439851150573e-07,
|
| 616 |
+
"logits/chosen": 2.846440076828003,
|
| 617 |
+
"logits/rejected": 2.9116950035095215,
|
| 618 |
+
"logps/chosen": -245.32241821289062,
|
| 619 |
+
"logps/rejected": -299.29278564453125,
|
| 620 |
+
"loss": 0.0,
|
| 621 |
+
"rewards/accuracies": 1.0,
|
| 622 |
+
"rewards/chosen": 4.315257549285889,
|
| 623 |
+
"rewards/margins": 33.424171447753906,
|
| 624 |
+
"rewards/rejected": -29.10890769958496,
|
| 625 |
+
"step": 410
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 3.2568093385214008,
|
| 629 |
+
"grad_norm": 0.0022656808141618967,
|
| 630 |
+
"learning_rate": 2.080225873944328e-07,
|
| 631 |
+
"logits/chosen": 2.8200035095214844,
|
| 632 |
+
"logits/rejected": 2.9111595153808594,
|
| 633 |
+
"logps/chosen": -258.51519775390625,
|
| 634 |
+
"logps/rejected": -303.6025390625,
|
| 635 |
+
"loss": 0.0,
|
| 636 |
+
"rewards/accuracies": 1.0,
|
| 637 |
+
"rewards/chosen": 4.961642265319824,
|
| 638 |
+
"rewards/margins": 33.01478958129883,
|
| 639 |
+
"rewards/rejected": -28.053142547607422,
|
| 640 |
+
"step": 420
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 3.3346303501945527,
|
| 644 |
+
"grad_norm": 0.006798978429287672,
|
| 645 |
+
"learning_rate": 1.6853038769745465e-07,
|
| 646 |
+
"logits/chosen": 2.8389997482299805,
|
| 647 |
+
"logits/rejected": 2.8933396339416504,
|
| 648 |
+
"logps/chosen": -245.7606658935547,
|
| 649 |
+
"logps/rejected": -295.68115234375,
|
| 650 |
+
"loss": 0.0,
|
| 651 |
+
"rewards/accuracies": 1.0,
|
| 652 |
+
"rewards/chosen": 4.181079387664795,
|
| 653 |
+
"rewards/margins": 32.27184295654297,
|
| 654 |
+
"rewards/rejected": -28.090763092041016,
|
| 655 |
+
"step": 430
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 3.412451361867704,
|
| 659 |
+
"grad_norm": 0.00100744201336056,
|
| 660 |
+
"learning_rate": 1.3284835711093535e-07,
|
| 661 |
+
"logits/chosen": 2.8530194759368896,
|
| 662 |
+
"logits/rejected": 2.9206631183624268,
|
| 663 |
+
"logps/chosen": -258.67431640625,
|
| 664 |
+
"logps/rejected": -298.80755615234375,
|
| 665 |
+
"loss": 0.0,
|
| 666 |
+
"rewards/accuracies": 1.0,
|
| 667 |
+
"rewards/chosen": 4.105883598327637,
|
| 668 |
+
"rewards/margins": 32.130775451660156,
|
| 669 |
+
"rewards/rejected": -28.024892807006836,
|
| 670 |
+
"step": 440
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 3.490272373540856,
|
| 674 |
+
"grad_norm": 0.0001550566521473229,
|
| 675 |
+
"learning_rate": 1.0114000680681356e-07,
|
| 676 |
+
"logits/chosen": 2.81189227104187,
|
| 677 |
+
"logits/rejected": 2.925079107284546,
|
| 678 |
+
"logps/chosen": -253.63217163085938,
|
| 679 |
+
"logps/rejected": -299.74273681640625,
|
| 680 |
+
"loss": 0.0,
|
| 681 |
+
"rewards/accuracies": 1.0,
|
| 682 |
+
"rewards/chosen": 6.281905174255371,
|
| 683 |
+
"rewards/margins": 33.87714385986328,
|
| 684 |
+
"rewards/rejected": -27.595239639282227,
|
| 685 |
+
"step": 450
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 3.5680933852140075,
|
| 689 |
+
"grad_norm": 0.01041385531425476,
|
| 690 |
+
"learning_rate": 7.355063876015199e-08,
|
| 691 |
+
"logits/chosen": 2.877830982208252,
|
| 692 |
+
"logits/rejected": 2.9270596504211426,
|
| 693 |
+
"logps/chosen": -277.89544677734375,
|
| 694 |
+
"logps/rejected": -303.5425109863281,
|
| 695 |
+
"loss": 0.0,
|
| 696 |
+
"rewards/accuracies": 1.0,
|
| 697 |
+
"rewards/chosen": 5.2588934898376465,
|
| 698 |
+
"rewards/margins": 32.801124572753906,
|
| 699 |
+
"rewards/rejected": -27.5422306060791,
|
| 700 |
+
"step": 460
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 3.6459143968871595,
|
| 704 |
+
"grad_norm": 0.01067538931965828,
|
| 705 |
+
"learning_rate": 5.020667990989525e-08,
|
| 706 |
+
"logits/chosen": 2.8863813877105713,
|
| 707 |
+
"logits/rejected": 2.9612936973571777,
|
| 708 |
+
"logps/chosen": -259.4737854003906,
|
| 709 |
+
"logps/rejected": -296.72637939453125,
|
| 710 |
+
"loss": 0.0,
|
| 711 |
+
"rewards/accuracies": 1.0,
|
| 712 |
+
"rewards/chosen": 4.224932670593262,
|
| 713 |
+
"rewards/margins": 32.182594299316406,
|
| 714 |
+
"rewards/rejected": -27.957666397094727,
|
| 715 |
+
"step": 470
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 3.7237354085603114,
|
| 719 |
+
"grad_norm": 0.004288141615688801,
|
| 720 |
+
"learning_rate": 3.121510281356143e-08,
|
| 721 |
+
"logits/chosen": 2.824970245361328,
|
| 722 |
+
"logits/rejected": 2.8980326652526855,
|
| 723 |
+
"logps/chosen": -251.7251434326172,
|
| 724 |
+
"logps/rejected": -293.5643615722656,
|
| 725 |
+
"loss": 0.0,
|
| 726 |
+
"rewards/accuracies": 1.0,
|
| 727 |
+
"rewards/chosen": 5.086753845214844,
|
| 728 |
+
"rewards/margins": 32.74205780029297,
|
| 729 |
+
"rewards/rejected": -27.65530776977539,
|
| 730 |
+
"step": 480
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 3.801556420233463,
|
| 734 |
+
"grad_norm": 9.263599349651486e-05,
|
| 735 |
+
"learning_rate": 1.6662935450695416e-08,
|
| 736 |
+
"logits/chosen": 2.7916643619537354,
|
| 737 |
+
"logits/rejected": 2.8788905143737793,
|
| 738 |
+
"logps/chosen": -239.5885467529297,
|
| 739 |
+
"logps/rejected": -293.6639099121094,
|
| 740 |
+
"loss": 0.0,
|
| 741 |
+
"rewards/accuracies": 1.0,
|
| 742 |
+
"rewards/chosen": 5.8184919357299805,
|
| 743 |
+
"rewards/margins": 32.92034912109375,
|
| 744 |
+
"rewards/rejected": -27.101858139038086,
|
| 745 |
+
"step": 490
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 3.8793774319066148,
|
| 749 |
+
"grad_norm": 0.003024149453267455,
|
| 750 |
+
"learning_rate": 6.6168624213784885e-09,
|
| 751 |
+
"logits/chosen": 2.828496217727661,
|
| 752 |
+
"logits/rejected": 2.8872532844543457,
|
| 753 |
+
"logps/chosen": -242.9530792236328,
|
| 754 |
+
"logps/rejected": -291.443359375,
|
| 755 |
+
"loss": 0.0,
|
| 756 |
+
"rewards/accuracies": 1.0,
|
| 757 |
+
"rewards/chosen": 4.576481342315674,
|
| 758 |
+
"rewards/margins": 31.855998992919922,
|
| 759 |
+
"rewards/rejected": -27.279516220092773,
|
| 760 |
+
"step": 500
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 3.9571984435797667,
|
| 764 |
+
"grad_norm": 0.003526742337271571,
|
| 765 |
+
"learning_rate": 1.1229193672881444e-09,
|
| 766 |
+
"logits/chosen": 2.8279409408569336,
|
| 767 |
+
"logits/rejected": 2.9110004901885986,
|
| 768 |
+
"logps/chosen": -243.04946899414062,
|
| 769 |
+
"logps/rejected": -305.1340026855469,
|
| 770 |
+
"loss": 0.0,
|
| 771 |
+
"rewards/accuracies": 1.0,
|
| 772 |
+
"rewards/chosen": 4.348401069641113,
|
| 773 |
+
"rewards/margins": 32.896034240722656,
|
| 774 |
+
"rewards/rejected": -28.547632217407227,
|
| 775 |
+
"step": 510
|
| 776 |
+
}
|
| 777 |
+
],
|
| 778 |
+
"logging_steps": 10,
|
| 779 |
+
"max_steps": 516,
|
| 780 |
+
"num_input_tokens_seen": 0,
|
| 781 |
+
"num_train_epochs": 4,
|
| 782 |
+
"save_steps": 500,
|
| 783 |
+
"stateful_callbacks": {
|
| 784 |
+
"TrainerControl": {
|
| 785 |
+
"args": {
|
| 786 |
+
"should_epoch_stop": false,
|
| 787 |
+
"should_evaluate": false,
|
| 788 |
+
"should_log": false,
|
| 789 |
+
"should_save": true,
|
| 790 |
+
"should_training_stop": true
|
| 791 |
+
},
|
| 792 |
+
"attributes": {}
|
| 793 |
+
}
|
| 794 |
+
},
|
| 795 |
+
"total_flos": 0.0,
|
| 796 |
+
"train_batch_size": 2,
|
| 797 |
+
"trial_name": null,
|
| 798 |
+
"trial_params": null
|
| 799 |
+
}
|