上传 trainer_state.json
Browse files- trainer_state.json +1453 -0
trainer_state.json
ADDED
@@ -0,0 +1,1453 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9982631930527722,
|
5 |
+
"eval_steps": 400,
|
6 |
+
"global_step": 467,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.01068804275217101,
|
13 |
+
"grad_norm": 84.02989783534706,
|
14 |
+
"learning_rate": 1.0638297872340425e-07,
|
15 |
+
"logits/chosen": -1.0397655963897705,
|
16 |
+
"logits/rejected": -1.0092562437057495,
|
17 |
+
"logps/chosen": -0.7628876566886902,
|
18 |
+
"logps/rejected": -0.7414335012435913,
|
19 |
+
"loss": 4.8657,
|
20 |
+
"rewards/accuracies": 0.3687500059604645,
|
21 |
+
"rewards/chosen": -7.628875732421875,
|
22 |
+
"rewards/margins": -0.21454186737537384,
|
23 |
+
"rewards/rejected": -7.414334774017334,
|
24 |
+
"step": 5
|
25 |
+
},
|
26 |
+
{
|
27 |
+
"epoch": 0.02137608550434202,
|
28 |
+
"grad_norm": 65.98135057247741,
|
29 |
+
"learning_rate": 2.127659574468085e-07,
|
30 |
+
"logits/chosen": -1.0671762228012085,
|
31 |
+
"logits/rejected": -0.9972389340400696,
|
32 |
+
"logps/chosen": -0.7587485909461975,
|
33 |
+
"logps/rejected": -0.7000675201416016,
|
34 |
+
"loss": 4.7692,
|
35 |
+
"rewards/accuracies": 0.41874998807907104,
|
36 |
+
"rewards/chosen": -7.587485313415527,
|
37 |
+
"rewards/margins": -0.5868101119995117,
|
38 |
+
"rewards/rejected": -7.000675201416016,
|
39 |
+
"step": 10
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.03206412825651302,
|
43 |
+
"grad_norm": 81.81057773129896,
|
44 |
+
"learning_rate": 3.1914893617021275e-07,
|
45 |
+
"logits/chosen": -0.9940463900566101,
|
46 |
+
"logits/rejected": -1.014291524887085,
|
47 |
+
"logps/chosen": -0.7531558871269226,
|
48 |
+
"logps/rejected": -0.7980529069900513,
|
49 |
+
"loss": 5.0125,
|
50 |
+
"rewards/accuracies": 0.5375000238418579,
|
51 |
+
"rewards/chosen": -7.531558990478516,
|
52 |
+
"rewards/margins": 0.44897016882896423,
|
53 |
+
"rewards/rejected": -7.980528831481934,
|
54 |
+
"step": 15
|
55 |
+
},
|
56 |
+
{
|
57 |
+
"epoch": 0.04275217100868404,
|
58 |
+
"grad_norm": 131.3611224894929,
|
59 |
+
"learning_rate": 4.25531914893617e-07,
|
60 |
+
"logits/chosen": -0.9941411018371582,
|
61 |
+
"logits/rejected": -0.9676485061645508,
|
62 |
+
"logps/chosen": -0.7748786807060242,
|
63 |
+
"logps/rejected": -0.7950050234794617,
|
64 |
+
"loss": 4.3887,
|
65 |
+
"rewards/accuracies": 0.5,
|
66 |
+
"rewards/chosen": -7.748786926269531,
|
67 |
+
"rewards/margins": 0.20126314461231232,
|
68 |
+
"rewards/rejected": -7.950050354003906,
|
69 |
+
"step": 20
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.053440213760855046,
|
73 |
+
"grad_norm": 41.98227986443809,
|
74 |
+
"learning_rate": 5.319148936170212e-07,
|
75 |
+
"logits/chosen": -1.06044602394104,
|
76 |
+
"logits/rejected": -1.0300180912017822,
|
77 |
+
"logps/chosen": -0.5242100358009338,
|
78 |
+
"logps/rejected": -0.5184083580970764,
|
79 |
+
"loss": 4.4531,
|
80 |
+
"rewards/accuracies": 0.5062500238418579,
|
81 |
+
"rewards/chosen": -5.242100715637207,
|
82 |
+
"rewards/margins": -0.05801659822463989,
|
83 |
+
"rewards/rejected": -5.184083461761475,
|
84 |
+
"step": 25
|
85 |
+
},
|
86 |
+
{
|
87 |
+
"epoch": 0.06412825651302605,
|
88 |
+
"grad_norm": 68.84250778408698,
|
89 |
+
"learning_rate": 6.382978723404255e-07,
|
90 |
+
"logits/chosen": -1.0604535341262817,
|
91 |
+
"logits/rejected": -1.0147985219955444,
|
92 |
+
"logps/chosen": -0.5427820086479187,
|
93 |
+
"logps/rejected": -0.538798451423645,
|
94 |
+
"loss": 4.0481,
|
95 |
+
"rewards/accuracies": 0.4312500059604645,
|
96 |
+
"rewards/chosen": -5.427820205688477,
|
97 |
+
"rewards/margins": -0.03983556479215622,
|
98 |
+
"rewards/rejected": -5.3879852294921875,
|
99 |
+
"step": 30
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.07481629926519706,
|
103 |
+
"grad_norm": 55.83166407300918,
|
104 |
+
"learning_rate": 7.446808510638297e-07,
|
105 |
+
"logits/chosen": -1.1716325283050537,
|
106 |
+
"logits/rejected": -1.0896761417388916,
|
107 |
+
"logps/chosen": -0.45515695214271545,
|
108 |
+
"logps/rejected": -0.46327242255210876,
|
109 |
+
"loss": 3.8398,
|
110 |
+
"rewards/accuracies": 0.48750001192092896,
|
111 |
+
"rewards/chosen": -4.55156946182251,
|
112 |
+
"rewards/margins": 0.08115490525960922,
|
113 |
+
"rewards/rejected": -4.632723808288574,
|
114 |
+
"step": 35
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.08550434201736808,
|
118 |
+
"grad_norm": 47.314667796032154,
|
119 |
+
"learning_rate": 8.51063829787234e-07,
|
120 |
+
"logits/chosen": -1.1367859840393066,
|
121 |
+
"logits/rejected": -1.0874762535095215,
|
122 |
+
"logps/chosen": -0.3599122166633606,
|
123 |
+
"logps/rejected": -0.4211342930793762,
|
124 |
+
"loss": 3.8501,
|
125 |
+
"rewards/accuracies": 0.543749988079071,
|
126 |
+
"rewards/chosen": -3.5991222858428955,
|
127 |
+
"rewards/margins": 0.6122205853462219,
|
128 |
+
"rewards/rejected": -4.211342811584473,
|
129 |
+
"step": 40
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.09619238476953908,
|
133 |
+
"grad_norm": 43.42431109758399,
|
134 |
+
"learning_rate": 9.574468085106384e-07,
|
135 |
+
"logits/chosen": -1.17836332321167,
|
136 |
+
"logits/rejected": -1.1281588077545166,
|
137 |
+
"logps/chosen": -0.377611368894577,
|
138 |
+
"logps/rejected": -0.42245084047317505,
|
139 |
+
"loss": 3.7936,
|
140 |
+
"rewards/accuracies": 0.53125,
|
141 |
+
"rewards/chosen": -3.7761130332946777,
|
142 |
+
"rewards/margins": 0.448394775390625,
|
143 |
+
"rewards/rejected": -4.224508762359619,
|
144 |
+
"step": 45
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 0.10688042752171009,
|
148 |
+
"grad_norm": 50.23949059041095,
|
149 |
+
"learning_rate": 9.998741174712533e-07,
|
150 |
+
"logits/chosen": -1.1873140335083008,
|
151 |
+
"logits/rejected": -1.1317476034164429,
|
152 |
+
"logps/chosen": -0.34768182039260864,
|
153 |
+
"logps/rejected": -0.38563376665115356,
|
154 |
+
"loss": 3.7644,
|
155 |
+
"rewards/accuracies": 0.5,
|
156 |
+
"rewards/chosen": -3.476818561553955,
|
157 |
+
"rewards/margins": 0.3795194625854492,
|
158 |
+
"rewards/rejected": -3.856337785720825,
|
159 |
+
"step": 50
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.11756847027388109,
|
163 |
+
"grad_norm": 77.86452279542212,
|
164 |
+
"learning_rate": 9.991050648838675e-07,
|
165 |
+
"logits/chosen": -1.2336595058441162,
|
166 |
+
"logits/rejected": -1.1919711828231812,
|
167 |
+
"logps/chosen": -0.33353298902511597,
|
168 |
+
"logps/rejected": -0.3969067931175232,
|
169 |
+
"loss": 3.6758,
|
170 |
+
"rewards/accuracies": 0.59375,
|
171 |
+
"rewards/chosen": -3.3353302478790283,
|
172 |
+
"rewards/margins": 0.6337377429008484,
|
173 |
+
"rewards/rejected": -3.9690680503845215,
|
174 |
+
"step": 55
|
175 |
+
},
|
176 |
+
{
|
177 |
+
"epoch": 0.1282565130260521,
|
178 |
+
"grad_norm": 48.54307450778913,
|
179 |
+
"learning_rate": 9.97637968732563e-07,
|
180 |
+
"logits/chosen": -1.2853938341140747,
|
181 |
+
"logits/rejected": -1.2468154430389404,
|
182 |
+
"logps/chosen": -0.3357910215854645,
|
183 |
+
"logps/rejected": -0.36223435401916504,
|
184 |
+
"loss": 3.6487,
|
185 |
+
"rewards/accuracies": 0.5874999761581421,
|
186 |
+
"rewards/chosen": -3.35791015625,
|
187 |
+
"rewards/margins": 0.26443368196487427,
|
188 |
+
"rewards/rejected": -3.6223437786102295,
|
189 |
+
"step": 60
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 0.13894455577822312,
|
193 |
+
"grad_norm": 67.81573981196492,
|
194 |
+
"learning_rate": 9.954748808839674e-07,
|
195 |
+
"logits/chosen": -1.203680396080017,
|
196 |
+
"logits/rejected": -1.1713194847106934,
|
197 |
+
"logps/chosen": -0.3974061906337738,
|
198 |
+
"logps/rejected": -0.4787389636039734,
|
199 |
+
"loss": 3.5226,
|
200 |
+
"rewards/accuracies": 0.6000000238418579,
|
201 |
+
"rewards/chosen": -3.9740614891052246,
|
202 |
+
"rewards/margins": 0.813327431678772,
|
203 |
+
"rewards/rejected": -4.787389278411865,
|
204 |
+
"step": 65
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 0.14963259853039412,
|
208 |
+
"grad_norm": 43.70954674356485,
|
209 |
+
"learning_rate": 9.926188266120295e-07,
|
210 |
+
"logits/chosen": -1.2420989274978638,
|
211 |
+
"logits/rejected": -1.211631178855896,
|
212 |
+
"logps/chosen": -0.43263110518455505,
|
213 |
+
"logps/rejected": -0.5376982688903809,
|
214 |
+
"loss": 3.6077,
|
215 |
+
"rewards/accuracies": 0.59375,
|
216 |
+
"rewards/chosen": -4.3263115882873535,
|
217 |
+
"rewards/margins": 1.0506718158721924,
|
218 |
+
"rewards/rejected": -5.376982688903809,
|
219 |
+
"step": 70
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.16032064128256512,
|
223 |
+
"grad_norm": 76.0561732944734,
|
224 |
+
"learning_rate": 9.890738003669027e-07,
|
225 |
+
"logits/chosen": -1.196466088294983,
|
226 |
+
"logits/rejected": -1.1188781261444092,
|
227 |
+
"logps/chosen": -0.4406164586544037,
|
228 |
+
"logps/rejected": -0.5367096662521362,
|
229 |
+
"loss": 3.5757,
|
230 |
+
"rewards/accuracies": 0.574999988079071,
|
231 |
+
"rewards/chosen": -4.40616512298584,
|
232 |
+
"rewards/margins": 0.9609323740005493,
|
233 |
+
"rewards/rejected": -5.3670973777771,
|
234 |
+
"step": 75
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 0.17100868403473615,
|
238 |
+
"grad_norm": 51.31544624480531,
|
239 |
+
"learning_rate": 9.848447601883433e-07,
|
240 |
+
"logits/chosen": -1.189353346824646,
|
241 |
+
"logits/rejected": -1.1714979410171509,
|
242 |
+
"logps/chosen": -0.38350600004196167,
|
243 |
+
"logps/rejected": -0.5140829682350159,
|
244 |
+
"loss": 3.4182,
|
245 |
+
"rewards/accuracies": 0.6625000238418579,
|
246 |
+
"rewards/chosen": -3.835059642791748,
|
247 |
+
"rewards/margins": 1.3057689666748047,
|
248 |
+
"rewards/rejected": -5.140829563140869,
|
249 |
+
"step": 80
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 0.18169672678690715,
|
253 |
+
"grad_norm": 72.27870189202869,
|
254 |
+
"learning_rate": 9.799376207714444e-07,
|
255 |
+
"logits/chosen": -1.2069041728973389,
|
256 |
+
"logits/rejected": -1.1819300651550293,
|
257 |
+
"logps/chosen": -0.4050617218017578,
|
258 |
+
"logps/rejected": -0.46999722719192505,
|
259 |
+
"loss": 3.4683,
|
260 |
+
"rewards/accuracies": 0.675000011920929,
|
261 |
+
"rewards/chosen": -4.05061674118042,
|
262 |
+
"rewards/margins": 0.6493551731109619,
|
263 |
+
"rewards/rejected": -4.699972152709961,
|
264 |
+
"step": 85
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 0.19238476953907815,
|
268 |
+
"grad_norm": 65.15839777923942,
|
269 |
+
"learning_rate": 9.743592451943998e-07,
|
270 |
+
"logits/chosen": -1.2592518329620361,
|
271 |
+
"logits/rejected": -1.2168278694152832,
|
272 |
+
"logps/chosen": -0.49755024909973145,
|
273 |
+
"logps/rejected": -0.6345678567886353,
|
274 |
+
"loss": 3.5477,
|
275 |
+
"rewards/accuracies": 0.612500011920929,
|
276 |
+
"rewards/chosen": -4.975502014160156,
|
277 |
+
"rewards/margins": 1.3701757192611694,
|
278 |
+
"rewards/rejected": -6.345677852630615,
|
279 |
+
"step": 90
|
280 |
+
},
|
281 |
+
{
|
282 |
+
"epoch": 0.20307281229124916,
|
283 |
+
"grad_norm": 86.65912439673568,
|
284 |
+
"learning_rate": 9.681174353198686e-07,
|
285 |
+
"logits/chosen": -1.324328064918518,
|
286 |
+
"logits/rejected": -1.2326462268829346,
|
287 |
+
"logps/chosen": -0.5199041962623596,
|
288 |
+
"logps/rejected": -0.5949305295944214,
|
289 |
+
"loss": 3.3966,
|
290 |
+
"rewards/accuracies": 0.668749988079071,
|
291 |
+
"rewards/chosen": -5.199041843414307,
|
292 |
+
"rewards/margins": 0.7502638697624207,
|
293 |
+
"rewards/rejected": -5.949306011199951,
|
294 |
+
"step": 95
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 0.21376085504342018,
|
298 |
+
"grad_norm": 84.91017386381003,
|
299 |
+
"learning_rate": 9.612209208833646e-07,
|
300 |
+
"logits/chosen": -1.223847508430481,
|
301 |
+
"logits/rejected": -1.193149209022522,
|
302 |
+
"logps/chosen": -0.5218724012374878,
|
303 |
+
"logps/rejected": -0.6197377443313599,
|
304 |
+
"loss": 3.4071,
|
305 |
+
"rewards/accuracies": 0.65625,
|
306 |
+
"rewards/chosen": -5.218724250793457,
|
307 |
+
"rewards/margins": 0.978653609752655,
|
308 |
+
"rewards/rejected": -6.197378158569336,
|
309 |
+
"step": 100
|
310 |
+
},
|
311 |
+
{
|
312 |
+
"epoch": 0.22444889779559118,
|
313 |
+
"grad_norm": 122.37976199382105,
|
314 |
+
"learning_rate": 9.536793472839324e-07,
|
315 |
+
"logits/chosen": -1.2392634153366089,
|
316 |
+
"logits/rejected": -1.178882360458374,
|
317 |
+
"logps/chosen": -0.47783392667770386,
|
318 |
+
"logps/rejected": -0.6210560202598572,
|
319 |
+
"loss": 3.3437,
|
320 |
+
"rewards/accuracies": 0.71875,
|
321 |
+
"rewards/chosen": -4.7783403396606445,
|
322 |
+
"rewards/margins": 1.432220220565796,
|
323 |
+
"rewards/rejected": -6.210559844970703,
|
324 |
+
"step": 105
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.23513694054776219,
|
328 |
+
"grad_norm": 117.75268692874619,
|
329 |
+
"learning_rate": 9.455032620941839e-07,
|
330 |
+
"logits/chosen": -1.2091577053070068,
|
331 |
+
"logits/rejected": -1.1453241109848022,
|
332 |
+
"logps/chosen": -0.5471813678741455,
|
333 |
+
"logps/rejected": -0.6967746615409851,
|
334 |
+
"loss": 3.3036,
|
335 |
+
"rewards/accuracies": 0.6812499761581421,
|
336 |
+
"rewards/chosen": -5.471813678741455,
|
337 |
+
"rewards/margins": 1.495932936668396,
|
338 |
+
"rewards/rejected": -6.967746734619141,
|
339 |
+
"step": 110
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 0.2458249832999332,
|
343 |
+
"grad_norm": 103.23266991280943,
|
344 |
+
"learning_rate": 9.367041003085648e-07,
|
345 |
+
"logits/chosen": -1.2898657321929932,
|
346 |
+
"logits/rejected": -1.224526047706604,
|
347 |
+
"logps/chosen": -0.5657516121864319,
|
348 |
+
"logps/rejected": -0.684479296207428,
|
349 |
+
"loss": 3.058,
|
350 |
+
"rewards/accuracies": 0.668749988079071,
|
351 |
+
"rewards/chosen": -5.657515525817871,
|
352 |
+
"rewards/margins": 1.18727707862854,
|
353 |
+
"rewards/rejected": -6.84479284286499,
|
354 |
+
"step": 115
|
355 |
+
},
|
356 |
+
{
|
357 |
+
"epoch": 0.2565130260521042,
|
358 |
+
"grad_norm": 87.5488547010946,
|
359 |
+
"learning_rate": 9.272941683504808e-07,
|
360 |
+
"logits/chosen": -1.2546017169952393,
|
361 |
+
"logits/rejected": -1.157871961593628,
|
362 |
+
"logps/chosen": -0.6318168640136719,
|
363 |
+
"logps/rejected": -0.8914499282836914,
|
364 |
+
"loss": 2.963,
|
365 |
+
"rewards/accuracies": 0.7749999761581421,
|
366 |
+
"rewards/chosen": -6.318168640136719,
|
367 |
+
"rewards/margins": 2.596330165863037,
|
368 |
+
"rewards/rejected": -8.914499282836914,
|
369 |
+
"step": 120
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 0.26720106880427524,
|
373 |
+
"grad_norm": 136.85902662680772,
|
374 |
+
"learning_rate": 9.172866268606513e-07,
|
375 |
+
"logits/chosen": -1.336275339126587,
|
376 |
+
"logits/rejected": -1.2872272729873657,
|
377 |
+
"logps/chosen": -0.6883447766304016,
|
378 |
+
"logps/rejected": -0.8685702085494995,
|
379 |
+
"loss": 2.782,
|
380 |
+
"rewards/accuracies": 0.7562500238418579,
|
381 |
+
"rewards/chosen": -6.883447170257568,
|
382 |
+
"rewards/margins": 1.8022544384002686,
|
383 |
+
"rewards/rejected": -8.685701370239258,
|
384 |
+
"step": 125
|
385 |
+
},
|
386 |
+
{
|
387 |
+
"epoch": 0.27788911155644624,
|
388 |
+
"grad_norm": 219.78249066687138,
|
389 |
+
"learning_rate": 9.066954722907638e-07,
|
390 |
+
"logits/chosen": -1.365232229232788,
|
391 |
+
"logits/rejected": -1.354945182800293,
|
392 |
+
"logps/chosen": -0.795397162437439,
|
393 |
+
"logps/rejected": -1.2445639371871948,
|
394 |
+
"loss": 2.7364,
|
395 |
+
"rewards/accuracies": 0.8062499761581421,
|
396 |
+
"rewards/chosen": -7.953972816467285,
|
397 |
+
"rewards/margins": 4.491666793823242,
|
398 |
+
"rewards/rejected": -12.445638656616211,
|
399 |
+
"step": 130
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 0.28857715430861725,
|
403 |
+
"grad_norm": 108.29484064425847,
|
404 |
+
"learning_rate": 8.955355173281707e-07,
|
405 |
+
"logits/chosen": -1.3360868692398071,
|
406 |
+
"logits/rejected": -1.2848550081253052,
|
407 |
+
"logps/chosen": -0.8393806219100952,
|
408 |
+
"logps/rejected": -1.0930553674697876,
|
409 |
+
"loss": 2.7236,
|
410 |
+
"rewards/accuracies": 0.7749999761581421,
|
411 |
+
"rewards/chosen": -8.393805503845215,
|
412 |
+
"rewards/margins": 2.536748170852661,
|
413 |
+
"rewards/rejected": -10.930554389953613,
|
414 |
+
"step": 135
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 0.29926519706078825,
|
418 |
+
"grad_norm": 111.40981489221316,
|
419 |
+
"learning_rate": 8.838223701790055e-07,
|
420 |
+
"logits/chosen": -1.4011452198028564,
|
421 |
+
"logits/rejected": -1.3746473789215088,
|
422 |
+
"logps/chosen": -0.8652567863464355,
|
423 |
+
"logps/rejected": -1.1113642454147339,
|
424 |
+
"loss": 2.6512,
|
425 |
+
"rewards/accuracies": 0.762499988079071,
|
426 |
+
"rewards/chosen": -8.652568817138672,
|
427 |
+
"rewards/margins": 2.4610743522644043,
|
428 |
+
"rewards/rejected": -11.113641738891602,
|
429 |
+
"step": 140
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.30995323981295925,
|
433 |
+
"grad_norm": 138.17602287907837,
|
434 |
+
"learning_rate": 8.71572412738697e-07,
|
435 |
+
"logits/chosen": -1.3082659244537354,
|
436 |
+
"logits/rejected": -1.2775059938430786,
|
437 |
+
"logps/chosen": -0.8726702928543091,
|
438 |
+
"logps/rejected": -1.1962165832519531,
|
439 |
+
"loss": 2.5784,
|
440 |
+
"rewards/accuracies": 0.7437499761581421,
|
441 |
+
"rewards/chosen": -8.726702690124512,
|
442 |
+
"rewards/margins": 3.2354626655578613,
|
443 |
+
"rewards/rejected": -11.962165832519531,
|
444 |
+
"step": 145
|
445 |
+
},
|
446 |
+
{
|
447 |
+
"epoch": 0.32064128256513025,
|
448 |
+
"grad_norm": 115.35023284841587,
|
449 |
+
"learning_rate": 8.588027776804058e-07,
|
450 |
+
"logits/chosen": -1.3314917087554932,
|
451 |
+
"logits/rejected": -1.306792974472046,
|
452 |
+
"logps/chosen": -1.0279521942138672,
|
453 |
+
"logps/rejected": -1.3382768630981445,
|
454 |
+
"loss": 2.47,
|
455 |
+
"rewards/accuracies": 0.793749988079071,
|
456 |
+
"rewards/chosen": -10.279521942138672,
|
457 |
+
"rewards/margins": 3.1032464504241943,
|
458 |
+
"rewards/rejected": -13.382769584655762,
|
459 |
+
"step": 150
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 0.33132932531730125,
|
463 |
+
"grad_norm": 144.67956167588085,
|
464 |
+
"learning_rate": 8.455313244934324e-07,
|
465 |
+
"logits/chosen": -1.3292877674102783,
|
466 |
+
"logits/rejected": -1.3038713932037354,
|
467 |
+
"logps/chosen": -1.035979986190796,
|
468 |
+
"logps/rejected": -1.4305903911590576,
|
469 |
+
"loss": 2.4435,
|
470 |
+
"rewards/accuracies": 0.7437499761581421,
|
471 |
+
"rewards/chosen": -10.359800338745117,
|
472 |
+
"rewards/margins": 3.946104049682617,
|
473 |
+
"rewards/rejected": -14.305903434753418,
|
474 |
+
"step": 155
|
475 |
+
},
|
476 |
+
{
|
477 |
+
"epoch": 0.3420173680694723,
|
478 |
+
"grad_norm": 106.03488572932915,
|
479 |
+
"learning_rate": 8.317766145051057e-07,
|
480 |
+
"logits/chosen": -1.3299791812896729,
|
481 |
+
"logits/rejected": -1.3083471059799194,
|
482 |
+
"logps/chosen": -1.1912587881088257,
|
483 |
+
"logps/rejected": -1.648496389389038,
|
484 |
+
"loss": 2.3299,
|
485 |
+
"rewards/accuracies": 0.7749999761581421,
|
486 |
+
"rewards/chosen": -11.912586212158203,
|
487 |
+
"rewards/margins": 4.5723772048950195,
|
488 |
+
"rewards/rejected": -16.484966278076172,
|
489 |
+
"step": 160
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 0.3527054108216433,
|
493 |
+
"grad_norm": 124.30842086260031,
|
494 |
+
"learning_rate": 8.175578849210894e-07,
|
495 |
+
"logits/chosen": -1.3360395431518555,
|
496 |
+
"logits/rejected": -1.3087151050567627,
|
497 |
+
"logps/chosen": -1.2806798219680786,
|
498 |
+
"logps/rejected": -1.7463128566741943,
|
499 |
+
"loss": 2.1935,
|
500 |
+
"rewards/accuracies": 0.75,
|
501 |
+
"rewards/chosen": -12.806798934936523,
|
502 |
+
"rewards/margins": 4.656330108642578,
|
503 |
+
"rewards/rejected": -17.463130950927734,
|
504 |
+
"step": 165
|
505 |
+
},
|
506 |
+
{
|
507 |
+
"epoch": 0.3633934535738143,
|
508 |
+
"grad_norm": 159.71509664144037,
|
509 |
+
"learning_rate": 8.028950219204099e-07,
|
510 |
+
"logits/chosen": -1.3390699625015259,
|
511 |
+
"logits/rejected": -1.3161685466766357,
|
512 |
+
"logps/chosen": -1.27903151512146,
|
513 |
+
"logps/rejected": -1.8541208505630493,
|
514 |
+
"loss": 2.1363,
|
515 |
+
"rewards/accuracies": 0.8187500238418579,
|
516 |
+
"rewards/chosen": -12.790315628051758,
|
517 |
+
"rewards/margins": 5.750893592834473,
|
518 |
+
"rewards/rejected": -18.541208267211914,
|
519 |
+
"step": 170
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 0.3740814963259853,
|
523 |
+
"grad_norm": 163.13967408983922,
|
524 |
+
"learning_rate": 7.878085328428368e-07,
|
525 |
+
"logits/chosen": -1.359133243560791,
|
526 |
+
"logits/rejected": -1.3104041814804077,
|
527 |
+
"logps/chosen": -1.4678348302841187,
|
528 |
+
"logps/rejected": -1.8786561489105225,
|
529 |
+
"loss": 2.0099,
|
530 |
+
"rewards/accuracies": 0.793749988079071,
|
531 |
+
"rewards/chosen": -14.678349494934082,
|
532 |
+
"rewards/margins": 4.108211517333984,
|
533 |
+
"rewards/rejected": -18.786561965942383,
|
534 |
+
"step": 175
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.3847695390781563,
|
538 |
+
"grad_norm": 213.96398552844718,
|
539 |
+
"learning_rate": 7.723195175075135e-07,
|
540 |
+
"logits/chosen": -1.3180896043777466,
|
541 |
+
"logits/rejected": -1.2993403673171997,
|
542 |
+
"logps/chosen": -1.395212173461914,
|
543 |
+
"logps/rejected": -1.9256620407104492,
|
544 |
+
"loss": 1.8566,
|
545 |
+
"rewards/accuracies": 0.800000011920929,
|
546 |
+
"rewards/chosen": -13.952122688293457,
|
547 |
+
"rewards/margins": 5.304497718811035,
|
548 |
+
"rewards/rejected": -19.25661849975586,
|
549 |
+
"step": 180
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 0.3954575818303273,
|
553 |
+
"grad_norm": 151.287610930137,
|
554 |
+
"learning_rate": 7.564496387029531e-07,
|
555 |
+
"logits/chosen": -1.354866623878479,
|
556 |
+
"logits/rejected": -1.2994146347045898,
|
557 |
+
"logps/chosen": -1.5145620107650757,
|
558 |
+
"logps/rejected": -2.0945541858673096,
|
559 |
+
"loss": 1.9317,
|
560 |
+
"rewards/accuracies": 0.856249988079071,
|
561 |
+
"rewards/chosen": -15.14561939239502,
|
562 |
+
"rewards/margins": 5.799921989440918,
|
563 |
+
"rewards/rejected": -20.945541381835938,
|
564 |
+
"step": 185
|
565 |
+
},
|
566 |
+
{
|
567 |
+
"epoch": 0.4061456245824983,
|
568 |
+
"grad_norm": 147.29173146280465,
|
569 |
+
"learning_rate": 7.402210918896689e-07,
|
570 |
+
"logits/chosen": -1.348362684249878,
|
571 |
+
"logits/rejected": -1.3590227365493774,
|
572 |
+
"logps/chosen": -1.7428133487701416,
|
573 |
+
"logps/rejected": -2.511070966720581,
|
574 |
+
"loss": 1.7298,
|
575 |
+
"rewards/accuracies": 0.862500011920929,
|
576 |
+
"rewards/chosen": -17.428131103515625,
|
577 |
+
"rewards/margins": 7.682579040527344,
|
578 |
+
"rewards/rejected": -25.1107120513916,
|
579 |
+
"step": 190
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 0.4168336673346693,
|
583 |
+
"grad_norm": 107.1032490301315,
|
584 |
+
"learning_rate": 7.236565741578162e-07,
|
585 |
+
"logits/chosen": -1.3077692985534668,
|
586 |
+
"logits/rejected": -1.2952277660369873,
|
587 |
+
"logps/chosen": -1.8426529169082642,
|
588 |
+
"logps/rejected": -2.392831325531006,
|
589 |
+
"loss": 1.9538,
|
590 |
+
"rewards/accuracies": 0.8062499761581421,
|
591 |
+
"rewards/chosen": -18.426528930664062,
|
592 |
+
"rewards/margins": 5.5017852783203125,
|
593 |
+
"rewards/rejected": -23.928314208984375,
|
594 |
+
"step": 195
|
595 |
+
},
|
596 |
+
{
|
597 |
+
"epoch": 0.42752171008684037,
|
598 |
+
"grad_norm": 190.9422038213508,
|
599 |
+
"learning_rate": 7.067792524832603e-07,
|
600 |
+
"logits/chosen": -1.3160083293914795,
|
601 |
+
"logits/rejected": -1.3034440279006958,
|
602 |
+
"logps/chosen": -1.7327241897583008,
|
603 |
+
"logps/rejected": -2.3605778217315674,
|
604 |
+
"loss": 1.6663,
|
605 |
+
"rewards/accuracies": 0.831250011920929,
|
606 |
+
"rewards/chosen": -17.327241897583008,
|
607 |
+
"rewards/margins": 6.278534889221191,
|
608 |
+
"rewards/rejected": -23.605777740478516,
|
609 |
+
"step": 200
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"epoch": 0.43820975283901137,
|
613 |
+
"grad_norm": 161.95141010920645,
|
614 |
+
"learning_rate": 6.896127313264642e-07,
|
615 |
+
"logits/chosen": -1.3677047491073608,
|
616 |
+
"logits/rejected": -1.3167794942855835,
|
617 |
+
"logps/chosen": -1.8963912725448608,
|
618 |
+
"logps/rejected": -2.526191234588623,
|
619 |
+
"loss": 1.8748,
|
620 |
+
"rewards/accuracies": 0.7875000238418579,
|
621 |
+
"rewards/chosen": -18.96391487121582,
|
622 |
+
"rewards/margins": 6.297998905181885,
|
623 |
+
"rewards/rejected": -25.261911392211914,
|
624 |
+
"step": 205
|
625 |
+
},
|
626 |
+
{
|
627 |
+
"epoch": 0.44889779559118237,
|
628 |
+
"grad_norm": 150.6462623725548,
|
629 |
+
"learning_rate": 6.721810196195174e-07,
|
630 |
+
"logits/chosen": -1.400187373161316,
|
631 |
+
"logits/rejected": -1.3892953395843506,
|
632 |
+
"logps/chosen": -2.041691541671753,
|
633 |
+
"logps/rejected": -2.6361751556396484,
|
634 |
+
"loss": 1.827,
|
635 |
+
"rewards/accuracies": 0.8500000238418579,
|
636 |
+
"rewards/chosen": -20.416913986206055,
|
637 |
+
"rewards/margins": 5.9448370933532715,
|
638 |
+
"rewards/rejected": -26.361751556396484,
|
639 |
+
"step": 210
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.45958583834335337,
|
643 |
+
"grad_norm": 126.80315819769801,
|
644 |
+
"learning_rate": 6.545084971874736e-07,
|
645 |
+
"logits/chosen": -1.3723266124725342,
|
646 |
+
"logits/rejected": -1.356567144393921,
|
647 |
+
"logps/chosen": -2.0259296894073486,
|
648 |
+
"logps/rejected": -2.754607677459717,
|
649 |
+
"loss": 1.6874,
|
650 |
+
"rewards/accuracies": 0.8062499761581421,
|
651 |
+
"rewards/chosen": -20.259296417236328,
|
652 |
+
"rewards/margins": 7.286777496337891,
|
653 |
+
"rewards/rejected": -27.546072006225586,
|
654 |
+
"step": 215
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 0.47027388109552437,
|
658 |
+
"grad_norm": 161.97036694522683,
|
659 |
+
"learning_rate": 6.3661988065096e-07,
|
660 |
+
"logits/chosen": -1.4270765781402588,
|
661 |
+
"logits/rejected": -1.4134238958358765,
|
662 |
+
"logps/chosen": -2.0706095695495605,
|
663 |
+
"logps/rejected": -2.8078479766845703,
|
664 |
+
"loss": 1.5613,
|
665 |
+
"rewards/accuracies": 0.800000011920929,
|
666 |
+
"rewards/chosen": -20.706096649169922,
|
667 |
+
"rewards/margins": 7.372382164001465,
|
668 |
+
"rewards/rejected": -28.078479766845703,
|
669 |
+
"step": 220
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 0.48096192384769537,
|
673 |
+
"grad_norm": 119.50283654492473,
|
674 |
+
"learning_rate": 6.185401888577487e-07,
|
675 |
+
"logits/chosen": -1.4146029949188232,
|
676 |
+
"logits/rejected": -1.3837422132492065,
|
677 |
+
"logps/chosen": -2.119856595993042,
|
678 |
+
"logps/rejected": -2.7663817405700684,
|
679 |
+
"loss": 1.7069,
|
680 |
+
"rewards/accuracies": 0.7875000238418579,
|
681 |
+
"rewards/chosen": -21.198566436767578,
|
682 |
+
"rewards/margins": 6.465248107910156,
|
683 |
+
"rewards/rejected": -27.663818359375,
|
684 |
+
"step": 225
|
685 |
+
},
|
686 |
+
{
|
687 |
+
"epoch": 0.4916499665998664,
|
688 |
+
"grad_norm": 109.77939760201329,
|
689 |
+
"learning_rate": 6.002947078916364e-07,
|
690 |
+
"logits/chosen": -1.4831273555755615,
|
691 |
+
"logits/rejected": -1.435591459274292,
|
692 |
+
"logps/chosen": -2.0505659580230713,
|
693 |
+
"logps/rejected": -2.685992956161499,
|
694 |
+
"loss": 1.5078,
|
695 |
+
"rewards/accuracies": 0.8187500238418579,
|
696 |
+
"rewards/chosen": -20.505659103393555,
|
697 |
+
"rewards/margins": 6.354269027709961,
|
698 |
+
"rewards/rejected": -26.859928131103516,
|
699 |
+
"step": 230
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 0.5023380093520374,
|
703 |
+
"grad_norm": 147.37811459958783,
|
704 |
+
"learning_rate": 5.819089557075688e-07,
|
705 |
+
"logits/chosen": -1.4949449300765991,
|
706 |
+
"logits/rejected": -1.469238042831421,
|
707 |
+
"logps/chosen": -2.045539617538452,
|
708 |
+
"logps/rejected": -2.816204071044922,
|
709 |
+
"loss": 1.4911,
|
710 |
+
"rewards/accuracies": 0.800000011920929,
|
711 |
+
"rewards/chosen": -20.455394744873047,
|
712 |
+
"rewards/margins": 7.706644535064697,
|
713 |
+
"rewards/rejected": -28.162038803100586,
|
714 |
+
"step": 235
|
715 |
+
},
|
716 |
+
{
|
717 |
+
"epoch": 0.5130260521042084,
|
718 |
+
"grad_norm": 132.440512722152,
|
719 |
+
"learning_rate": 5.634086464424742e-07,
|
720 |
+
"logits/chosen": -1.461572289466858,
|
721 |
+
"logits/rejected": -1.4638675451278687,
|
722 |
+
"logps/chosen": -1.9458844661712646,
|
723 |
+
"logps/rejected": -2.7108004093170166,
|
724 |
+
"loss": 1.5773,
|
725 |
+
"rewards/accuracies": 0.824999988079071,
|
726 |
+
"rewards/chosen": -19.458845138549805,
|
727 |
+
"rewards/margins": 7.6491570472717285,
|
728 |
+
"rewards/rejected": -27.10800552368164,
|
729 |
+
"step": 240
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 0.5237140948563794,
|
733 |
+
"grad_norm": 236.09806301372015,
|
734 |
+
"learning_rate": 5.448196544517167e-07,
|
735 |
+
"logits/chosen": -1.5521364212036133,
|
736 |
+
"logits/rejected": -1.4946634769439697,
|
737 |
+
"logps/chosen": -2.0292551517486572,
|
738 |
+
"logps/rejected": -2.9242515563964844,
|
739 |
+
"loss": 1.395,
|
740 |
+
"rewards/accuracies": 0.8812500238418579,
|
741 |
+
"rewards/chosen": -20.292551040649414,
|
742 |
+
"rewards/margins": 8.949962615966797,
|
743 |
+
"rewards/rejected": -29.242517471313477,
|
744 |
+
"step": 245
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 0.5344021376085505,
|
748 |
+
"grad_norm": 142.79859396103615,
|
749 |
+
"learning_rate": 5.26167978121472e-07,
|
750 |
+
"logits/chosen": -1.4808663129806519,
|
751 |
+
"logits/rejected": -1.4642443656921387,
|
752 |
+
"logps/chosen": -2.046252489089966,
|
753 |
+
"logps/rejected": -2.9118173122406006,
|
754 |
+
"loss": 1.3738,
|
755 |
+
"rewards/accuracies": 0.8687499761581421,
|
756 |
+
"rewards/chosen": -20.462526321411133,
|
757 |
+
"rewards/margins": 8.655647277832031,
|
758 |
+
"rewards/rejected": -29.1181697845459,
|
759 |
+
"step": 250
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 0.5450901803607214,
|
763 |
+
"grad_norm": 155.0378761351227,
|
764 |
+
"learning_rate": 5.074797035076318e-07,
|
765 |
+
"logits/chosen": -1.5233964920043945,
|
766 |
+
"logits/rejected": -1.5014541149139404,
|
767 |
+
"logps/chosen": -2.2131590843200684,
|
768 |
+
"logps/rejected": -2.9324965476989746,
|
769 |
+
"loss": 1.5353,
|
770 |
+
"rewards/accuracies": 0.8374999761581421,
|
771 |
+
"rewards/chosen": -22.131589889526367,
|
772 |
+
"rewards/margins": 7.193373680114746,
|
773 |
+
"rewards/rejected": -29.324962615966797,
|
774 |
+
"step": 255
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 0.5557782231128925,
|
778 |
+
"grad_norm": 130.04315061980418,
|
779 |
+
"learning_rate": 4.887809678520975e-07,
|
780 |
+
"logits/chosen": -1.5009315013885498,
|
781 |
+
"logits/rejected": -1.474130630493164,
|
782 |
+
"logps/chosen": -2.1018893718719482,
|
783 |
+
"logps/rejected": -2.8382973670959473,
|
784 |
+
"loss": 1.5292,
|
785 |
+
"rewards/accuracies": 0.862500011920929,
|
786 |
+
"rewards/chosen": -21.018896102905273,
|
787 |
+
"rewards/margins": 7.364079475402832,
|
788 |
+
"rewards/rejected": -28.382976531982422,
|
789 |
+
"step": 260
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 0.5664662658650634,
|
793 |
+
"grad_norm": 119.27303382130957,
|
794 |
+
"learning_rate": 4.700979230274829e-07,
|
795 |
+
"logits/chosen": -1.4742642641067505,
|
796 |
+
"logits/rejected": -1.4596500396728516,
|
797 |
+
"logps/chosen": -2.2155094146728516,
|
798 |
+
"logps/rejected": -3.0179553031921387,
|
799 |
+
"loss": 1.1693,
|
800 |
+
"rewards/accuracies": 0.862500011920929,
|
801 |
+
"rewards/chosen": -22.15509605407715,
|
802 |
+
"rewards/margins": 8.024459838867188,
|
803 |
+
"rewards/rejected": -30.179553985595703,
|
804 |
+
"step": 265
|
805 |
+
},
|
806 |
+
{
|
807 |
+
"epoch": 0.5771543086172345,
|
808 |
+
"grad_norm": 131.97372035922703,
|
809 |
+
"learning_rate": 4.514566989613559e-07,
|
810 |
+
"logits/chosen": -1.4763275384902954,
|
811 |
+
"logits/rejected": -1.4486229419708252,
|
812 |
+
"logps/chosen": -1.990666151046753,
|
813 |
+
"logps/rejected": -2.838904857635498,
|
814 |
+
"loss": 1.3013,
|
815 |
+
"rewards/accuracies": 0.875,
|
816 |
+
"rewards/chosen": -19.906661987304688,
|
817 |
+
"rewards/margins": 8.482388496398926,
|
818 |
+
"rewards/rejected": -28.389049530029297,
|
819 |
+
"step": 270
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 0.5878423513694054,
|
823 |
+
"grad_norm": 104.56387348722761,
|
824 |
+
"learning_rate": 4.328833670911724e-07,
|
825 |
+
"logits/chosen": -1.4566619396209717,
|
826 |
+
"logits/rejected": -1.4216809272766113,
|
827 |
+
"logps/chosen": -1.9770195484161377,
|
828 |
+
"logps/rejected": -2.719089984893799,
|
829 |
+
"loss": 1.4258,
|
830 |
+
"rewards/accuracies": 0.831250011920929,
|
831 |
+
"rewards/chosen": -19.77019500732422,
|
832 |
+
"rewards/margins": 7.4207048416137695,
|
833 |
+
"rewards/rejected": -27.190898895263672,
|
834 |
+
"step": 275
|
835 |
+
},
|
836 |
+
{
|
837 |
+
"epoch": 0.5985303941215765,
|
838 |
+
"grad_norm": 133.9300472658157,
|
839 |
+
"learning_rate": 4.144039039010124e-07,
|
840 |
+
"logits/chosen": -1.542373776435852,
|
841 |
+
"logits/rejected": -1.5213029384613037,
|
842 |
+
"logps/chosen": -2.0355706214904785,
|
843 |
+
"logps/rejected": -2.883401393890381,
|
844 |
+
"loss": 1.3022,
|
845 |
+
"rewards/accuracies": 0.831250011920929,
|
846 |
+
"rewards/chosen": -20.3557071685791,
|
847 |
+
"rewards/margins": 8.478304862976074,
|
848 |
+
"rewards/rejected": -28.834014892578125,
|
849 |
+
"step": 280
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.6092184368737475,
|
853 |
+
"grad_norm": 118.2550806898935,
|
854 |
+
"learning_rate": 3.960441545911204e-07,
|
855 |
+
"logits/chosen": -1.501997947692871,
|
856 |
+
"logits/rejected": -1.4718389511108398,
|
857 |
+
"logps/chosen": -2.043879270553589,
|
858 |
+
"logps/rejected": -2.9162051677703857,
|
859 |
+
"loss": 1.0025,
|
860 |
+
"rewards/accuracies": 0.8999999761581421,
|
861 |
+
"rewards/chosen": -20.438793182373047,
|
862 |
+
"rewards/margins": 8.723258972167969,
|
863 |
+
"rewards/rejected": -29.16205406188965,
|
864 |
+
"step": 285
|
865 |
+
},
|
866 |
+
{
|
867 |
+
"epoch": 0.6199064796259185,
|
868 |
+
"grad_norm": 112.08802157558269,
|
869 |
+
"learning_rate": 3.778297969310529e-07,
|
870 |
+
"logits/chosen": -1.5190720558166504,
|
871 |
+
"logits/rejected": -1.4788782596588135,
|
872 |
+
"logps/chosen": -2.014256000518799,
|
873 |
+
"logps/rejected": -2.7785563468933105,
|
874 |
+
"loss": 1.3157,
|
875 |
+
"rewards/accuracies": 0.831250011920929,
|
876 |
+
"rewards/chosen": -20.142559051513672,
|
877 |
+
"rewards/margins": 7.64300537109375,
|
878 |
+
"rewards/rejected": -27.785564422607422,
|
879 |
+
"step": 290
|
880 |
+
},
|
881 |
+
{
|
882 |
+
"epoch": 0.6305945223780896,
|
883 |
+
"grad_norm": 122.16012718306597,
|
884 |
+
"learning_rate": 3.5978630534699865e-07,
|
885 |
+
"logits/chosen": -1.4519203901290894,
|
886 |
+
"logits/rejected": -1.4377477169036865,
|
887 |
+
"logps/chosen": -2.052429437637329,
|
888 |
+
"logps/rejected": -2.7946228981018066,
|
889 |
+
"loss": 1.2035,
|
890 |
+
"rewards/accuracies": 0.8125,
|
891 |
+
"rewards/chosen": -20.524295806884766,
|
892 |
+
"rewards/margins": 7.421932220458984,
|
893 |
+
"rewards/rejected": -27.94622802734375,
|
894 |
+
"step": 295
|
895 |
+
},
|
896 |
+
{
|
897 |
+
"epoch": 0.6412825651302605,
|
898 |
+
"grad_norm": 127.59960775053135,
|
899 |
+
"learning_rate": 3.4193891529348795e-07,
|
900 |
+
"logits/chosen": -1.3882105350494385,
|
901 |
+
"logits/rejected": -1.365216612815857,
|
902 |
+
"logps/chosen": -2.0868842601776123,
|
903 |
+
"logps/rejected": -2.7921385765075684,
|
904 |
+
"loss": 1.5608,
|
905 |
+
"rewards/accuracies": 0.8062499761581421,
|
906 |
+
"rewards/chosen": -20.86884307861328,
|
907 |
+
"rewards/margins": 7.0525383949279785,
|
908 |
+
"rewards/rejected": -27.9213809967041,
|
909 |
+
"step": 300
|
910 |
+
},
|
911 |
+
{
|
912 |
+
"epoch": 0.6519706078824316,
|
913 |
+
"grad_norm": 104.5591334100815,
|
914 |
+
"learning_rate": 3.243125879593286e-07,
|
915 |
+
"logits/chosen": -1.5032074451446533,
|
916 |
+
"logits/rejected": -1.458888053894043,
|
917 |
+
"logps/chosen": -2.047602415084839,
|
918 |
+
"logps/rejected": -2.731414794921875,
|
919 |
+
"loss": 1.3552,
|
920 |
+
"rewards/accuracies": 0.8125,
|
921 |
+
"rewards/chosen": -20.476024627685547,
|
922 |
+
"rewards/margins": 6.8381242752075195,
|
923 |
+
"rewards/rejected": -27.31414794921875,
|
924 |
+
"step": 305
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 0.6626586506346025,
|
928 |
+
"grad_norm": 128.9435075925882,
|
929 |
+
"learning_rate": 3.069319753571269e-07,
|
930 |
+
"logits/chosen": -1.5344185829162598,
|
931 |
+
"logits/rejected": -1.520206093788147,
|
932 |
+
"logps/chosen": -2.182490587234497,
|
933 |
+
"logps/rejected": -2.994981288909912,
|
934 |
+
"loss": 1.3424,
|
935 |
+
"rewards/accuracies": 0.90625,
|
936 |
+
"rewards/chosen": -21.824905395507812,
|
937 |
+
"rewards/margins": 8.124906539916992,
|
938 |
+
"rewards/rejected": -29.949810028076172,
|
939 |
+
"step": 310
|
940 |
+
},
|
941 |
+
{
|
942 |
+
"epoch": 0.6733466933867736,
|
943 |
+
"grad_norm": 117.862174484777,
|
944 |
+
"learning_rate": 2.898213858452173e-07,
|
945 |
+
"logits/chosen": -1.5316812992095947,
|
946 |
+
"logits/rejected": -1.4777030944824219,
|
947 |
+
"logps/chosen": -2.1463074684143066,
|
948 |
+
"logps/rejected": -2.8947274684906006,
|
949 |
+
"loss": 1.2773,
|
950 |
+
"rewards/accuracies": 0.856249988079071,
|
951 |
+
"rewards/chosen": -21.46307373046875,
|
952 |
+
"rewards/margins": 7.484200954437256,
|
953 |
+
"rewards/rejected": -28.947277069091797,
|
954 |
+
"step": 315
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 0.6840347361389446,
|
958 |
+
"grad_norm": 154.98050630680447,
|
959 |
+
"learning_rate": 2.730047501302266e-07,
|
960 |
+
"logits/chosen": -1.5196678638458252,
|
961 |
+
"logits/rejected": -1.5156781673431396,
|
962 |
+
"logps/chosen": -2.249702215194702,
|
963 |
+
"logps/rejected": -3.2510273456573486,
|
964 |
+
"loss": 1.0665,
|
965 |
+
"rewards/accuracies": 0.893750011920929,
|
966 |
+
"rewards/chosen": -22.497024536132812,
|
967 |
+
"rewards/margins": 10.013248443603516,
|
968 |
+
"rewards/rejected": -32.51027297973633,
|
969 |
+
"step": 320
|
970 |
+
},
|
971 |
+
{
|
972 |
+
"epoch": 0.6947227788911156,
|
973 |
+
"grad_norm": 180.45567365921923,
|
974 |
+
"learning_rate": 2.5650558779781635e-07,
|
975 |
+
"logits/chosen": -1.549286127090454,
|
976 |
+
"logits/rejected": -1.5002086162567139,
|
977 |
+
"logps/chosen": -2.399672031402588,
|
978 |
+
"logps/rejected": -3.504227876663208,
|
979 |
+
"loss": 1.2384,
|
980 |
+
"rewards/accuracies": 0.8812500238418579,
|
981 |
+
"rewards/chosen": -23.996719360351562,
|
982 |
+
"rewards/margins": 11.045561790466309,
|
983 |
+
"rewards/rejected": -35.04228210449219,
|
984 |
+
"step": 325
|
985 |
+
},
|
986 |
+
{
|
987 |
+
"epoch": 0.7054108216432866,
|
988 |
+
"grad_norm": 105.90784807533655,
|
989 |
+
"learning_rate": 2.403469744184154e-07,
|
990 |
+
"logits/chosen": -1.4591710567474365,
|
991 |
+
"logits/rejected": -1.4205106496810913,
|
992 |
+
"logps/chosen": -2.2293872833251953,
|
993 |
+
"logps/rejected": -3.00614857673645,
|
994 |
+
"loss": 1.2359,
|
995 |
+
"rewards/accuracies": 0.862500011920929,
|
996 |
+
"rewards/chosen": -22.293874740600586,
|
997 |
+
"rewards/margins": 7.767613410949707,
|
998 |
+
"rewards/rejected": -30.061487197875977,
|
999 |
+
"step": 330
|
1000 |
+
},
|
1001 |
+
{
|
1002 |
+
"epoch": 0.7160988643954576,
|
1003 |
+
"grad_norm": 124.3214763138239,
|
1004 |
+
"learning_rate": 2.2455150927394878e-07,
|
1005 |
+
"logits/chosen": -1.495527744293213,
|
1006 |
+
"logits/rejected": -1.4803019762039185,
|
1007 |
+
"logps/chosen": -2.202894449234009,
|
1008 |
+
"logps/rejected": -3.0441994667053223,
|
1009 |
+
"loss": 1.1359,
|
1010 |
+
"rewards/accuracies": 0.862500011920929,
|
1011 |
+
"rewards/chosen": -22.02894401550293,
|
1012 |
+
"rewards/margins": 8.413049697875977,
|
1013 |
+
"rewards/rejected": -30.441997528076172,
|
1014 |
+
"step": 335
|
1015 |
+
},
|
1016 |
+
{
|
1017 |
+
"epoch": 0.7267869071476286,
|
1018 |
+
"grad_norm": 156.03049853633283,
|
1019 |
+
"learning_rate": 2.0914128375069722e-07,
|
1020 |
+
"logits/chosen": -1.5111547708511353,
|
1021 |
+
"logits/rejected": -1.4789291620254517,
|
1022 |
+
"logps/chosen": -2.1601741313934326,
|
1023 |
+
"logps/rejected": -3.021275281906128,
|
1024 |
+
"loss": 1.2419,
|
1025 |
+
"rewards/accuracies": 0.8687499761581421,
|
1026 |
+
"rewards/chosen": -21.601741790771484,
|
1027 |
+
"rewards/margins": 8.611013412475586,
|
1028 |
+
"rewards/rejected": -30.212757110595703,
|
1029 |
+
"step": 340
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"epoch": 0.7374749498997996,
|
1033 |
+
"grad_norm": 130.76746865900543,
|
1034 |
+
"learning_rate": 1.9413785044249676e-07,
|
1035 |
+
"logits/chosen": -1.5371043682098389,
|
1036 |
+
"logits/rejected": -1.5177440643310547,
|
1037 |
+
"logps/chosen": -2.2243714332580566,
|
1038 |
+
"logps/rejected": -3.2498092651367188,
|
1039 |
+
"loss": 1.2299,
|
1040 |
+
"rewards/accuracies": 0.90625,
|
1041 |
+
"rewards/chosen": -22.243711471557617,
|
1042 |
+
"rewards/margins": 10.25438117980957,
|
1043 |
+
"rewards/rejected": -32.49809265136719,
|
1044 |
+
"step": 345
|
1045 |
+
},
|
1046 |
+
{
|
1047 |
+
"epoch": 0.7481629926519706,
|
1048 |
+
"grad_norm": 219.49500818808932,
|
1049 |
+
"learning_rate": 1.7956219300748792e-07,
|
1050 |
+
"logits/chosen": -1.5225059986114502,
|
1051 |
+
"logits/rejected": -1.5286823511123657,
|
1052 |
+
"logps/chosen": -2.12577486038208,
|
1053 |
+
"logps/rejected": -3.068633794784546,
|
1054 |
+
"loss": 1.164,
|
1055 |
+
"rewards/accuracies": 0.887499988079071,
|
1056 |
+
"rewards/chosen": -21.257749557495117,
|
1057 |
+
"rewards/margins": 9.428587913513184,
|
1058 |
+
"rewards/rejected": -30.686336517333984,
|
1059 |
+
"step": 350
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 0.7588510354041417,
|
1063 |
+
"grad_norm": 148.30498812771748,
|
1064 |
+
"learning_rate": 1.6543469682057104e-07,
|
1065 |
+
"logits/chosen": -1.4495574235916138,
|
1066 |
+
"logits/rejected": -1.4652235507965088,
|
1067 |
+
"logps/chosen": -2.066810131072998,
|
1068 |
+
"logps/rejected": -2.938152313232422,
|
1069 |
+
"loss": 1.1351,
|
1070 |
+
"rewards/accuracies": 0.8812500238418579,
|
1071 |
+
"rewards/chosen": -20.668102264404297,
|
1072 |
+
"rewards/margins": 8.713422775268555,
|
1073 |
+
"rewards/rejected": -29.38152503967285,
|
1074 |
+
"step": 355
|
1075 |
+
},
|
1076 |
+
{
|
1077 |
+
"epoch": 0.7695390781563126,
|
1078 |
+
"grad_norm": 116.32905868645479,
|
1079 |
+
"learning_rate": 1.5177512046261666e-07,
|
1080 |
+
"logits/chosen": -1.5009536743164062,
|
1081 |
+
"logits/rejected": -1.5022971630096436,
|
1082 |
+
"logps/chosen": -2.0494394302368164,
|
1083 |
+
"logps/rejected": -3.0328898429870605,
|
1084 |
+
"loss": 1.1972,
|
1085 |
+
"rewards/accuracies": 0.8687499761581421,
|
1086 |
+
"rewards/chosen": -20.494396209716797,
|
1087 |
+
"rewards/margins": 9.834505081176758,
|
1088 |
+
"rewards/rejected": -30.328899383544922,
|
1089 |
+
"step": 360
|
1090 |
+
},
|
1091 |
+
{
|
1092 |
+
"epoch": 0.7802271209084837,
|
1093 |
+
"grad_norm": 108.7134031593527,
|
1094 |
+
"learning_rate": 1.3860256808630427e-07,
|
1095 |
+
"logits/chosen": -1.5464454889297485,
|
1096 |
+
"logits/rejected": -1.4812748432159424,
|
1097 |
+
"logps/chosen": -2.1187500953674316,
|
1098 |
+
"logps/rejected": -3.0532031059265137,
|
1099 |
+
"loss": 1.253,
|
1100 |
+
"rewards/accuracies": 0.8500000238418579,
|
1101 |
+
"rewards/chosen": -21.187496185302734,
|
1102 |
+
"rewards/margins": 9.344534873962402,
|
1103 |
+
"rewards/rejected": -30.532033920288086,
|
1104 |
+
"step": 365
|
1105 |
+
},
|
1106 |
+
{
|
1107 |
+
"epoch": 0.7909151636606546,
|
1108 |
+
"grad_norm": 145.2134549922322,
|
1109 |
+
"learning_rate": 1.2593546269723647e-07,
|
1110 |
+
"logits/chosen": -1.474498987197876,
|
1111 |
+
"logits/rejected": -1.4640263319015503,
|
1112 |
+
"logps/chosen": -2.100693941116333,
|
1113 |
+
"logps/rejected": -2.9064152240753174,
|
1114 |
+
"loss": 1.1015,
|
1115 |
+
"rewards/accuracies": 0.887499988079071,
|
1116 |
+
"rewards/chosen": -21.00693702697754,
|
1117 |
+
"rewards/margins": 8.05721378326416,
|
1118 |
+
"rewards/rejected": -29.064151763916016,
|
1119 |
+
"step": 370
|
1120 |
+
},
|
1121 |
+
{
|
1122 |
+
"epoch": 0.8016032064128257,
|
1123 |
+
"grad_norm": 130.4190340462438,
|
1124 |
+
"learning_rate": 1.1379152038770029e-07,
|
1125 |
+
"logits/chosen": -1.4943348169326782,
|
1126 |
+
"logits/rejected": -1.4995317459106445,
|
1127 |
+
"logps/chosen": -2.2358219623565674,
|
1128 |
+
"logps/rejected": -3.1511778831481934,
|
1129 |
+
"loss": 1.1776,
|
1130 |
+
"rewards/accuracies": 0.887499988079071,
|
1131 |
+
"rewards/chosen": -22.35822105407715,
|
1132 |
+
"rewards/margins": 9.153559684753418,
|
1133 |
+
"rewards/rejected": -31.51177978515625,
|
1134 |
+
"step": 375
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"epoch": 0.8122912491649966,
|
1138 |
+
"grad_norm": 171.4769450829484,
|
1139 |
+
"learning_rate": 1.0218772555910954e-07,
|
1140 |
+
"logits/chosen": -1.5206212997436523,
|
1141 |
+
"logits/rejected": -1.5025417804718018,
|
1142 |
+
"logps/chosen": -2.1182684898376465,
|
1143 |
+
"logps/rejected": -2.9665935039520264,
|
1144 |
+
"loss": 1.2678,
|
1145 |
+
"rewards/accuracies": 0.8687499761581421,
|
1146 |
+
"rewards/chosen": -21.18268394470215,
|
1147 |
+
"rewards/margins": 8.48325252532959,
|
1148 |
+
"rewards/rejected": -29.665935516357422,
|
1149 |
+
"step": 380
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 0.8229792919171677,
|
1153 |
+
"grad_norm": 116.74061042296164,
|
1154 |
+
"learning_rate": 9.114030716778432e-08,
|
1155 |
+
"logits/chosen": -1.5152442455291748,
|
1156 |
+
"logits/rejected": -1.4944274425506592,
|
1157 |
+
"logps/chosen": -2.0695395469665527,
|
1158 |
+
"logps/rejected": -3.1040778160095215,
|
1159 |
+
"loss": 1.0078,
|
1160 |
+
"rewards/accuracies": 0.8812500238418579,
|
1161 |
+
"rewards/chosen": -20.69539451599121,
|
1162 |
+
"rewards/margins": 10.345380783081055,
|
1163 |
+
"rewards/rejected": -31.0407772064209,
|
1164 |
+
"step": 385
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 0.8336673346693386,
|
1168 |
+
"grad_norm": 112.33137236518094,
|
1169 |
+
"learning_rate": 8.066471602728803e-08,
|
1170 |
+
"logits/chosen": -1.5178992748260498,
|
1171 |
+
"logits/rejected": -1.5032987594604492,
|
1172 |
+
"logps/chosen": -2.170666456222534,
|
1173 |
+
"logps/rejected": -3.162553310394287,
|
1174 |
+
"loss": 1.2224,
|
1175 |
+
"rewards/accuracies": 0.8687499761581421,
|
1176 |
+
"rewards/chosen": -21.7066650390625,
|
1177 |
+
"rewards/margins": 9.918868064880371,
|
1178 |
+
"rewards/rejected": -31.62552833557129,
|
1179 |
+
"step": 390
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 0.8443553774215097,
|
1183 |
+
"grad_norm": 186.4853149645633,
|
1184 |
+
"learning_rate": 7.077560319906694e-08,
|
1185 |
+
"logits/chosen": -1.5229237079620361,
|
1186 |
+
"logits/rejected": -1.505030632019043,
|
1187 |
+
"logps/chosen": -2.0735864639282227,
|
1188 |
+
"logps/rejected": -2.9833900928497314,
|
1189 |
+
"loss": 1.1526,
|
1190 |
+
"rewards/accuracies": 0.856249988079071,
|
1191 |
+
"rewards/chosen": -20.73586654663086,
|
1192 |
+
"rewards/margins": 9.098031997680664,
|
1193 |
+
"rewards/rejected": -29.833898544311523,
|
1194 |
+
"step": 395
|
1195 |
+
},
|
1196 |
+
{
|
1197 |
+
"epoch": 0.8550434201736807,
|
1198 |
+
"grad_norm": 96.6762011696182,
|
1199 |
+
"learning_rate": 6.148679950161672e-08,
|
1200 |
+
"logits/chosen": -1.5230014324188232,
|
1201 |
+
"logits/rejected": -1.5108994245529175,
|
1202 |
+
"logps/chosen": -2.109778642654419,
|
1203 |
+
"logps/rejected": -2.941293716430664,
|
1204 |
+
"loss": 1.1544,
|
1205 |
+
"rewards/accuracies": 0.862500011920929,
|
1206 |
+
"rewards/chosen": -21.09778594970703,
|
1207 |
+
"rewards/margins": 8.315154075622559,
|
1208 |
+
"rewards/rejected": -29.412939071655273,
|
1209 |
+
"step": 400
|
1210 |
+
},
|
1211 |
+
{
|
1212 |
+
"epoch": 0.8550434201736807,
|
1213 |
+
"eval_logits/chosen": -1.710108757019043,
|
1214 |
+
"eval_logits/rejected": -1.7197374105453491,
|
1215 |
+
"eval_logps/chosen": -2.08453369140625,
|
1216 |
+
"eval_logps/rejected": -2.94063138961792,
|
1217 |
+
"eval_loss": 1.1388919353485107,
|
1218 |
+
"eval_rewards/accuracies": 0.8678861856460571,
|
1219 |
+
"eval_rewards/chosen": -20.845340728759766,
|
1220 |
+
"eval_rewards/margins": 8.560976028442383,
|
1221 |
+
"eval_rewards/rejected": -29.406312942504883,
|
1222 |
+
"eval_runtime": 123.1178,
|
1223 |
+
"eval_samples_per_second": 15.928,
|
1224 |
+
"eval_steps_per_second": 0.999,
|
1225 |
+
"step": 400
|
1226 |
+
},
|
1227 |
+
{
|
1228 |
+
"epoch": 0.8657314629258517,
|
1229 |
+
"grad_norm": 143.6008976778444,
|
1230 |
+
"learning_rate": 5.2811296166831666e-08,
|
1231 |
+
"logits/chosen": -1.4884498119354248,
|
1232 |
+
"logits/rejected": -1.508371114730835,
|
1233 |
+
"logps/chosen": -2.2530012130737305,
|
1234 |
+
"logps/rejected": -3.071047782897949,
|
1235 |
+
"loss": 1.1786,
|
1236 |
+
"rewards/accuracies": 0.887499988079071,
|
1237 |
+
"rewards/chosen": -22.530010223388672,
|
1238 |
+
"rewards/margins": 8.18046760559082,
|
1239 |
+
"rewards/rejected": -30.710479736328125,
|
1240 |
+
"step": 405
|
1241 |
+
},
|
1242 |
+
{
|
1243 |
+
"epoch": 0.8764195056780227,
|
1244 |
+
"grad_norm": 145.0618422989251,
|
1245 |
+
"learning_rate": 4.4761226670592066e-08,
|
1246 |
+
"logits/chosen": -1.4991635084152222,
|
1247 |
+
"logits/rejected": -1.4899944067001343,
|
1248 |
+
"logps/chosen": -2.1319069862365723,
|
1249 |
+
"logps/rejected": -2.977902889251709,
|
1250 |
+
"loss": 1.073,
|
1251 |
+
"rewards/accuracies": 0.893750011920929,
|
1252 |
+
"rewards/chosen": -21.31907081604004,
|
1253 |
+
"rewards/margins": 8.4599609375,
|
1254 |
+
"rewards/rejected": -29.77903175354004,
|
1255 |
+
"step": 410
|
1256 |
+
},
|
1257 |
+
{
|
1258 |
+
"epoch": 0.8871075484301937,
|
1259 |
+
"grad_norm": 121.04879854574509,
|
1260 |
+
"learning_rate": 3.734784976300165e-08,
|
1261 |
+
"logits/chosen": -1.5042963027954102,
|
1262 |
+
"logits/rejected": -1.4524286985397339,
|
1263 |
+
"logps/chosen": -2.0573108196258545,
|
1264 |
+
"logps/rejected": -3.0475449562072754,
|
1265 |
+
"loss": 1.2613,
|
1266 |
+
"rewards/accuracies": 0.875,
|
1267 |
+
"rewards/chosen": -20.573108673095703,
|
1268 |
+
"rewards/margins": 9.902341842651367,
|
1269 |
+
"rewards/rejected": -30.475452423095703,
|
1270 |
+
"step": 415
|
1271 |
+
},
|
1272 |
+
{
|
1273 |
+
"epoch": 0.8977955911823647,
|
1274 |
+
"grad_norm": 139.53176945804844,
|
1275 |
+
"learning_rate": 3.058153372200695e-08,
|
1276 |
+
"logits/chosen": -1.5310368537902832,
|
1277 |
+
"logits/rejected": -1.4835598468780518,
|
1278 |
+
"logps/chosen": -2.005772352218628,
|
1279 |
+
"logps/rejected": -3.0218088626861572,
|
1280 |
+
"loss": 1.1364,
|
1281 |
+
"rewards/accuracies": 0.9125000238418579,
|
1282 |
+
"rewards/chosen": -20.057723999023438,
|
1283 |
+
"rewards/margins": 10.160362243652344,
|
1284 |
+
"rewards/rejected": -30.218088150024414,
|
1285 |
+
"step": 420
|
1286 |
+
},
|
1287 |
+
{
|
1288 |
+
"epoch": 0.9084836339345357,
|
1289 |
+
"grad_norm": 136.53214771188803,
|
1290 |
+
"learning_rate": 2.4471741852423233e-08,
|
1291 |
+
"logits/chosen": -1.525810718536377,
|
1292 |
+
"logits/rejected": -1.51809823513031,
|
1293 |
+
"logps/chosen": -2.186115026473999,
|
1294 |
+
"logps/rejected": -3.016331911087036,
|
1295 |
+
"loss": 1.4659,
|
1296 |
+
"rewards/accuracies": 0.862500011920929,
|
1297 |
+
"rewards/chosen": -21.86115074157715,
|
1298 |
+
"rewards/margins": 8.302164077758789,
|
1299 |
+
"rewards/rejected": -30.163314819335938,
|
1300 |
+
"step": 425
|
1301 |
+
},
|
1302 |
+
{
|
1303 |
+
"epoch": 0.9191716766867067,
|
1304 |
+
"grad_norm": 159.75999561219473,
|
1305 |
+
"learning_rate": 1.9027019250647036e-08,
|
1306 |
+
"logits/chosen": -1.5137279033660889,
|
1307 |
+
"logits/rejected": -1.5020883083343506,
|
1308 |
+
"logps/chosen": -2.2320022583007812,
|
1309 |
+
"logps/rejected": -3.141892671585083,
|
1310 |
+
"loss": 1.1123,
|
1311 |
+
"rewards/accuracies": 0.90625,
|
1312 |
+
"rewards/chosen": -22.320022583007812,
|
1313 |
+
"rewards/margins": 9.098905563354492,
|
1314 |
+
"rewards/rejected": -31.418926239013672,
|
1315 |
+
"step": 430
|
1316 |
+
},
|
1317 |
+
{
|
1318 |
+
"epoch": 0.9298597194388778,
|
1319 |
+
"grad_norm": 103.73846522782931,
|
1320 |
+
"learning_rate": 1.4254980853566246e-08,
|
1321 |
+
"logits/chosen": -1.469313383102417,
|
1322 |
+
"logits/rejected": -1.4278925657272339,
|
1323 |
+
"logps/chosen": -1.988031029701233,
|
1324 |
+
"logps/rejected": -2.880598545074463,
|
1325 |
+
"loss": 1.1778,
|
1326 |
+
"rewards/accuracies": 0.8812500238418579,
|
1327 |
+
"rewards/chosen": -19.880308151245117,
|
1328 |
+
"rewards/margins": 8.925679206848145,
|
1329 |
+
"rewards/rejected": -28.805988311767578,
|
1330 |
+
"step": 435
|
1331 |
+
},
|
1332 |
+
{
|
1333 |
+
"epoch": 0.9405477621910487,
|
1334 |
+
"grad_norm": 103.8822744213963,
|
1335 |
+
"learning_rate": 1.016230078838226e-08,
|
1336 |
+
"logits/chosen": -1.4933596849441528,
|
1337 |
+
"logits/rejected": -1.4389324188232422,
|
1338 |
+
"logps/chosen": -2.1481704711914062,
|
1339 |
+
"logps/rejected": -3.0063669681549072,
|
1340 |
+
"loss": 1.0981,
|
1341 |
+
"rewards/accuracies": 0.90625,
|
1342 |
+
"rewards/chosen": -21.481706619262695,
|
1343 |
+
"rewards/margins": 8.581964492797852,
|
1344 |
+
"rewards/rejected": -30.063669204711914,
|
1345 |
+
"step": 440
|
1346 |
+
},
|
1347 |
+
{
|
1348 |
+
"epoch": 0.9512358049432198,
|
1349 |
+
"grad_norm": 105.17668873683024,
|
1350 |
+
"learning_rate": 6.754703038239329e-09,
|
1351 |
+
"logits/chosen": -1.4460530281066895,
|
1352 |
+
"logits/rejected": -1.431138277053833,
|
1353 |
+
"logps/chosen": -2.159626007080078,
|
1354 |
+
"logps/rejected": -3.2153289318084717,
|
1355 |
+
"loss": 1.1378,
|
1356 |
+
"rewards/accuracies": 0.893750011920929,
|
1357 |
+
"rewards/chosen": -21.59626007080078,
|
1358 |
+
"rewards/margins": 10.557029724121094,
|
1359 |
+
"rewards/rejected": -32.153289794921875,
|
1360 |
+
"step": 445
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 0.9619238476953907,
|
1364 |
+
"grad_norm": 149.56429045938268,
|
1365 |
+
"learning_rate": 4.036953436716895e-09,
|
1366 |
+
"logits/chosen": -1.5528205633163452,
|
1367 |
+
"logits/rejected": -1.5331635475158691,
|
1368 |
+
"logps/chosen": -2.0413358211517334,
|
1369 |
+
"logps/rejected": -2.8829147815704346,
|
1370 |
+
"loss": 1.1398,
|
1371 |
+
"rewards/accuracies": 0.875,
|
1372 |
+
"rewards/chosen": -20.413360595703125,
|
1373 |
+
"rewards/margins": 8.415786743164062,
|
1374 |
+
"rewards/rejected": -28.829147338867188,
|
1375 |
+
"step": 450
|
1376 |
+
},
|
1377 |
+
{
|
1378 |
+
"epoch": 0.9726118904475618,
|
1379 |
+
"grad_norm": 144.02520089734548,
|
1380 |
+
"learning_rate": 2.0128530023804656e-09,
|
1381 |
+
"logits/chosen": -1.5203301906585693,
|
1382 |
+
"logits/rejected": -1.4858264923095703,
|
1383 |
+
"logps/chosen": -2.0674800872802734,
|
1384 |
+
"logps/rejected": -3.0988433361053467,
|
1385 |
+
"loss": 1.0231,
|
1386 |
+
"rewards/accuracies": 0.9312499761581421,
|
1387 |
+
"rewards/chosen": -20.674800872802734,
|
1388 |
+
"rewards/margins": 10.313634872436523,
|
1389 |
+
"rewards/rejected": -30.988433837890625,
|
1390 |
+
"step": 455
|
1391 |
+
},
|
1392 |
+
{
|
1393 |
+
"epoch": 0.9832999331997327,
|
1394 |
+
"grad_norm": 123.56718367330731,
|
1395 |
+
"learning_rate": 6.852326227130833e-10,
|
1396 |
+
"logits/chosen": -1.5208638906478882,
|
1397 |
+
"logits/rejected": -1.5143053531646729,
|
1398 |
+
"logps/chosen": -2.192969799041748,
|
1399 |
+
"logps/rejected": -3.1692872047424316,
|
1400 |
+
"loss": 1.1094,
|
1401 |
+
"rewards/accuracies": 0.862500011920929,
|
1402 |
+
"rewards/chosen": -21.929697036743164,
|
1403 |
+
"rewards/margins": 9.763177871704102,
|
1404 |
+
"rewards/rejected": -31.692874908447266,
|
1405 |
+
"step": 460
|
1406 |
+
},
|
1407 |
+
{
|
1408 |
+
"epoch": 0.9939879759519038,
|
1409 |
+
"grad_norm": 128.9716796420016,
|
1410 |
+
"learning_rate": 5.594909486328348e-11,
|
1411 |
+
"logits/chosen": -1.4912371635437012,
|
1412 |
+
"logits/rejected": -1.496371865272522,
|
1413 |
+
"logps/chosen": -2.177973508834839,
|
1414 |
+
"logps/rejected": -3.1520798206329346,
|
1415 |
+
"loss": 1.2437,
|
1416 |
+
"rewards/accuracies": 0.856249988079071,
|
1417 |
+
"rewards/chosen": -21.779735565185547,
|
1418 |
+
"rewards/margins": 9.741061210632324,
|
1419 |
+
"rewards/rejected": -31.520797729492188,
|
1420 |
+
"step": 465
|
1421 |
+
},
|
1422 |
+
{
|
1423 |
+
"epoch": 0.9982631930527722,
|
1424 |
+
"step": 467,
|
1425 |
+
"total_flos": 0.0,
|
1426 |
+
"train_loss": 2.0977548006004643,
|
1427 |
+
"train_runtime": 12789.999,
|
1428 |
+
"train_samples_per_second": 4.681,
|
1429 |
+
"train_steps_per_second": 0.037
|
1430 |
+
}
|
1431 |
+
],
|
1432 |
+
"logging_steps": 5,
|
1433 |
+
"max_steps": 467,
|
1434 |
+
"num_input_tokens_seen": 0,
|
1435 |
+
"num_train_epochs": 1,
|
1436 |
+
"save_steps": 1000000,
|
1437 |
+
"stateful_callbacks": {
|
1438 |
+
"TrainerControl": {
|
1439 |
+
"args": {
|
1440 |
+
"should_epoch_stop": false,
|
1441 |
+
"should_evaluate": false,
|
1442 |
+
"should_log": false,
|
1443 |
+
"should_save": true,
|
1444 |
+
"should_training_stop": true
|
1445 |
+
},
|
1446 |
+
"attributes": {}
|
1447 |
+
}
|
1448 |
+
},
|
1449 |
+
"total_flos": 0.0,
|
1450 |
+
"train_batch_size": 2,
|
1451 |
+
"trial_name": null,
|
1452 |
+
"trial_params": null
|
1453 |
+
}
|