Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +1467 -0
trainer_state.json
ADDED
@@ -0,0 +1,1467 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.998691442030882,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 477,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.010468463752944255,
|
13 |
+
"grad_norm": 3.7953430363493075,
|
14 |
+
"learning_rate": 5.208333333333333e-08,
|
15 |
+
"logits/chosen": -2.994403839111328,
|
16 |
+
"logits/rejected": -2.932447671890259,
|
17 |
+
"logps/chosen": -1.0262531042099,
|
18 |
+
"logps/rejected": -1.1902999877929688,
|
19 |
+
"loss": -0.0244,
|
20 |
+
"rewards/accuracies": 0.612500011920929,
|
21 |
+
"rewards/chosen": -1.0262531042099,
|
22 |
+
"rewards/margins": 0.16404692828655243,
|
23 |
+
"rewards/rejected": -1.1902999877929688,
|
24 |
+
"step": 5
|
25 |
+
},
|
26 |
+
{
|
27 |
+
"epoch": 0.02093692750588851,
|
28 |
+
"grad_norm": 4.19613852299559,
|
29 |
+
"learning_rate": 1.0416666666666667e-07,
|
30 |
+
"logits/chosen": -3.0355896949768066,
|
31 |
+
"logits/rejected": -2.978191375732422,
|
32 |
+
"logps/chosen": -1.1014703512191772,
|
33 |
+
"logps/rejected": -1.1318297386169434,
|
34 |
+
"loss": -0.0274,
|
35 |
+
"rewards/accuracies": 0.512499988079071,
|
36 |
+
"rewards/chosen": -1.1014703512191772,
|
37 |
+
"rewards/margins": 0.030359352007508278,
|
38 |
+
"rewards/rejected": -1.1318297386169434,
|
39 |
+
"step": 10
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.031405391258832765,
|
43 |
+
"grad_norm": 4.396342901267451,
|
44 |
+
"learning_rate": 1.5624999999999999e-07,
|
45 |
+
"logits/chosen": -3.0005362033843994,
|
46 |
+
"logits/rejected": -2.9955382347106934,
|
47 |
+
"logps/chosen": -1.0661541223526,
|
48 |
+
"logps/rejected": -1.0591233968734741,
|
49 |
+
"loss": -0.0311,
|
50 |
+
"rewards/accuracies": 0.44999998807907104,
|
51 |
+
"rewards/chosen": -1.0661541223526,
|
52 |
+
"rewards/margins": -0.007030715234577656,
|
53 |
+
"rewards/rejected": -1.0591233968734741,
|
54 |
+
"step": 15
|
55 |
+
},
|
56 |
+
{
|
57 |
+
"epoch": 0.04187385501177702,
|
58 |
+
"grad_norm": 4.0816701499978265,
|
59 |
+
"learning_rate": 2.0833333333333333e-07,
|
60 |
+
"logits/chosen": -3.015394687652588,
|
61 |
+
"logits/rejected": -2.977292537689209,
|
62 |
+
"logps/chosen": -1.0042058229446411,
|
63 |
+
"logps/rejected": -1.0408815145492554,
|
64 |
+
"loss": -0.0327,
|
65 |
+
"rewards/accuracies": 0.48750001192092896,
|
66 |
+
"rewards/chosen": -1.0042058229446411,
|
67 |
+
"rewards/margins": 0.03667566925287247,
|
68 |
+
"rewards/rejected": -1.0408815145492554,
|
69 |
+
"step": 20
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.05234231876472128,
|
73 |
+
"grad_norm": 4.229656757392526,
|
74 |
+
"learning_rate": 2.604166666666667e-07,
|
75 |
+
"logits/chosen": -3.0237503051757812,
|
76 |
+
"logits/rejected": -2.973411798477173,
|
77 |
+
"logps/chosen": -0.9534778594970703,
|
78 |
+
"logps/rejected": -1.1941173076629639,
|
79 |
+
"loss": -0.0414,
|
80 |
+
"rewards/accuracies": 0.6000000238418579,
|
81 |
+
"rewards/chosen": -0.9534778594970703,
|
82 |
+
"rewards/margins": 0.24063941836357117,
|
83 |
+
"rewards/rejected": -1.1941173076629639,
|
84 |
+
"step": 25
|
85 |
+
},
|
86 |
+
{
|
87 |
+
"epoch": 0.06281078251766553,
|
88 |
+
"grad_norm": 5.6524920417618585,
|
89 |
+
"learning_rate": 3.1249999999999997e-07,
|
90 |
+
"logits/chosen": -3.0303757190704346,
|
91 |
+
"logits/rejected": -2.9622457027435303,
|
92 |
+
"logps/chosen": -0.875920295715332,
|
93 |
+
"logps/rejected": -0.9086316227912903,
|
94 |
+
"loss": -0.0368,
|
95 |
+
"rewards/accuracies": 0.550000011920929,
|
96 |
+
"rewards/chosen": -0.875920295715332,
|
97 |
+
"rewards/margins": 0.03271136432886124,
|
98 |
+
"rewards/rejected": -0.9086316227912903,
|
99 |
+
"step": 30
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.07327924627060979,
|
103 |
+
"grad_norm": 5.177058208668614,
|
104 |
+
"learning_rate": 3.645833333333333e-07,
|
105 |
+
"logits/chosen": -3.0142247676849365,
|
106 |
+
"logits/rejected": -2.9961647987365723,
|
107 |
+
"logps/chosen": -0.86571204662323,
|
108 |
+
"logps/rejected": -1.1118712425231934,
|
109 |
+
"loss": -0.0475,
|
110 |
+
"rewards/accuracies": 0.612500011920929,
|
111 |
+
"rewards/chosen": -0.86571204662323,
|
112 |
+
"rewards/margins": 0.24615919589996338,
|
113 |
+
"rewards/rejected": -1.1118712425231934,
|
114 |
+
"step": 35
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.08374771002355404,
|
118 |
+
"grad_norm": 4.595347219497465,
|
119 |
+
"learning_rate": 4.1666666666666667e-07,
|
120 |
+
"logits/chosen": -2.9783272743225098,
|
121 |
+
"logits/rejected": -2.948113203048706,
|
122 |
+
"logps/chosen": -0.8789661526679993,
|
123 |
+
"logps/rejected": -0.9445465803146362,
|
124 |
+
"loss": -0.0484,
|
125 |
+
"rewards/accuracies": 0.550000011920929,
|
126 |
+
"rewards/chosen": -0.8789661526679993,
|
127 |
+
"rewards/margins": 0.06558056175708771,
|
128 |
+
"rewards/rejected": -0.9445465803146362,
|
129 |
+
"step": 40
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.0942161737764983,
|
133 |
+
"grad_norm": 4.777810219545724,
|
134 |
+
"learning_rate": 4.6874999999999996e-07,
|
135 |
+
"logits/chosen": -3.0266075134277344,
|
136 |
+
"logits/rejected": -3.020308494567871,
|
137 |
+
"logps/chosen": -0.8326341509819031,
|
138 |
+
"logps/rejected": -0.9679509401321411,
|
139 |
+
"loss": -0.0617,
|
140 |
+
"rewards/accuracies": 0.612500011920929,
|
141 |
+
"rewards/chosen": -0.8326341509819031,
|
142 |
+
"rewards/margins": 0.13531669974327087,
|
143 |
+
"rewards/rejected": -0.9679509401321411,
|
144 |
+
"step": 45
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 0.10468463752944256,
|
148 |
+
"grad_norm": 5.443655913072939,
|
149 |
+
"learning_rate": 4.999731868769026e-07,
|
150 |
+
"logits/chosen": -2.88401198387146,
|
151 |
+
"logits/rejected": -2.9223647117614746,
|
152 |
+
"logps/chosen": -0.9018585085868835,
|
153 |
+
"logps/rejected": -1.0319406986236572,
|
154 |
+
"loss": -0.0525,
|
155 |
+
"rewards/accuracies": 0.550000011920929,
|
156 |
+
"rewards/chosen": -0.9018585085868835,
|
157 |
+
"rewards/margins": 0.13008219003677368,
|
158 |
+
"rewards/rejected": -1.0319406986236572,
|
159 |
+
"step": 50
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.11515310128238682,
|
163 |
+
"grad_norm": 8.004493874473818,
|
164 |
+
"learning_rate": 4.996716052911017e-07,
|
165 |
+
"logits/chosen": -3.0439066886901855,
|
166 |
+
"logits/rejected": -3.038985013961792,
|
167 |
+
"logps/chosen": -0.9202953577041626,
|
168 |
+
"logps/rejected": -1.087965726852417,
|
169 |
+
"loss": -0.0482,
|
170 |
+
"rewards/accuracies": 0.5874999761581421,
|
171 |
+
"rewards/chosen": -0.9202953577041626,
|
172 |
+
"rewards/margins": 0.1676703691482544,
|
173 |
+
"rewards/rejected": -1.087965726852417,
|
174 |
+
"step": 55
|
175 |
+
},
|
176 |
+
{
|
177 |
+
"epoch": 0.12562156503533106,
|
178 |
+
"grad_norm": 8.438457497880151,
|
179 |
+
"learning_rate": 4.990353313429303e-07,
|
180 |
+
"logits/chosen": -2.9756836891174316,
|
181 |
+
"logits/rejected": -2.904886245727539,
|
182 |
+
"logps/chosen": -0.9683998227119446,
|
183 |
+
"logps/rejected": -1.149547815322876,
|
184 |
+
"loss": -0.0681,
|
185 |
+
"rewards/accuracies": 0.6000000238418579,
|
186 |
+
"rewards/chosen": -0.9683998227119446,
|
187 |
+
"rewards/margins": 0.181147962808609,
|
188 |
+
"rewards/rejected": -1.149547815322876,
|
189 |
+
"step": 60
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 0.1360900287882753,
|
193 |
+
"grad_norm": 6.673228848159463,
|
194 |
+
"learning_rate": 4.980652179769217e-07,
|
195 |
+
"logits/chosen": -2.966130495071411,
|
196 |
+
"logits/rejected": -2.9898931980133057,
|
197 |
+
"logps/chosen": -0.9524378776550293,
|
198 |
+
"logps/rejected": -1.1887619495391846,
|
199 |
+
"loss": -0.0605,
|
200 |
+
"rewards/accuracies": 0.574999988079071,
|
201 |
+
"rewards/chosen": -0.9524378776550293,
|
202 |
+
"rewards/margins": 0.23632416129112244,
|
203 |
+
"rewards/rejected": -1.1887619495391846,
|
204 |
+
"step": 65
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 0.14655849254121958,
|
208 |
+
"grad_norm": 7.12442296176354,
|
209 |
+
"learning_rate": 4.967625656594781e-07,
|
210 |
+
"logits/chosen": -3.0336520671844482,
|
211 |
+
"logits/rejected": -2.9966013431549072,
|
212 |
+
"logps/chosen": -1.0305348634719849,
|
213 |
+
"logps/rejected": -1.2037804126739502,
|
214 |
+
"loss": -0.0808,
|
215 |
+
"rewards/accuracies": 0.5874999761581421,
|
216 |
+
"rewards/chosen": -1.0305348634719849,
|
217 |
+
"rewards/margins": 0.1732456386089325,
|
218 |
+
"rewards/rejected": -1.2037804126739502,
|
219 |
+
"step": 70
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.15702695629416383,
|
223 |
+
"grad_norm": 8.805089068512533,
|
224 |
+
"learning_rate": 4.951291206355559e-07,
|
225 |
+
"logits/chosen": -2.9939000606536865,
|
226 |
+
"logits/rejected": -2.958610773086548,
|
227 |
+
"logps/chosen": -1.0724363327026367,
|
228 |
+
"logps/rejected": -1.3205912113189697,
|
229 |
+
"loss": -0.0632,
|
230 |
+
"rewards/accuracies": 0.5874999761581421,
|
231 |
+
"rewards/chosen": -1.0724363327026367,
|
232 |
+
"rewards/margins": 0.2481549233198166,
|
233 |
+
"rewards/rejected": -1.3205912113189697,
|
234 |
+
"step": 75
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 0.16749542004710807,
|
238 |
+
"grad_norm": 6.848905163137727,
|
239 |
+
"learning_rate": 4.93167072587771e-07,
|
240 |
+
"logits/chosen": -2.9966750144958496,
|
241 |
+
"logits/rejected": -2.9649088382720947,
|
242 |
+
"logps/chosen": -1.0316729545593262,
|
243 |
+
"logps/rejected": -1.4772520065307617,
|
244 |
+
"loss": -0.0784,
|
245 |
+
"rewards/accuracies": 0.637499988079071,
|
246 |
+
"rewards/chosen": -1.0316729545593262,
|
247 |
+
"rewards/margins": 0.4455792009830475,
|
248 |
+
"rewards/rejected": -1.4772520065307617,
|
249 |
+
"step": 80
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 0.17796388380005235,
|
253 |
+
"grad_norm": 8.38726120157511,
|
254 |
+
"learning_rate": 4.908790517010636e-07,
|
255 |
+
"logits/chosen": -2.995182514190674,
|
256 |
+
"logits/rejected": -2.9591526985168457,
|
257 |
+
"logps/chosen": -0.920505166053772,
|
258 |
+
"logps/rejected": -1.3156445026397705,
|
259 |
+
"loss": -0.0829,
|
260 |
+
"rewards/accuracies": 0.625,
|
261 |
+
"rewards/chosen": -0.920505166053772,
|
262 |
+
"rewards/margins": 0.39513930678367615,
|
263 |
+
"rewards/rejected": -1.3156445026397705,
|
264 |
+
"step": 85
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 0.1884323475529966,
|
268 |
+
"grad_norm": 5.335037541677766,
|
269 |
+
"learning_rate": 4.882681251368548e-07,
|
270 |
+
"logits/chosen": -2.9661142826080322,
|
271 |
+
"logits/rejected": -2.9613051414489746,
|
272 |
+
"logps/chosen": -1.0484893321990967,
|
273 |
+
"logps/rejected": -1.3956043720245361,
|
274 |
+
"loss": -0.0851,
|
275 |
+
"rewards/accuracies": 0.6000000238418579,
|
276 |
+
"rewards/chosen": -1.0484893321990967,
|
277 |
+
"rewards/margins": 0.34711501002311707,
|
278 |
+
"rewards/rejected": -1.3956043720245361,
|
279 |
+
"step": 90
|
280 |
+
},
|
281 |
+
{
|
282 |
+
"epoch": 0.19890081130594087,
|
283 |
+
"grad_norm": 5.9994342157674705,
|
284 |
+
"learning_rate": 4.853377929214243e-07,
|
285 |
+
"logits/chosen": -3.0459396839141846,
|
286 |
+
"logits/rejected": -3.03053617477417,
|
287 |
+
"logps/chosen": -1.004883050918579,
|
288 |
+
"logps/rejected": -1.2524257898330688,
|
289 |
+
"loss": -0.0726,
|
290 |
+
"rewards/accuracies": 0.5375000238418579,
|
291 |
+
"rewards/chosen": -1.004883050918579,
|
292 |
+
"rewards/margins": 0.24754276871681213,
|
293 |
+
"rewards/rejected": -1.2524257898330688,
|
294 |
+
"step": 95
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 0.2093692750588851,
|
298 |
+
"grad_norm": 7.84898491024868,
|
299 |
+
"learning_rate": 4.820919832540181e-07,
|
300 |
+
"logits/chosen": -3.0662217140197754,
|
301 |
+
"logits/rejected": -3.080054521560669,
|
302 |
+
"logps/chosen": -1.0252565145492554,
|
303 |
+
"logps/rejected": -1.3465287685394287,
|
304 |
+
"loss": -0.0815,
|
305 |
+
"rewards/accuracies": 0.512499988079071,
|
306 |
+
"rewards/chosen": -1.0252565145492554,
|
307 |
+
"rewards/margins": 0.32127222418785095,
|
308 |
+
"rewards/rejected": -1.3465287685394287,
|
309 |
+
"step": 100
|
310 |
+
},
|
311 |
+
{
|
312 |
+
"epoch": 0.21983773881182936,
|
313 |
+
"grad_norm": 6.574476192964948,
|
314 |
+
"learning_rate": 4.785350472409791e-07,
|
315 |
+
"logits/chosen": -3.0290255546569824,
|
316 |
+
"logits/rejected": -3.0409884452819824,
|
317 |
+
"logps/chosen": -1.2733252048492432,
|
318 |
+
"logps/rejected": -1.5657093524932861,
|
319 |
+
"loss": -0.1053,
|
320 |
+
"rewards/accuracies": 0.625,
|
321 |
+
"rewards/chosen": -1.2733252048492432,
|
322 |
+
"rewards/margins": 0.29238420724868774,
|
323 |
+
"rewards/rejected": -1.5657093524932861,
|
324 |
+
"step": 105
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.23030620256477363,
|
328 |
+
"grad_norm": 10.583840544155157,
|
329 |
+
"learning_rate": 4.7467175306295647e-07,
|
330 |
+
"logits/chosen": -2.9230265617370605,
|
331 |
+
"logits/rejected": -2.9670557975769043,
|
332 |
+
"logps/chosen": -1.0758285522460938,
|
333 |
+
"logps/rejected": -1.6084600687026978,
|
334 |
+
"loss": -0.105,
|
335 |
+
"rewards/accuracies": 0.625,
|
336 |
+
"rewards/chosen": -1.0758285522460938,
|
337 |
+
"rewards/margins": 0.5326316356658936,
|
338 |
+
"rewards/rejected": -1.6084600687026978,
|
339 |
+
"step": 110
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 0.24077466631771788,
|
343 |
+
"grad_norm": 8.422749604707546,
|
344 |
+
"learning_rate": 4.70507279583015e-07,
|
345 |
+
"logits/chosen": -3.0686347484588623,
|
346 |
+
"logits/rejected": -3.062587022781372,
|
347 |
+
"logps/chosen": -1.0952173471450806,
|
348 |
+
"logps/rejected": -1.419394612312317,
|
349 |
+
"loss": -0.0899,
|
350 |
+
"rewards/accuracies": 0.6625000238418579,
|
351 |
+
"rewards/chosen": -1.0952173471450806,
|
352 |
+
"rewards/margins": 0.32417723536491394,
|
353 |
+
"rewards/rejected": -1.419394612312317,
|
354 |
+
"step": 115
|
355 |
+
},
|
356 |
+
{
|
357 |
+
"epoch": 0.2512431300706621,
|
358 |
+
"grad_norm": 7.245428385744917,
|
359 |
+
"learning_rate": 4.6604720940421207e-07,
|
360 |
+
"logits/chosen": -2.9976067543029785,
|
361 |
+
"logits/rejected": -2.9936585426330566,
|
362 |
+
"logps/chosen": -1.1029393672943115,
|
363 |
+
"logps/rejected": -1.669960379600525,
|
364 |
+
"loss": -0.0782,
|
365 |
+
"rewards/accuracies": 0.612500011920929,
|
366 |
+
"rewards/chosen": -1.1029393672943115,
|
367 |
+
"rewards/margins": 0.5670207738876343,
|
368 |
+
"rewards/rejected": -1.669960379600525,
|
369 |
+
"step": 120
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 0.26171159382360637,
|
373 |
+
"grad_norm": 9.260649226954074,
|
374 |
+
"learning_rate": 4.612975213859487e-07,
|
375 |
+
"logits/chosen": -2.89277982711792,
|
376 |
+
"logits/rejected": -2.8418478965759277,
|
377 |
+
"logps/chosen": -1.1914057731628418,
|
378 |
+
"logps/rejected": -1.5322601795196533,
|
379 |
+
"loss": -0.0806,
|
380 |
+
"rewards/accuracies": 0.6499999761581421,
|
381 |
+
"rewards/chosen": -1.1914057731628418,
|
382 |
+
"rewards/margins": 0.3408544659614563,
|
383 |
+
"rewards/rejected": -1.5322601795196533,
|
384 |
+
"step": 125
|
385 |
+
},
|
386 |
+
{
|
387 |
+
"epoch": 0.2721800575765506,
|
388 |
+
"grad_norm": 7.897383530777984,
|
389 |
+
"learning_rate": 4.5626458262912735e-07,
|
390 |
+
"logits/chosen": -2.9802322387695312,
|
391 |
+
"logits/rejected": -3.0263705253601074,
|
392 |
+
"logps/chosen": -1.0792829990386963,
|
393 |
+
"logps/rejected": -1.620160698890686,
|
394 |
+
"loss": -0.0969,
|
395 |
+
"rewards/accuracies": 0.7124999761581421,
|
396 |
+
"rewards/chosen": -1.0792829990386963,
|
397 |
+
"rewards/margins": 0.540877640247345,
|
398 |
+
"rewards/rejected": -1.620160698890686,
|
399 |
+
"step": 130
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 0.2826485213294949,
|
403 |
+
"grad_norm": 8.56684631151293,
|
404 |
+
"learning_rate": 4.5095513994085974e-07,
|
405 |
+
"logits/chosen": -2.9500975608825684,
|
406 |
+
"logits/rejected": -2.930224895477295,
|
407 |
+
"logps/chosen": -1.156005859375,
|
408 |
+
"logps/rejected": -1.5593708753585815,
|
409 |
+
"loss": -0.0941,
|
410 |
+
"rewards/accuracies": 0.612500011920929,
|
411 |
+
"rewards/chosen": -1.156005859375,
|
412 |
+
"rewards/margins": 0.40336498618125916,
|
413 |
+
"rewards/rejected": -1.5593708753585815,
|
414 |
+
"step": 135
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 0.29311698508243916,
|
418 |
+
"grad_norm": 11.777411857127742,
|
419 |
+
"learning_rate": 4.453763107901675e-07,
|
420 |
+
"logits/chosen": -2.9826929569244385,
|
421 |
+
"logits/rejected": -2.996265411376953,
|
422 |
+
"logps/chosen": -1.2722768783569336,
|
423 |
+
"logps/rejected": -1.6690384149551392,
|
424 |
+
"loss": -0.0801,
|
425 |
+
"rewards/accuracies": 0.6000000238418579,
|
426 |
+
"rewards/chosen": -1.2722768783569336,
|
427 |
+
"rewards/margins": 0.39676159620285034,
|
428 |
+
"rewards/rejected": -1.6690384149551392,
|
429 |
+
"step": 140
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.3035854488353834,
|
433 |
+
"grad_norm": 13.441666638709599,
|
434 |
+
"learning_rate": 4.395355737667985e-07,
|
435 |
+
"logits/chosen": -2.8897225856781006,
|
436 |
+
"logits/rejected": -2.924546241760254,
|
437 |
+
"logps/chosen": -1.1064995527267456,
|
438 |
+
"logps/rejected": -1.5362865924835205,
|
439 |
+
"loss": -0.1008,
|
440 |
+
"rewards/accuracies": 0.637499988079071,
|
441 |
+
"rewards/chosen": -1.1064995527267456,
|
442 |
+
"rewards/margins": 0.4297870695590973,
|
443 |
+
"rewards/rejected": -1.5362865924835205,
|
444 |
+
"step": 145
|
445 |
+
},
|
446 |
+
{
|
447 |
+
"epoch": 0.31405391258832765,
|
448 |
+
"grad_norm": 13.628880017668104,
|
449 |
+
"learning_rate": 4.3344075855595097e-07,
|
450 |
+
"logits/chosen": -2.99322247505188,
|
451 |
+
"logits/rejected": -2.963665246963501,
|
452 |
+
"logps/chosen": -1.135622501373291,
|
453 |
+
"logps/rejected": -1.865152359008789,
|
454 |
+
"loss": -0.0913,
|
455 |
+
"rewards/accuracies": 0.675000011920929,
|
456 |
+
"rewards/chosen": -1.135622501373291,
|
457 |
+
"rewards/margins": 0.729529857635498,
|
458 |
+
"rewards/rejected": -1.865152359008789,
|
459 |
+
"step": 150
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 0.3245223763412719,
|
463 |
+
"grad_norm": 18.47300522097184,
|
464 |
+
"learning_rate": 4.271000354423425e-07,
|
465 |
+
"logits/chosen": -2.86858868598938,
|
466 |
+
"logits/rejected": -2.850736141204834,
|
467 |
+
"logps/chosen": -1.3002798557281494,
|
468 |
+
"logps/rejected": -1.8037567138671875,
|
469 |
+
"loss": -0.1109,
|
470 |
+
"rewards/accuracies": 0.6875,
|
471 |
+
"rewards/chosen": -1.3002798557281494,
|
472 |
+
"rewards/margins": 0.5034767985343933,
|
473 |
+
"rewards/rejected": -1.8037567138671875,
|
474 |
+
"step": 155
|
475 |
+
},
|
476 |
+
{
|
477 |
+
"epoch": 0.33499084009421615,
|
478 |
+
"grad_norm": 13.245417836958321,
|
479 |
+
"learning_rate": 4.2052190435769554e-07,
|
480 |
+
"logits/chosen": -2.8188085556030273,
|
481 |
+
"logits/rejected": -2.847963333129883,
|
482 |
+
"logps/chosen": -1.3339486122131348,
|
483 |
+
"logps/rejected": -2.279085636138916,
|
484 |
+
"loss": -0.1074,
|
485 |
+
"rewards/accuracies": 0.6499999761581421,
|
486 |
+
"rewards/chosen": -1.3339486122131348,
|
487 |
+
"rewards/margins": 0.9451370239257812,
|
488 |
+
"rewards/rejected": -2.279085636138916,
|
489 |
+
"step": 160
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 0.34545930384716045,
|
493 |
+
"grad_norm": 16.119605348885486,
|
494 |
+
"learning_rate": 4.137151834863213e-07,
|
495 |
+
"logits/chosen": -2.949324369430542,
|
496 |
+
"logits/rejected": -2.931403875350952,
|
497 |
+
"logps/chosen": -1.4816526174545288,
|
498 |
+
"logps/rejected": -1.9713685512542725,
|
499 |
+
"loss": -0.1042,
|
500 |
+
"rewards/accuracies": 0.637499988079071,
|
501 |
+
"rewards/chosen": -1.4816526174545288,
|
502 |
+
"rewards/margins": 0.4897160530090332,
|
503 |
+
"rewards/rejected": -1.9713685512542725,
|
504 |
+
"step": 165
|
505 |
+
},
|
506 |
+
{
|
507 |
+
"epoch": 0.3559277676001047,
|
508 |
+
"grad_norm": 9.715633120289555,
|
509 |
+
"learning_rate": 4.0668899744407567e-07,
|
510 |
+
"logits/chosen": -2.9010634422302246,
|
511 |
+
"logits/rejected": -2.8985071182250977,
|
512 |
+
"logps/chosen": -1.2567403316497803,
|
513 |
+
"logps/rejected": -1.9025369882583618,
|
514 |
+
"loss": -0.1039,
|
515 |
+
"rewards/accuracies": 0.699999988079071,
|
516 |
+
"rewards/chosen": -1.2567403316497803,
|
517 |
+
"rewards/margins": 0.6457964777946472,
|
518 |
+
"rewards/rejected": -1.9025369882583618,
|
519 |
+
"step": 170
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 0.36639623135304894,
|
523 |
+
"grad_norm": 17.275270203000964,
|
524 |
+
"learning_rate": 3.994527650465352e-07,
|
525 |
+
"logits/chosen": -2.82110595703125,
|
526 |
+
"logits/rejected": -2.841233491897583,
|
527 |
+
"logps/chosen": -1.009877324104309,
|
528 |
+
"logps/rejected": -1.7530879974365234,
|
529 |
+
"loss": -0.1124,
|
530 |
+
"rewards/accuracies": 0.699999988079071,
|
531 |
+
"rewards/chosen": -1.009877324104309,
|
532 |
+
"rewards/margins": 0.7432108521461487,
|
533 |
+
"rewards/rejected": -1.7530879974365234,
|
534 |
+
"step": 175
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.3768646951059932,
|
538 |
+
"grad_norm": 14.01202945114004,
|
539 |
+
"learning_rate": 3.920161866827889e-07,
|
540 |
+
"logits/chosen": -2.848515272140503,
|
541 |
+
"logits/rejected": -2.8287739753723145,
|
542 |
+
"logps/chosen": -1.2261879444122314,
|
543 |
+
"logps/rejected": -1.877882719039917,
|
544 |
+
"loss": -0.1098,
|
545 |
+
"rewards/accuracies": 0.6625000238418579,
|
546 |
+
"rewards/chosen": -1.2261879444122314,
|
547 |
+
"rewards/margins": 0.6516948342323303,
|
548 |
+
"rewards/rejected": -1.877882719039917,
|
549 |
+
"step": 180
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 0.38733315885893743,
|
553 |
+
"grad_norm": 12.209017183487612,
|
554 |
+
"learning_rate": 3.8438923131177237e-07,
|
555 |
+
"logits/chosen": -2.842813491821289,
|
556 |
+
"logits/rejected": -2.884312152862549,
|
557 |
+
"logps/chosen": -1.138585090637207,
|
558 |
+
"logps/rejected": -1.9834768772125244,
|
559 |
+
"loss": -0.122,
|
560 |
+
"rewards/accuracies": 0.6875,
|
561 |
+
"rewards/chosen": -1.138585090637207,
|
562 |
+
"rewards/margins": 0.8448915481567383,
|
563 |
+
"rewards/rejected": -1.9834768772125244,
|
564 |
+
"step": 185
|
565 |
+
},
|
566 |
+
{
|
567 |
+
"epoch": 0.39780162261188173,
|
568 |
+
"grad_norm": 20.510663419119325,
|
569 |
+
"learning_rate": 3.765821230985757e-07,
|
570 |
+
"logits/chosen": -2.815495252609253,
|
571 |
+
"logits/rejected": -2.8491616249084473,
|
572 |
+
"logps/chosen": -1.116502285003662,
|
573 |
+
"logps/rejected": -2.221632480621338,
|
574 |
+
"loss": -0.1159,
|
575 |
+
"rewards/accuracies": 0.7250000238418579,
|
576 |
+
"rewards/chosen": -1.116502285003662,
|
577 |
+
"rewards/margins": 1.1051301956176758,
|
578 |
+
"rewards/rejected": -2.221632480621338,
|
579 |
+
"step": 190
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 0.408270086364826,
|
583 |
+
"grad_norm": 10.576124534133774,
|
584 |
+
"learning_rate": 3.6860532770864005e-07,
|
585 |
+
"logits/chosen": -2.8467681407928467,
|
586 |
+
"logits/rejected": -2.814634323120117,
|
587 |
+
"logps/chosen": -1.2795817852020264,
|
588 |
+
"logps/rejected": -2.408722400665283,
|
589 |
+
"loss": -0.098,
|
590 |
+
"rewards/accuracies": 0.675000011920929,
|
591 |
+
"rewards/chosen": -1.2795817852020264,
|
592 |
+
"rewards/margins": 1.1291409730911255,
|
593 |
+
"rewards/rejected": -2.408722400665283,
|
594 |
+
"step": 195
|
595 |
+
},
|
596 |
+
{
|
597 |
+
"epoch": 0.4187385501177702,
|
598 |
+
"grad_norm": 9.7795437319874,
|
599 |
+
"learning_rate": 3.604695382782159e-07,
|
600 |
+
"logits/chosen": -2.747274398803711,
|
601 |
+
"logits/rejected": -2.6726021766662598,
|
602 |
+
"logps/chosen": -1.3564021587371826,
|
603 |
+
"logps/rejected": -1.8328683376312256,
|
604 |
+
"loss": -0.1253,
|
605 |
+
"rewards/accuracies": 0.5625,
|
606 |
+
"rewards/chosen": -1.3564021587371826,
|
607 |
+
"rewards/margins": 0.47646617889404297,
|
608 |
+
"rewards/rejected": -1.8328683376312256,
|
609 |
+
"step": 200
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"epoch": 0.42920701387071447,
|
613 |
+
"grad_norm": 10.854270078977583,
|
614 |
+
"learning_rate": 3.5218566107988867e-07,
|
615 |
+
"logits/chosen": -2.751420021057129,
|
616 |
+
"logits/rejected": -2.7701942920684814,
|
617 |
+
"logps/chosen": -1.4730335474014282,
|
618 |
+
"logps/rejected": -2.165682315826416,
|
619 |
+
"loss": -0.1223,
|
620 |
+
"rewards/accuracies": 0.6000000238418579,
|
621 |
+
"rewards/chosen": -1.4730335474014282,
|
622 |
+
"rewards/margins": 0.6926488280296326,
|
623 |
+
"rewards/rejected": -2.165682315826416,
|
624 |
+
"step": 205
|
625 |
+
},
|
626 |
+
{
|
627 |
+
"epoch": 0.4396754776236587,
|
628 |
+
"grad_norm": 8.55442674152993,
|
629 |
+
"learning_rate": 3.4376480090239047e-07,
|
630 |
+
"logits/chosen": -2.741508960723877,
|
631 |
+
"logits/rejected": -2.699432849884033,
|
632 |
+
"logps/chosen": -1.2727248668670654,
|
633 |
+
"logps/rejected": -1.861159086227417,
|
634 |
+
"loss": -0.1068,
|
635 |
+
"rewards/accuracies": 0.6499999761581421,
|
636 |
+
"rewards/chosen": -1.2727248668670654,
|
637 |
+
"rewards/margins": 0.5884342193603516,
|
638 |
+
"rewards/rejected": -1.861159086227417,
|
639 |
+
"step": 210
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.45014394137660296,
|
643 |
+
"grad_norm": 11.243462810743234,
|
644 |
+
"learning_rate": 3.3521824616429284e-07,
|
645 |
+
"logits/chosen": -2.6302266120910645,
|
646 |
+
"logits/rejected": -2.601926326751709,
|
647 |
+
"logps/chosen": -1.2547543048858643,
|
648 |
+
"logps/rejected": -2.3449490070343018,
|
649 |
+
"loss": -0.1052,
|
650 |
+
"rewards/accuracies": 0.7250000238418579,
|
651 |
+
"rewards/chosen": -1.2547543048858643,
|
652 |
+
"rewards/margins": 1.090194582939148,
|
653 |
+
"rewards/rejected": -2.3449490070343018,
|
654 |
+
"step": 215
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 0.46061240512954726,
|
658 |
+
"grad_norm": 54.27098259781432,
|
659 |
+
"learning_rate": 3.265574537815398e-07,
|
660 |
+
"logits/chosen": -2.580867052078247,
|
661 |
+
"logits/rejected": -2.5696752071380615,
|
662 |
+
"logps/chosen": -1.3456079959869385,
|
663 |
+
"logps/rejected": -2.1853251457214355,
|
664 |
+
"loss": -0.1262,
|
665 |
+
"rewards/accuracies": 0.6000000238418579,
|
666 |
+
"rewards/chosen": -1.3456079959869385,
|
667 |
+
"rewards/margins": 0.8397170901298523,
|
668 |
+
"rewards/rejected": -2.1853251457214355,
|
669 |
+
"step": 220
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 0.4710808688824915,
|
673 |
+
"grad_norm": 48.88998141840299,
|
674 |
+
"learning_rate": 3.1779403380910425e-07,
|
675 |
+
"logits/chosen": -2.6540842056274414,
|
676 |
+
"logits/rejected": -2.501495838165283,
|
677 |
+
"logps/chosen": -1.7648048400878906,
|
678 |
+
"logps/rejected": -2.8733654022216797,
|
679 |
+
"loss": -0.1162,
|
680 |
+
"rewards/accuracies": 0.800000011920929,
|
681 |
+
"rewards/chosen": -1.7648048400878906,
|
682 |
+
"rewards/margins": 1.1085606813430786,
|
683 |
+
"rewards/rejected": -2.8733654022216797,
|
684 |
+
"step": 225
|
685 |
+
},
|
686 |
+
{
|
687 |
+
"epoch": 0.48154933263543576,
|
688 |
+
"grad_norm": 14.602355735002494,
|
689 |
+
"learning_rate": 3.0893973387735683e-07,
|
690 |
+
"logits/chosen": -2.587488889694214,
|
691 |
+
"logits/rejected": -2.5237855911254883,
|
692 |
+
"logps/chosen": -1.4629151821136475,
|
693 |
+
"logps/rejected": -2.598280191421509,
|
694 |
+
"loss": -0.1311,
|
695 |
+
"rewards/accuracies": 0.6499999761581421,
|
696 |
+
"rewards/chosen": -1.4629151821136475,
|
697 |
+
"rewards/margins": 1.1353652477264404,
|
698 |
+
"rewards/rejected": -2.598280191421509,
|
699 |
+
"step": 230
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 0.49201779638838,
|
703 |
+
"grad_norm": 11.479507096481315,
|
704 |
+
"learning_rate": 3.000064234440111e-07,
|
705 |
+
"logits/chosen": -2.4094467163085938,
|
706 |
+
"logits/rejected": -2.3350281715393066,
|
707 |
+
"logps/chosen": -1.1608326435089111,
|
708 |
+
"logps/rejected": -2.2014541625976562,
|
709 |
+
"loss": -0.1266,
|
710 |
+
"rewards/accuracies": 0.637499988079071,
|
711 |
+
"rewards/chosen": -1.1608326435089111,
|
712 |
+
"rewards/margins": 1.040621280670166,
|
713 |
+
"rewards/rejected": -2.2014541625976562,
|
714 |
+
"step": 235
|
715 |
+
},
|
716 |
+
{
|
717 |
+
"epoch": 0.5024862601413242,
|
718 |
+
"grad_norm": 10.608303165039153,
|
719 |
+
"learning_rate": 2.910060778827554e-07,
|
720 |
+
"logits/chosen": -2.469841718673706,
|
721 |
+
"logits/rejected": -2.4375293254852295,
|
722 |
+
"logps/chosen": -1.57926344871521,
|
723 |
+
"logps/rejected": -2.4959769248962402,
|
724 |
+
"loss": -0.1232,
|
725 |
+
"rewards/accuracies": 0.6625000238418579,
|
726 |
+
"rewards/chosen": -1.57926344871521,
|
727 |
+
"rewards/margins": 0.9167135953903198,
|
728 |
+
"rewards/rejected": -2.4959769248962402,
|
729 |
+
"step": 240
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 0.5129547238942685,
|
733 |
+
"grad_norm": 25.353094981844627,
|
734 |
+
"learning_rate": 2.8195076242990116e-07,
|
735 |
+
"logits/chosen": -2.5656850337982178,
|
736 |
+
"logits/rejected": -2.5005340576171875,
|
737 |
+
"logps/chosen": -1.3019143342971802,
|
738 |
+
"logps/rejected": -2.8710713386535645,
|
739 |
+
"loss": -0.1347,
|
740 |
+
"rewards/accuracies": 0.6875,
|
741 |
+
"rewards/chosen": -1.3019143342971802,
|
742 |
+
"rewards/margins": 1.5691570043563843,
|
743 |
+
"rewards/rejected": -2.8710713386535645,
|
744 |
+
"step": 245
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 0.5234231876472127,
|
748 |
+
"grad_norm": 21.49026192198408,
|
749 |
+
"learning_rate": 2.7285261601056697e-07,
|
750 |
+
"logits/chosen": -2.4698336124420166,
|
751 |
+
"logits/rejected": -2.3579485416412354,
|
752 |
+
"logps/chosen": -1.7306108474731445,
|
753 |
+
"logps/rejected": -3.098844051361084,
|
754 |
+
"loss": -0.1102,
|
755 |
+
"rewards/accuracies": 0.699999988079071,
|
756 |
+
"rewards/chosen": -1.7306108474731445,
|
757 |
+
"rewards/margins": 1.368233323097229,
|
758 |
+
"rewards/rejected": -3.098844051361084,
|
759 |
+
"step": 250
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 0.533891651400157,
|
763 |
+
"grad_norm": 21.368862546983063,
|
764 |
+
"learning_rate": 2.6372383496608186e-07,
|
765 |
+
"logits/chosen": -2.613272190093994,
|
766 |
+
"logits/rejected": -2.5582284927368164,
|
767 |
+
"logps/chosen": -1.365275502204895,
|
768 |
+
"logps/rejected": -2.1760246753692627,
|
769 |
+
"loss": -0.1128,
|
770 |
+
"rewards/accuracies": 0.5874999761581421,
|
771 |
+
"rewards/chosen": -1.365275502204895,
|
772 |
+
"rewards/margins": 0.8107490539550781,
|
773 |
+
"rewards/rejected": -2.1760246753692627,
|
774 |
+
"step": 255
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 0.5443601151531012,
|
778 |
+
"grad_norm": 10.208688557762956,
|
779 |
+
"learning_rate": 2.5457665670441937e-07,
|
780 |
+
"logits/chosen": -2.587008237838745,
|
781 |
+
"logits/rejected": -2.517343521118164,
|
782 |
+
"logps/chosen": -1.1947706937789917,
|
783 |
+
"logps/rejected": -2.122412919998169,
|
784 |
+
"loss": -0.1285,
|
785 |
+
"rewards/accuracies": 0.6875,
|
786 |
+
"rewards/chosen": -1.1947706937789917,
|
787 |
+
"rewards/margins": 0.9276420474052429,
|
788 |
+
"rewards/rejected": -2.122412919998169,
|
789 |
+
"step": 260
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 0.5548285789060455,
|
793 |
+
"grad_norm": 17.908677113828368,
|
794 |
+
"learning_rate": 2.454233432955807e-07,
|
795 |
+
"logits/chosen": -2.5146946907043457,
|
796 |
+
"logits/rejected": -2.5216221809387207,
|
797 |
+
"logps/chosen": -1.1950175762176514,
|
798 |
+
"logps/rejected": -2.243485927581787,
|
799 |
+
"loss": -0.1212,
|
800 |
+
"rewards/accuracies": 0.7124999761581421,
|
801 |
+
"rewards/chosen": -1.1950175762176514,
|
802 |
+
"rewards/margins": 1.0484683513641357,
|
803 |
+
"rewards/rejected": -2.243485927581787,
|
804 |
+
"step": 265
|
805 |
+
},
|
806 |
+
{
|
807 |
+
"epoch": 0.5652970426589898,
|
808 |
+
"grad_norm": 10.277763557702748,
|
809 |
+
"learning_rate": 2.3627616503391812e-07,
|
810 |
+
"logits/chosen": -2.4133479595184326,
|
811 |
+
"logits/rejected": -2.4332380294799805,
|
812 |
+
"logps/chosen": -2.0134479999542236,
|
813 |
+
"logps/rejected": -2.5305628776550293,
|
814 |
+
"loss": -0.1209,
|
815 |
+
"rewards/accuracies": 0.637499988079071,
|
816 |
+
"rewards/chosen": -2.0134479999542236,
|
817 |
+
"rewards/margins": 0.5171147584915161,
|
818 |
+
"rewards/rejected": -2.5305628776550293,
|
819 |
+
"step": 270
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 0.575765506411934,
|
823 |
+
"grad_norm": 20.809963361604478,
|
824 |
+
"learning_rate": 2.2714738398943308e-07,
|
825 |
+
"logits/chosen": -2.465207815170288,
|
826 |
+
"logits/rejected": -2.3378491401672363,
|
827 |
+
"logps/chosen": -1.4752328395843506,
|
828 |
+
"logps/rejected": -3.626649856567383,
|
829 |
+
"loss": -0.1508,
|
830 |
+
"rewards/accuracies": 0.75,
|
831 |
+
"rewards/chosen": -1.4752328395843506,
|
832 |
+
"rewards/margins": 2.1514172554016113,
|
833 |
+
"rewards/rejected": -3.626649856567383,
|
834 |
+
"step": 275
|
835 |
+
},
|
836 |
+
{
|
837 |
+
"epoch": 0.5862339701648783,
|
838 |
+
"grad_norm": 14.238466786958266,
|
839 |
+
"learning_rate": 2.1804923757009882e-07,
|
840 |
+
"logits/chosen": -2.537182331085205,
|
841 |
+
"logits/rejected": -2.421999454498291,
|
842 |
+
"logps/chosen": -1.1645643711090088,
|
843 |
+
"logps/rejected": -2.6537137031555176,
|
844 |
+
"loss": -0.127,
|
845 |
+
"rewards/accuracies": 0.6875,
|
846 |
+
"rewards/chosen": -1.1645643711090088,
|
847 |
+
"rewards/margins": 1.4891494512557983,
|
848 |
+
"rewards/rejected": -2.6537137031555176,
|
849 |
+
"step": 280
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.5967024339178225,
|
853 |
+
"grad_norm": 16.046921334576403,
|
854 |
+
"learning_rate": 2.089939221172446e-07,
|
855 |
+
"logits/chosen": -2.5481698513031006,
|
856 |
+
"logits/rejected": -2.5069780349731445,
|
857 |
+
"logps/chosen": -1.4470568895339966,
|
858 |
+
"logps/rejected": -2.9359803199768066,
|
859 |
+
"loss": -0.1148,
|
860 |
+
"rewards/accuracies": 0.737500011920929,
|
861 |
+
"rewards/chosen": -1.4470568895339966,
|
862 |
+
"rewards/margins": 1.4889236688613892,
|
863 |
+
"rewards/rejected": -2.9359803199768066,
|
864 |
+
"step": 285
|
865 |
+
},
|
866 |
+
{
|
867 |
+
"epoch": 0.6071708976707668,
|
868 |
+
"grad_norm": 18.892819573440462,
|
869 |
+
"learning_rate": 1.9999357655598891e-07,
|
870 |
+
"logits/chosen": -2.431614398956299,
|
871 |
+
"logits/rejected": -2.342071056365967,
|
872 |
+
"logps/chosen": -1.7402465343475342,
|
873 |
+
"logps/rejected": -2.2307562828063965,
|
874 |
+
"loss": -0.1091,
|
875 |
+
"rewards/accuracies": 0.625,
|
876 |
+
"rewards/chosen": -1.7402465343475342,
|
877 |
+
"rewards/margins": 0.49050965905189514,
|
878 |
+
"rewards/rejected": -2.2307562828063965,
|
879 |
+
"step": 290
|
880 |
+
},
|
881 |
+
{
|
882 |
+
"epoch": 0.6176393614237111,
|
883 |
+
"grad_norm": 12.912710021152682,
|
884 |
+
"learning_rate": 1.9106026612264315e-07,
|
885 |
+
"logits/chosen": -2.553067684173584,
|
886 |
+
"logits/rejected": -2.392414093017578,
|
887 |
+
"logps/chosen": -1.4181064367294312,
|
888 |
+
"logps/rejected": -2.5548157691955566,
|
889 |
+
"loss": -0.1331,
|
890 |
+
"rewards/accuracies": 0.7250000238418579,
|
891 |
+
"rewards/chosen": -1.4181064367294312,
|
892 |
+
"rewards/margins": 1.136709451675415,
|
893 |
+
"rewards/rejected": -2.5548157691955566,
|
894 |
+
"step": 295
|
895 |
+
},
|
896 |
+
{
|
897 |
+
"epoch": 0.6281078251766553,
|
898 |
+
"grad_norm": 11.978919101167111,
|
899 |
+
"learning_rate": 1.8220596619089573e-07,
|
900 |
+
"logits/chosen": -2.3048555850982666,
|
901 |
+
"logits/rejected": -2.2152676582336426,
|
902 |
+
"logps/chosen": -1.5337917804718018,
|
903 |
+
"logps/rejected": -2.5993523597717285,
|
904 |
+
"loss": -0.1186,
|
905 |
+
"rewards/accuracies": 0.737500011920929,
|
906 |
+
"rewards/chosen": -1.5337917804718018,
|
907 |
+
"rewards/margins": 1.0655608177185059,
|
908 |
+
"rewards/rejected": -2.5993523597717285,
|
909 |
+
"step": 300
|
910 |
+
},
|
911 |
+
{
|
912 |
+
"epoch": 0.6385762889295996,
|
913 |
+
"grad_norm": 27.362413259228795,
|
914 |
+
"learning_rate": 1.7344254621846017e-07,
|
915 |
+
"logits/chosen": -2.394150733947754,
|
916 |
+
"logits/rejected": -2.315016269683838,
|
917 |
+
"logps/chosen": -1.6578819751739502,
|
918 |
+
"logps/rejected": -2.4831995964050293,
|
919 |
+
"loss": -0.1346,
|
920 |
+
"rewards/accuracies": 0.625,
|
921 |
+
"rewards/chosen": -1.6578819751739502,
|
922 |
+
"rewards/margins": 0.8253179788589478,
|
923 |
+
"rewards/rejected": -2.4831995964050293,
|
924 |
+
"step": 305
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 0.6490447526825438,
|
928 |
+
"grad_norm": 12.794098747918724,
|
929 |
+
"learning_rate": 1.647817538357072e-07,
|
930 |
+
"logits/chosen": -2.384284496307373,
|
931 |
+
"logits/rejected": -2.131286382675171,
|
932 |
+
"logps/chosen": -1.2339116334915161,
|
933 |
+
"logps/rejected": -2.9859485626220703,
|
934 |
+
"loss": -0.1415,
|
935 |
+
"rewards/accuracies": 0.7875000238418579,
|
936 |
+
"rewards/chosen": -1.2339116334915161,
|
937 |
+
"rewards/margins": 1.7520368099212646,
|
938 |
+
"rewards/rejected": -2.9859485626220703,
|
939 |
+
"step": 310
|
940 |
+
},
|
941 |
+
{
|
942 |
+
"epoch": 0.6595132164354881,
|
943 |
+
"grad_norm": 17.663997235638817,
|
944 |
+
"learning_rate": 1.562351990976095e-07,
|
945 |
+
"logits/chosen": -2.458538293838501,
|
946 |
+
"logits/rejected": -2.3150718212127686,
|
947 |
+
"logps/chosen": -1.5998607873916626,
|
948 |
+
"logps/rejected": -2.276341199874878,
|
949 |
+
"loss": -0.1284,
|
950 |
+
"rewards/accuracies": 0.574999988079071,
|
951 |
+
"rewards/chosen": -1.5998607873916626,
|
952 |
+
"rewards/margins": 0.6764806509017944,
|
953 |
+
"rewards/rejected": -2.276341199874878,
|
954 |
+
"step": 315
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 0.6699816801884323,
|
958 |
+
"grad_norm": 17.859774700477008,
|
959 |
+
"learning_rate": 1.478143389201113e-07,
|
960 |
+
"logits/chosen": -2.4031972885131836,
|
961 |
+
"logits/rejected": -2.2891368865966797,
|
962 |
+
"logps/chosen": -1.5045208930969238,
|
963 |
+
"logps/rejected": -2.6329269409179688,
|
964 |
+
"loss": -0.136,
|
965 |
+
"rewards/accuracies": 0.7124999761581421,
|
966 |
+
"rewards/chosen": -1.5045208930969238,
|
967 |
+
"rewards/margins": 1.128406286239624,
|
968 |
+
"rewards/rejected": -2.6329269409179688,
|
969 |
+
"step": 320
|
970 |
+
},
|
971 |
+
{
|
972 |
+
"epoch": 0.6804501439413766,
|
973 |
+
"grad_norm": 22.79098412535109,
|
974 |
+
"learning_rate": 1.3953046172178413e-07,
|
975 |
+
"logits/chosen": -2.346923828125,
|
976 |
+
"logits/rejected": -2.2279601097106934,
|
977 |
+
"logps/chosen": -1.3641127347946167,
|
978 |
+
"logps/rejected": -2.919264316558838,
|
979 |
+
"loss": -0.1149,
|
980 |
+
"rewards/accuracies": 0.7749999761581421,
|
981 |
+
"rewards/chosen": -1.3641127347946167,
|
982 |
+
"rewards/margins": 1.5551520586013794,
|
983 |
+
"rewards/rejected": -2.919264316558838,
|
984 |
+
"step": 325
|
985 |
+
},
|
986 |
+
{
|
987 |
+
"epoch": 0.6909186076943209,
|
988 |
+
"grad_norm": 19.92126992190782,
|
989 |
+
"learning_rate": 1.3139467229135998e-07,
|
990 |
+
"logits/chosen": -2.4331743717193604,
|
991 |
+
"logits/rejected": -2.319657802581787,
|
992 |
+
"logps/chosen": -1.1333062648773193,
|
993 |
+
"logps/rejected": -2.6917026042938232,
|
994 |
+
"loss": -0.1293,
|
995 |
+
"rewards/accuracies": 0.7124999761581421,
|
996 |
+
"rewards/chosen": -1.1333062648773193,
|
997 |
+
"rewards/margins": 1.558396577835083,
|
998 |
+
"rewards/rejected": -2.6917026042938232,
|
999 |
+
"step": 330
|
1000 |
+
},
|
1001 |
+
{
|
1002 |
+
"epoch": 0.7013870714472651,
|
1003 |
+
"grad_norm": 66.91823491595629,
|
1004 |
+
"learning_rate": 1.2341787690142435e-07,
|
1005 |
+
"logits/chosen": -2.1722617149353027,
|
1006 |
+
"logits/rejected": -2.239929676055908,
|
1007 |
+
"logps/chosen": -1.8514454364776611,
|
1008 |
+
"logps/rejected": -2.463306427001953,
|
1009 |
+
"loss": -0.1216,
|
1010 |
+
"rewards/accuracies": 0.6499999761581421,
|
1011 |
+
"rewards/chosen": -1.8514454364776611,
|
1012 |
+
"rewards/margins": 0.6118608713150024,
|
1013 |
+
"rewards/rejected": -2.463306427001953,
|
1014 |
+
"step": 335
|
1015 |
+
},
|
1016 |
+
{
|
1017 |
+
"epoch": 0.7118555352002094,
|
1018 |
+
"grad_norm": 31.938913094618993,
|
1019 |
+
"learning_rate": 1.1561076868822755e-07,
|
1020 |
+
"logits/chosen": -2.334263563156128,
|
1021 |
+
"logits/rejected": -2.202540159225464,
|
1022 |
+
"logps/chosen": -1.2181419134140015,
|
1023 |
+
"logps/rejected": -3.0234227180480957,
|
1024 |
+
"loss": -0.1432,
|
1025 |
+
"rewards/accuracies": 0.6875,
|
1026 |
+
"rewards/chosen": -1.2181419134140015,
|
1027 |
+
"rewards/margins": 1.8052809238433838,
|
1028 |
+
"rewards/rejected": -3.0234227180480957,
|
1029 |
+
"step": 340
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"epoch": 0.7223239989531536,
|
1033 |
+
"grad_norm": 16.20254395042608,
|
1034 |
+
"learning_rate": 1.0798381331721107e-07,
|
1035 |
+
"logits/chosen": -2.3701541423797607,
|
1036 |
+
"logits/rejected": -2.2050604820251465,
|
1037 |
+
"logps/chosen": -1.4401299953460693,
|
1038 |
+
"logps/rejected": -2.4977269172668457,
|
1039 |
+
"loss": -0.114,
|
1040 |
+
"rewards/accuracies": 0.675000011920929,
|
1041 |
+
"rewards/chosen": -1.4401299953460693,
|
1042 |
+
"rewards/margins": 1.0575966835021973,
|
1043 |
+
"rewards/rejected": -2.4977269172668457,
|
1044 |
+
"step": 345
|
1045 |
+
},
|
1046 |
+
{
|
1047 |
+
"epoch": 0.7327924627060979,
|
1048 |
+
"grad_norm": 15.428080471435864,
|
1049 |
+
"learning_rate": 1.0054723495346482e-07,
|
1050 |
+
"logits/chosen": -2.2697322368621826,
|
1051 |
+
"logits/rejected": -1.9184463024139404,
|
1052 |
+
"logps/chosen": -1.9560050964355469,
|
1053 |
+
"logps/rejected": -2.635880947113037,
|
1054 |
+
"loss": -0.1343,
|
1055 |
+
"rewards/accuracies": 0.637499988079071,
|
1056 |
+
"rewards/chosen": -1.9560050964355469,
|
1057 |
+
"rewards/margins": 0.6798759698867798,
|
1058 |
+
"rewards/rejected": -2.635880947113037,
|
1059 |
+
"step": 350
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 0.7432609264590422,
|
1063 |
+
"grad_norm": 42.2232530553668,
|
1064 |
+
"learning_rate": 9.331100255592436e-08,
|
1065 |
+
"logits/chosen": -2.2260258197784424,
|
1066 |
+
"logits/rejected": -2.0879032611846924,
|
1067 |
+
"logps/chosen": -1.52364182472229,
|
1068 |
+
"logps/rejected": -3.164949655532837,
|
1069 |
+
"loss": -0.1182,
|
1070 |
+
"rewards/accuracies": 0.737500011920929,
|
1071 |
+
"rewards/chosen": -1.52364182472229,
|
1072 |
+
"rewards/margins": 1.6413074731826782,
|
1073 |
+
"rewards/rejected": -3.164949655532837,
|
1074 |
+
"step": 355
|
1075 |
+
},
|
1076 |
+
{
|
1077 |
+
"epoch": 0.7537293902119864,
|
1078 |
+
"grad_norm": 15.626689472317864,
|
1079 |
+
"learning_rate": 8.628481651367875e-08,
|
1080 |
+
"logits/chosen": -2.132807731628418,
|
1081 |
+
"logits/rejected": -2.0273263454437256,
|
1082 |
+
"logps/chosen": -1.9577980041503906,
|
1083 |
+
"logps/rejected": -2.6277079582214355,
|
1084 |
+
"loss": -0.1234,
|
1085 |
+
"rewards/accuracies": 0.6625000238418579,
|
1086 |
+
"rewards/chosen": -1.9577980041503906,
|
1087 |
+
"rewards/margins": 0.6699098348617554,
|
1088 |
+
"rewards/rejected": -2.6277079582214355,
|
1089 |
+
"step": 360
|
1090 |
+
},
|
1091 |
+
{
|
1092 |
+
"epoch": 0.7641978539649307,
|
1093 |
+
"grad_norm": 41.37098781296867,
|
1094 |
+
"learning_rate": 7.947809564230445e-08,
|
1095 |
+
"logits/chosen": -2.2800426483154297,
|
1096 |
+
"logits/rejected": -2.0097782611846924,
|
1097 |
+
"logps/chosen": -1.793992042541504,
|
1098 |
+
"logps/rejected": -3.1277503967285156,
|
1099 |
+
"loss": -0.139,
|
1100 |
+
"rewards/accuracies": 0.7124999761581421,
|
1101 |
+
"rewards/chosen": -1.793992042541504,
|
1102 |
+
"rewards/margins": 1.3337581157684326,
|
1103 |
+
"rewards/rejected": -3.1277503967285156,
|
1104 |
+
"step": 365
|
1105 |
+
},
|
1106 |
+
{
|
1107 |
+
"epoch": 0.7746663177178749,
|
1108 |
+
"grad_norm": 62.29595045789601,
|
1109 |
+
"learning_rate": 7.289996455765748e-08,
|
1110 |
+
"logits/chosen": -2.3463916778564453,
|
1111 |
+
"logits/rejected": -2.204310655593872,
|
1112 |
+
"logps/chosen": -1.4694790840148926,
|
1113 |
+
"logps/rejected": -2.4598724842071533,
|
1114 |
+
"loss": -0.1446,
|
1115 |
+
"rewards/accuracies": 0.6875,
|
1116 |
+
"rewards/chosen": -1.4694790840148926,
|
1117 |
+
"rewards/margins": 0.990393340587616,
|
1118 |
+
"rewards/rejected": -2.4598724842071533,
|
1119 |
+
"step": 370
|
1120 |
+
},
|
1121 |
+
{
|
1122 |
+
"epoch": 0.7851347814708192,
|
1123 |
+
"grad_norm": 26.402921223340687,
|
1124 |
+
"learning_rate": 6.655924144404906e-08,
|
1125 |
+
"logits/chosen": -2.1613121032714844,
|
1126 |
+
"logits/rejected": -2.173642158508301,
|
1127 |
+
"logps/chosen": -1.5132877826690674,
|
1128 |
+
"logps/rejected": -2.912444591522217,
|
1129 |
+
"loss": -0.1292,
|
1130 |
+
"rewards/accuracies": 0.6000000238418579,
|
1131 |
+
"rewards/chosen": -1.5132877826690674,
|
1132 |
+
"rewards/margins": 1.399156928062439,
|
1133 |
+
"rewards/rejected": -2.912444591522217,
|
1134 |
+
"step": 375
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"epoch": 0.7956032452237635,
|
1138 |
+
"grad_norm": 30.264254092211626,
|
1139 |
+
"learning_rate": 6.046442623320145e-08,
|
1140 |
+
"logits/chosen": -2.3032054901123047,
|
1141 |
+
"logits/rejected": -2.158975839614868,
|
1142 |
+
"logps/chosen": -1.456990361213684,
|
1143 |
+
"logps/rejected": -2.298990488052368,
|
1144 |
+
"loss": -0.1205,
|
1145 |
+
"rewards/accuracies": 0.75,
|
1146 |
+
"rewards/chosen": -1.456990361213684,
|
1147 |
+
"rewards/margins": 0.8420003056526184,
|
1148 |
+
"rewards/rejected": -2.298990488052368,
|
1149 |
+
"step": 380
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 0.8060717089767077,
|
1153 |
+
"grad_norm": 14.190868550898228,
|
1154 |
+
"learning_rate": 5.4623689209832484e-08,
|
1155 |
+
"logits/chosen": -2.239023447036743,
|
1156 |
+
"logits/rejected": -2.178297758102417,
|
1157 |
+
"logps/chosen": -1.2550222873687744,
|
1158 |
+
"logps/rejected": -2.4818851947784424,
|
1159 |
+
"loss": -0.1307,
|
1160 |
+
"rewards/accuracies": 0.699999988079071,
|
1161 |
+
"rewards/chosen": -1.2550222873687744,
|
1162 |
+
"rewards/margins": 1.226862907409668,
|
1163 |
+
"rewards/rejected": -2.4818851947784424,
|
1164 |
+
"step": 385
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 0.816540172729652,
|
1168 |
+
"grad_norm": 23.083356140391576,
|
1169 |
+
"learning_rate": 4.904486005914027e-08,
|
1170 |
+
"logits/chosen": -2.3200623989105225,
|
1171 |
+
"logits/rejected": -2.108736515045166,
|
1172 |
+
"logps/chosen": -1.408341646194458,
|
1173 |
+
"logps/rejected": -2.439392566680908,
|
1174 |
+
"loss": -0.1144,
|
1175 |
+
"rewards/accuracies": 0.6000000238418579,
|
1176 |
+
"rewards/chosen": -1.408341646194458,
|
1177 |
+
"rewards/margins": 1.0310509204864502,
|
1178 |
+
"rewards/rejected": -2.439392566680908,
|
1179 |
+
"step": 390
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 0.8270086364825961,
|
1183 |
+
"grad_norm": 22.17340356091122,
|
1184 |
+
"learning_rate": 4.373541737087263e-08,
|
1185 |
+
"logits/chosen": -2.2692439556121826,
|
1186 |
+
"logits/rejected": -2.039247751235962,
|
1187 |
+
"logps/chosen": -1.2795946598052979,
|
1188 |
+
"logps/rejected": -2.8815693855285645,
|
1189 |
+
"loss": -0.1234,
|
1190 |
+
"rewards/accuracies": 0.6875,
|
1191 |
+
"rewards/chosen": -1.2795946598052979,
|
1192 |
+
"rewards/margins": 1.6019747257232666,
|
1193 |
+
"rewards/rejected": -2.8815693855285645,
|
1194 |
+
"step": 395
|
1195 |
+
},
|
1196 |
+
{
|
1197 |
+
"epoch": 0.8374771002355405,
|
1198 |
+
"grad_norm": 16.463230317307886,
|
1199 |
+
"learning_rate": 3.8702478614051345e-08,
|
1200 |
+
"logits/chosen": -2.365762710571289,
|
1201 |
+
"logits/rejected": -2.099416971206665,
|
1202 |
+
"logps/chosen": -1.3784708976745605,
|
1203 |
+
"logps/rejected": -2.2646727561950684,
|
1204 |
+
"loss": -0.1132,
|
1205 |
+
"rewards/accuracies": 0.6625000238418579,
|
1206 |
+
"rewards/chosen": -1.3784708976745605,
|
1207 |
+
"rewards/margins": 0.8862019777297974,
|
1208 |
+
"rewards/rejected": -2.2646727561950684,
|
1209 |
+
"step": 400
|
1210 |
+
},
|
1211 |
+
{
|
1212 |
+
"epoch": 0.8479455639884846,
|
1213 |
+
"grad_norm": 18.03293576484585,
|
1214 |
+
"learning_rate": 3.3952790595787986e-08,
|
1215 |
+
"logits/chosen": -2.3217966556549072,
|
1216 |
+
"logits/rejected": -1.9924516677856445,
|
1217 |
+
"logps/chosen": -1.5933209657669067,
|
1218 |
+
"logps/rejected": -3.219921112060547,
|
1219 |
+
"loss": -0.1356,
|
1220 |
+
"rewards/accuracies": 0.762499988079071,
|
1221 |
+
"rewards/chosen": -1.5933209657669067,
|
1222 |
+
"rewards/margins": 1.626599907875061,
|
1223 |
+
"rewards/rejected": -3.219921112060547,
|
1224 |
+
"step": 405
|
1225 |
+
},
|
1226 |
+
{
|
1227 |
+
"epoch": 0.8584140277414289,
|
1228 |
+
"grad_norm": 10.746995288687499,
|
1229 |
+
"learning_rate": 2.9492720416985e-08,
|
1230 |
+
"logits/chosen": -2.2354073524475098,
|
1231 |
+
"logits/rejected": -2.0534746646881104,
|
1232 |
+
"logps/chosen": -1.3867294788360596,
|
1233 |
+
"logps/rejected": -2.943979024887085,
|
1234 |
+
"loss": -0.1385,
|
1235 |
+
"rewards/accuracies": 0.699999988079071,
|
1236 |
+
"rewards/chosen": -1.3867294788360596,
|
1237 |
+
"rewards/margins": 1.5572497844696045,
|
1238 |
+
"rewards/rejected": -2.943979024887085,
|
1239 |
+
"step": 410
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"epoch": 0.8688824914943732,
|
1243 |
+
"grad_norm": 33.58919538932842,
|
1244 |
+
"learning_rate": 2.5328246937043525e-08,
|
1245 |
+
"logits/chosen": -2.1925342082977295,
|
1246 |
+
"logits/rejected": -2.1190714836120605,
|
1247 |
+
"logps/chosen": -1.6617501974105835,
|
1248 |
+
"logps/rejected": -2.4904022216796875,
|
1249 |
+
"loss": -0.1213,
|
1250 |
+
"rewards/accuracies": 0.637499988079071,
|
1251 |
+
"rewards/chosen": -1.6617501974105835,
|
1252 |
+
"rewards/margins": 0.828652024269104,
|
1253 |
+
"rewards/rejected": -2.4904022216796875,
|
1254 |
+
"step": 415
|
1255 |
+
},
|
1256 |
+
{
|
1257 |
+
"epoch": 0.8793509552473174,
|
1258 |
+
"grad_norm": 21.053333844149208,
|
1259 |
+
"learning_rate": 2.1464952759020856e-08,
|
1260 |
+
"logits/chosen": -2.1517724990844727,
|
1261 |
+
"logits/rejected": -2.0602805614471436,
|
1262 |
+
"logps/chosen": -1.5017045736312866,
|
1263 |
+
"logps/rejected": -2.838300943374634,
|
1264 |
+
"loss": -0.1388,
|
1265 |
+
"rewards/accuracies": 0.7124999761581421,
|
1266 |
+
"rewards/chosen": -1.5017045736312866,
|
1267 |
+
"rewards/margins": 1.3365962505340576,
|
1268 |
+
"rewards/rejected": -2.838300943374634,
|
1269 |
+
"step": 420
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 0.8898194190002617,
|
1273 |
+
"grad_norm": 22.480002670218727,
|
1274 |
+
"learning_rate": 1.7908016745981856e-08,
|
1275 |
+
"logits/chosen": -2.3761887550354004,
|
1276 |
+
"logits/rejected": -2.0966172218322754,
|
1277 |
+
"logps/chosen": -1.2259721755981445,
|
1278 |
+
"logps/rejected": -2.8700621128082275,
|
1279 |
+
"loss": -0.1468,
|
1280 |
+
"rewards/accuracies": 0.637499988079071,
|
1281 |
+
"rewards/chosen": -1.2259721755981445,
|
1282 |
+
"rewards/margins": 1.644089937210083,
|
1283 |
+
"rewards/rejected": -2.8700621128082275,
|
1284 |
+
"step": 425
|
1285 |
+
},
|
1286 |
+
{
|
1287 |
+
"epoch": 0.9002878827532059,
|
1288 |
+
"grad_norm": 21.22984719288559,
|
1289 |
+
"learning_rate": 1.4662207078575684e-08,
|
1290 |
+
"logits/chosen": -2.2501072883605957,
|
1291 |
+
"logits/rejected": -1.9055511951446533,
|
1292 |
+
"logps/chosen": -1.4240307807922363,
|
1293 |
+
"logps/rejected": -2.8532519340515137,
|
1294 |
+
"loss": -0.1372,
|
1295 |
+
"rewards/accuracies": 0.737500011920929,
|
1296 |
+
"rewards/chosen": -1.4240307807922363,
|
1297 |
+
"rewards/margins": 1.4292211532592773,
|
1298 |
+
"rewards/rejected": -2.8532519340515137,
|
1299 |
+
"step": 430
|
1300 |
+
},
|
1301 |
+
{
|
1302 |
+
"epoch": 0.9107563465061502,
|
1303 |
+
"grad_norm": 40.370663526287096,
|
1304 |
+
"learning_rate": 1.1731874863145142e-08,
|
1305 |
+
"logits/chosen": -2.192682981491089,
|
1306 |
+
"logits/rejected": -2.085170030593872,
|
1307 |
+
"logps/chosen": -1.2839529514312744,
|
1308 |
+
"logps/rejected": -2.737715721130371,
|
1309 |
+
"loss": -0.1369,
|
1310 |
+
"rewards/accuracies": 0.612500011920929,
|
1311 |
+
"rewards/chosen": -1.2839529514312744,
|
1312 |
+
"rewards/margins": 1.4537627696990967,
|
1313 |
+
"rewards/rejected": -2.737715721130371,
|
1314 |
+
"step": 435
|
1315 |
+
},
|
1316 |
+
{
|
1317 |
+
"epoch": 0.9212248102590945,
|
1318 |
+
"grad_norm": 18.579655203147492,
|
1319 |
+
"learning_rate": 9.12094829893642e-09,
|
1320 |
+
"logits/chosen": -2.225111484527588,
|
1321 |
+
"logits/rejected": -2.090142250061035,
|
1322 |
+
"logps/chosen": -1.4594054222106934,
|
1323 |
+
"logps/rejected": -2.5845236778259277,
|
1324 |
+
"loss": -0.1324,
|
1325 |
+
"rewards/accuracies": 0.612500011920929,
|
1326 |
+
"rewards/chosen": -1.4594054222106934,
|
1327 |
+
"rewards/margins": 1.1251181364059448,
|
1328 |
+
"rewards/rejected": -2.5845236778259277,
|
1329 |
+
"step": 440
|
1330 |
+
},
|
1331 |
+
{
|
1332 |
+
"epoch": 0.9316932740120387,
|
1333 |
+
"grad_norm": 36.396751960507515,
|
1334 |
+
"learning_rate": 6.832927412229017e-09,
|
1335 |
+
"logits/chosen": -2.068014144897461,
|
1336 |
+
"logits/rejected": -1.9256622791290283,
|
1337 |
+
"logps/chosen": -1.551655650138855,
|
1338 |
+
"logps/rejected": -2.514082193374634,
|
1339 |
+
"loss": -0.1397,
|
1340 |
+
"rewards/accuracies": 0.637499988079071,
|
1341 |
+
"rewards/chosen": -1.551655650138855,
|
1342 |
+
"rewards/margins": 0.9624261856079102,
|
1343 |
+
"rewards/rejected": -2.514082193374634,
|
1344 |
+
"step": 445
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"epoch": 0.942161737764983,
|
1348 |
+
"grad_norm": 14.063862533994572,
|
1349 |
+
"learning_rate": 4.8708793644441086e-09,
|
1350 |
+
"logits/chosen": -2.364987373352051,
|
1351 |
+
"logits/rejected": -2.075181245803833,
|
1352 |
+
"logps/chosen": -1.2103084325790405,
|
1353 |
+
"logps/rejected": -2.0531344413757324,
|
1354 |
+
"loss": -0.1367,
|
1355 |
+
"rewards/accuracies": 0.6625000238418579,
|
1356 |
+
"rewards/chosen": -1.2103084325790405,
|
1357 |
+
"rewards/margins": 0.8428258895874023,
|
1358 |
+
"rewards/rejected": -2.0531344413757324,
|
1359 |
+
"step": 450
|
1360 |
+
},
|
1361 |
+
{
|
1362 |
+
"epoch": 0.9526302015179272,
|
1363 |
+
"grad_norm": 25.580016907055246,
|
1364 |
+
"learning_rate": 3.2374343405217884e-09,
|
1365 |
+
"logits/chosen": -2.308936357498169,
|
1366 |
+
"logits/rejected": -2.066481590270996,
|
1367 |
+
"logps/chosen": -1.6806007623672485,
|
1368 |
+
"logps/rejected": -2.6968746185302734,
|
1369 |
+
"loss": -0.1109,
|
1370 |
+
"rewards/accuracies": 0.699999988079071,
|
1371 |
+
"rewards/chosen": -1.6806007623672485,
|
1372 |
+
"rewards/margins": 1.016274094581604,
|
1373 |
+
"rewards/rejected": -2.6968746185302734,
|
1374 |
+
"step": 455
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 0.9630986652708715,
|
1378 |
+
"grad_norm": 15.882341955307094,
|
1379 |
+
"learning_rate": 1.9347820230782295e-09,
|
1380 |
+
"logits/chosen": -2.4680800437927246,
|
1381 |
+
"logits/rejected": -1.8986313343048096,
|
1382 |
+
"logps/chosen": -1.2367839813232422,
|
1383 |
+
"logps/rejected": -3.457548141479492,
|
1384 |
+
"loss": -0.1378,
|
1385 |
+
"rewards/accuracies": 0.824999988079071,
|
1386 |
+
"rewards/chosen": -1.2367839813232422,
|
1387 |
+
"rewards/margins": 2.22076416015625,
|
1388 |
+
"rewards/rejected": -3.457548141479492,
|
1389 |
+
"step": 460
|
1390 |
+
},
|
1391 |
+
{
|
1392 |
+
"epoch": 0.9735671290238157,
|
1393 |
+
"grad_norm": 26.022689528234064,
|
1394 |
+
"learning_rate": 9.64668657069706e-10,
|
1395 |
+
"logits/chosen": -2.198855400085449,
|
1396 |
+
"logits/rejected": -1.8760795593261719,
|
1397 |
+
"logps/chosen": -1.3434503078460693,
|
1398 |
+
"logps/rejected": -3.093702793121338,
|
1399 |
+
"loss": -0.1446,
|
1400 |
+
"rewards/accuracies": 0.7749999761581421,
|
1401 |
+
"rewards/chosen": -1.3434503078460693,
|
1402 |
+
"rewards/margins": 1.750252366065979,
|
1403 |
+
"rewards/rejected": -3.093702793121338,
|
1404 |
+
"step": 465
|
1405 |
+
},
|
1406 |
+
{
|
1407 |
+
"epoch": 0.98403559277676,
|
1408 |
+
"grad_norm": 10.255280078989255,
|
1409 |
+
"learning_rate": 3.2839470889836627e-10,
|
1410 |
+
"logits/chosen": -2.194026470184326,
|
1411 |
+
"logits/rejected": -2.0016260147094727,
|
1412 |
+
"logps/chosen": -1.472402811050415,
|
1413 |
+
"logps/rejected": -2.952603340148926,
|
1414 |
+
"loss": -0.1302,
|
1415 |
+
"rewards/accuracies": 0.625,
|
1416 |
+
"rewards/chosen": -1.472402811050415,
|
1417 |
+
"rewards/margins": 1.4802005290985107,
|
1418 |
+
"rewards/rejected": -2.952603340148926,
|
1419 |
+
"step": 470
|
1420 |
+
},
|
1421 |
+
{
|
1422 |
+
"epoch": 0.9945040565297043,
|
1423 |
+
"grad_norm": 22.001525465495153,
|
1424 |
+
"learning_rate": 2.6813123097352287e-11,
|
1425 |
+
"logits/chosen": -2.2689156532287598,
|
1426 |
+
"logits/rejected": -2.0389671325683594,
|
1427 |
+
"logps/chosen": -1.387656807899475,
|
1428 |
+
"logps/rejected": -2.36881947517395,
|
1429 |
+
"loss": -0.1428,
|
1430 |
+
"rewards/accuracies": 0.625,
|
1431 |
+
"rewards/chosen": -1.387656807899475,
|
1432 |
+
"rewards/margins": 0.9811626672744751,
|
1433 |
+
"rewards/rejected": -2.36881947517395,
|
1434 |
+
"step": 475
|
1435 |
+
},
|
1436 |
+
{
|
1437 |
+
"epoch": 0.998691442030882,
|
1438 |
+
"step": 477,
|
1439 |
+
"total_flos": 0.0,
|
1440 |
+
"train_loss": -0.10732274213551975,
|
1441 |
+
"train_runtime": 21833.2592,
|
1442 |
+
"train_samples_per_second": 2.8,
|
1443 |
+
"train_steps_per_second": 0.022
|
1444 |
+
}
|
1445 |
+
],
|
1446 |
+
"logging_steps": 5,
|
1447 |
+
"max_steps": 477,
|
1448 |
+
"num_input_tokens_seen": 0,
|
1449 |
+
"num_train_epochs": 1,
|
1450 |
+
"save_steps": 239,
|
1451 |
+
"stateful_callbacks": {
|
1452 |
+
"TrainerControl": {
|
1453 |
+
"args": {
|
1454 |
+
"should_epoch_stop": false,
|
1455 |
+
"should_evaluate": false,
|
1456 |
+
"should_log": false,
|
1457 |
+
"should_save": true,
|
1458 |
+
"should_training_stop": false
|
1459 |
+
},
|
1460 |
+
"attributes": {}
|
1461 |
+
}
|
1462 |
+
},
|
1463 |
+
"total_flos": 0.0,
|
1464 |
+
"train_batch_size": 1,
|
1465 |
+
"trial_name": null,
|
1466 |
+
"trial_params": null
|
1467 |
+
}
|