Training in progress, step 51000, checkpoint
Browse files
last-checkpoint/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65ec0b3a186d550c4ba55716532995005afbb92f004b5646d6a052d0dddc69f0
|
| 3 |
+
size 14645
|
last-checkpoint/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad0500ea63996002f0ef9cbc01487b7c8e3fdf9b62bcb36aaedc4fd0d5aa6c40
|
| 3 |
+
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -35008,6 +35008,706 @@
|
|
| 35008 |
"learning_rate": 0.000451547516582407,
|
| 35009 |
"loss": 0.5963,
|
| 35010 |
"step": 50000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35011 |
}
|
| 35012 |
],
|
| 35013 |
"logging_steps": 10,
|
|
@@ -35027,7 +35727,7 @@
|
|
| 35027 |
"attributes": {}
|
| 35028 |
}
|
| 35029 |
},
|
| 35030 |
-
"total_flos":
|
| 35031 |
"train_batch_size": 2048,
|
| 35032 |
"trial_name": null,
|
| 35033 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.208896534775129,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 51000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 35008 |
"learning_rate": 0.000451547516582407,
|
| 35009 |
"loss": 0.5963,
|
| 35010 |
"step": 50000
|
| 35011 |
+
},
|
| 35012 |
+
{
|
| 35013 |
+
"epoch": 0.20484148439420005,
|
| 35014 |
+
"grad_norm": 0.243549644947052,
|
| 35015 |
+
"learning_rate": 0.0004515284030288294,
|
| 35016 |
+
"loss": 0.5854,
|
| 35017 |
+
"step": 50010
|
| 35018 |
+
},
|
| 35019 |
+
{
|
| 35020 |
+
"epoch": 0.20488244449905793,
|
| 35021 |
+
"grad_norm": 0.279680460691452,
|
| 35022 |
+
"learning_rate": 0.00045150928611073136,
|
| 35023 |
+
"loss": 0.5868,
|
| 35024 |
+
"step": 50020
|
| 35025 |
+
},
|
| 35026 |
+
{
|
| 35027 |
+
"epoch": 0.20492340460391578,
|
| 35028 |
+
"grad_norm": 0.24950598180294037,
|
| 35029 |
+
"learning_rate": 0.00045149016582843185,
|
| 35030 |
+
"loss": 0.5811,
|
| 35031 |
+
"step": 50030
|
| 35032 |
+
},
|
| 35033 |
+
{
|
| 35034 |
+
"epoch": 0.20496436470877366,
|
| 35035 |
+
"grad_norm": 0.244454026222229,
|
| 35036 |
+
"learning_rate": 0.00045147104218225015,
|
| 35037 |
+
"loss": 0.5708,
|
| 35038 |
+
"step": 50040
|
| 35039 |
+
},
|
| 35040 |
+
{
|
| 35041 |
+
"epoch": 0.2050053248136315,
|
| 35042 |
+
"grad_norm": 0.2745433449745178,
|
| 35043 |
+
"learning_rate": 0.0004514519151725056,
|
| 35044 |
+
"loss": 0.5695,
|
| 35045 |
+
"step": 50050
|
| 35046 |
+
},
|
| 35047 |
+
{
|
| 35048 |
+
"epoch": 0.2050462849184894,
|
| 35049 |
+
"grad_norm": 0.2784786522388458,
|
| 35050 |
+
"learning_rate": 0.00045143278479951746,
|
| 35051 |
+
"loss": 0.5628,
|
| 35052 |
+
"step": 50060
|
| 35053 |
+
},
|
| 35054 |
+
{
|
| 35055 |
+
"epoch": 0.20508724502334727,
|
| 35056 |
+
"grad_norm": 0.25503087043762207,
|
| 35057 |
+
"learning_rate": 0.0004514136510636052,
|
| 35058 |
+
"loss": 0.556,
|
| 35059 |
+
"step": 50070
|
| 35060 |
+
},
|
| 35061 |
+
{
|
| 35062 |
+
"epoch": 0.20512820512820512,
|
| 35063 |
+
"grad_norm": 0.23562008142471313,
|
| 35064 |
+
"learning_rate": 0.00045139451396508814,
|
| 35065 |
+
"loss": 0.5536,
|
| 35066 |
+
"step": 50080
|
| 35067 |
+
},
|
| 35068 |
+
{
|
| 35069 |
+
"epoch": 0.205169165233063,
|
| 35070 |
+
"grad_norm": 0.24655170738697052,
|
| 35071 |
+
"learning_rate": 0.00045137537350428583,
|
| 35072 |
+
"loss": 0.5422,
|
| 35073 |
+
"step": 50090
|
| 35074 |
+
},
|
| 35075 |
+
{
|
| 35076 |
+
"epoch": 0.20521012533792088,
|
| 35077 |
+
"grad_norm": 0.267242431640625,
|
| 35078 |
+
"learning_rate": 0.0004513562296815178,
|
| 35079 |
+
"loss": 0.5341,
|
| 35080 |
+
"step": 50100
|
| 35081 |
+
},
|
| 35082 |
+
{
|
| 35083 |
+
"epoch": 0.20525108544277873,
|
| 35084 |
+
"grad_norm": 0.27842870354652405,
|
| 35085 |
+
"learning_rate": 0.00045133708249710366,
|
| 35086 |
+
"loss": 0.5275,
|
| 35087 |
+
"step": 50110
|
| 35088 |
+
},
|
| 35089 |
+
{
|
| 35090 |
+
"epoch": 0.2052920455476366,
|
| 35091 |
+
"grad_norm": 0.27561154961586,
|
| 35092 |
+
"learning_rate": 0.00045131793195136305,
|
| 35093 |
+
"loss": 0.5152,
|
| 35094 |
+
"step": 50120
|
| 35095 |
+
},
|
| 35096 |
+
{
|
| 35097 |
+
"epoch": 0.20533300565249446,
|
| 35098 |
+
"grad_norm": 0.23686790466308594,
|
| 35099 |
+
"learning_rate": 0.0004512987780446158,
|
| 35100 |
+
"loss": 0.5038,
|
| 35101 |
+
"step": 50130
|
| 35102 |
+
},
|
| 35103 |
+
{
|
| 35104 |
+
"epoch": 0.20537396575735234,
|
| 35105 |
+
"grad_norm": 0.31466424465179443,
|
| 35106 |
+
"learning_rate": 0.00045127962077718157,
|
| 35107 |
+
"loss": 0.4938,
|
| 35108 |
+
"step": 50140
|
| 35109 |
+
},
|
| 35110 |
+
{
|
| 35111 |
+
"epoch": 0.20541492586221022,
|
| 35112 |
+
"grad_norm": 0.24923282861709595,
|
| 35113 |
+
"learning_rate": 0.0004512604601493803,
|
| 35114 |
+
"loss": 0.4856,
|
| 35115 |
+
"step": 50150
|
| 35116 |
+
},
|
| 35117 |
+
{
|
| 35118 |
+
"epoch": 0.20545588596706807,
|
| 35119 |
+
"grad_norm": 0.3219786584377289,
|
| 35120 |
+
"learning_rate": 0.0004512412961615317,
|
| 35121 |
+
"loss": 0.4657,
|
| 35122 |
+
"step": 50160
|
| 35123 |
+
},
|
| 35124 |
+
{
|
| 35125 |
+
"epoch": 0.20549684607192595,
|
| 35126 |
+
"grad_norm": 0.2668122947216034,
|
| 35127 |
+
"learning_rate": 0.00045122212881395585,
|
| 35128 |
+
"loss": 0.4373,
|
| 35129 |
+
"step": 50170
|
| 35130 |
+
},
|
| 35131 |
+
{
|
| 35132 |
+
"epoch": 0.2055378061767838,
|
| 35133 |
+
"grad_norm": 0.23565934598445892,
|
| 35134 |
+
"learning_rate": 0.00045120295810697277,
|
| 35135 |
+
"loss": 0.4352,
|
| 35136 |
+
"step": 50180
|
| 35137 |
+
},
|
| 35138 |
+
{
|
| 35139 |
+
"epoch": 0.20557876628164168,
|
| 35140 |
+
"grad_norm": 0.23589932918548584,
|
| 35141 |
+
"learning_rate": 0.00045118378404090237,
|
| 35142 |
+
"loss": 0.4234,
|
| 35143 |
+
"step": 50190
|
| 35144 |
+
},
|
| 35145 |
+
{
|
| 35146 |
+
"epoch": 0.20561972638649956,
|
| 35147 |
+
"grad_norm": 0.23775924742221832,
|
| 35148 |
+
"learning_rate": 0.00045116460661606493,
|
| 35149 |
+
"loss": 0.4069,
|
| 35150 |
+
"step": 50200
|
| 35151 |
+
},
|
| 35152 |
+
{
|
| 35153 |
+
"epoch": 0.2056606864913574,
|
| 35154 |
+
"grad_norm": 0.23857219517230988,
|
| 35155 |
+
"learning_rate": 0.0004511454258327805,
|
| 35156 |
+
"loss": 0.3881,
|
| 35157 |
+
"step": 50210
|
| 35158 |
+
},
|
| 35159 |
+
{
|
| 35160 |
+
"epoch": 0.2057016465962153,
|
| 35161 |
+
"grad_norm": 0.23606610298156738,
|
| 35162 |
+
"learning_rate": 0.0004511262416913694,
|
| 35163 |
+
"loss": 0.3744,
|
| 35164 |
+
"step": 50220
|
| 35165 |
+
},
|
| 35166 |
+
{
|
| 35167 |
+
"epoch": 0.20574260670107317,
|
| 35168 |
+
"grad_norm": 0.28327473998069763,
|
| 35169 |
+
"learning_rate": 0.0004511070541921518,
|
| 35170 |
+
"loss": 0.3607,
|
| 35171 |
+
"step": 50230
|
| 35172 |
+
},
|
| 35173 |
+
{
|
| 35174 |
+
"epoch": 0.20578356680593102,
|
| 35175 |
+
"grad_norm": 0.2499786615371704,
|
| 35176 |
+
"learning_rate": 0.0004510878633354481,
|
| 35177 |
+
"loss": 0.3425,
|
| 35178 |
+
"step": 50240
|
| 35179 |
+
},
|
| 35180 |
+
{
|
| 35181 |
+
"epoch": 0.2058245269107889,
|
| 35182 |
+
"grad_norm": 0.22934164106845856,
|
| 35183 |
+
"learning_rate": 0.0004510686691215787,
|
| 35184 |
+
"loss": 0.3257,
|
| 35185 |
+
"step": 50250
|
| 35186 |
+
},
|
| 35187 |
+
{
|
| 35188 |
+
"epoch": 0.20586548701564675,
|
| 35189 |
+
"grad_norm": 0.2540028393268585,
|
| 35190 |
+
"learning_rate": 0.00045104947155086403,
|
| 35191 |
+
"loss": 0.3092,
|
| 35192 |
+
"step": 50260
|
| 35193 |
+
},
|
| 35194 |
+
{
|
| 35195 |
+
"epoch": 0.20590644712050463,
|
| 35196 |
+
"grad_norm": 0.23107609152793884,
|
| 35197 |
+
"learning_rate": 0.0004510302706236246,
|
| 35198 |
+
"loss": 0.313,
|
| 35199 |
+
"step": 50270
|
| 35200 |
+
},
|
| 35201 |
+
{
|
| 35202 |
+
"epoch": 0.2059474072253625,
|
| 35203 |
+
"grad_norm": 0.2587755620479584,
|
| 35204 |
+
"learning_rate": 0.000451011066340181,
|
| 35205 |
+
"loss": 0.3126,
|
| 35206 |
+
"step": 50280
|
| 35207 |
+
},
|
| 35208 |
+
{
|
| 35209 |
+
"epoch": 0.20598836733022036,
|
| 35210 |
+
"grad_norm": 0.25043630599975586,
|
| 35211 |
+
"learning_rate": 0.0004509918587008537,
|
| 35212 |
+
"loss": 0.3531,
|
| 35213 |
+
"step": 50290
|
| 35214 |
+
},
|
| 35215 |
+
{
|
| 35216 |
+
"epoch": 0.20602932743507824,
|
| 35217 |
+
"grad_norm": 0.2484988570213318,
|
| 35218 |
+
"learning_rate": 0.0004509726477059635,
|
| 35219 |
+
"loss": 0.3966,
|
| 35220 |
+
"step": 50300
|
| 35221 |
+
},
|
| 35222 |
+
{
|
| 35223 |
+
"epoch": 0.20607028753993611,
|
| 35224 |
+
"grad_norm": 0.2553873062133789,
|
| 35225 |
+
"learning_rate": 0.0004509534333558312,
|
| 35226 |
+
"loss": 0.4431,
|
| 35227 |
+
"step": 50310
|
| 35228 |
+
},
|
| 35229 |
+
{
|
| 35230 |
+
"epoch": 0.20611124764479397,
|
| 35231 |
+
"grad_norm": 0.25729450583457947,
|
| 35232 |
+
"learning_rate": 0.0004509342156507774,
|
| 35233 |
+
"loss": 0.5248,
|
| 35234 |
+
"step": 50320
|
| 35235 |
+
},
|
| 35236 |
+
{
|
| 35237 |
+
"epoch": 0.20615220774965184,
|
| 35238 |
+
"grad_norm": 0.28946325182914734,
|
| 35239 |
+
"learning_rate": 0.00045091499459112307,
|
| 35240 |
+
"loss": 0.5807,
|
| 35241 |
+
"step": 50330
|
| 35242 |
+
},
|
| 35243 |
+
{
|
| 35244 |
+
"epoch": 0.2061931678545097,
|
| 35245 |
+
"grad_norm": 0.2637884020805359,
|
| 35246 |
+
"learning_rate": 0.0004508957701771891,
|
| 35247 |
+
"loss": 0.6062,
|
| 35248 |
+
"step": 50340
|
| 35249 |
+
},
|
| 35250 |
+
{
|
| 35251 |
+
"epoch": 0.20623412795936757,
|
| 35252 |
+
"grad_norm": 0.30340537428855896,
|
| 35253 |
+
"learning_rate": 0.00045087654240929636,
|
| 35254 |
+
"loss": 0.632,
|
| 35255 |
+
"step": 50350
|
| 35256 |
+
},
|
| 35257 |
+
{
|
| 35258 |
+
"epoch": 0.20627508806422545,
|
| 35259 |
+
"grad_norm": 0.2894531488418579,
|
| 35260 |
+
"learning_rate": 0.00045085731128776595,
|
| 35261 |
+
"loss": 0.657,
|
| 35262 |
+
"step": 50360
|
| 35263 |
+
},
|
| 35264 |
+
{
|
| 35265 |
+
"epoch": 0.2063160481690833,
|
| 35266 |
+
"grad_norm": 0.27278509736061096,
|
| 35267 |
+
"learning_rate": 0.0004508380768129189,
|
| 35268 |
+
"loss": 0.6787,
|
| 35269 |
+
"step": 50370
|
| 35270 |
+
},
|
| 35271 |
+
{
|
| 35272 |
+
"epoch": 0.20635700827394118,
|
| 35273 |
+
"grad_norm": 0.31178078055381775,
|
| 35274 |
+
"learning_rate": 0.00045081883898507626,
|
| 35275 |
+
"loss": 0.6832,
|
| 35276 |
+
"step": 50380
|
| 35277 |
+
},
|
| 35278 |
+
{
|
| 35279 |
+
"epoch": 0.20639796837879906,
|
| 35280 |
+
"grad_norm": 0.33291715383529663,
|
| 35281 |
+
"learning_rate": 0.0004507995978045593,
|
| 35282 |
+
"loss": 0.6923,
|
| 35283 |
+
"step": 50390
|
| 35284 |
+
},
|
| 35285 |
+
{
|
| 35286 |
+
"epoch": 0.2064389284836569,
|
| 35287 |
+
"grad_norm": 0.26210567355155945,
|
| 35288 |
+
"learning_rate": 0.00045078035327168926,
|
| 35289 |
+
"loss": 0.7106,
|
| 35290 |
+
"step": 50400
|
| 35291 |
+
},
|
| 35292 |
+
{
|
| 35293 |
+
"epoch": 0.2064798885885148,
|
| 35294 |
+
"grad_norm": 0.27464473247528076,
|
| 35295 |
+
"learning_rate": 0.0004507611053867874,
|
| 35296 |
+
"loss": 0.7063,
|
| 35297 |
+
"step": 50410
|
| 35298 |
+
},
|
| 35299 |
+
{
|
| 35300 |
+
"epoch": 0.20652084869337264,
|
| 35301 |
+
"grad_norm": 0.25764334201812744,
|
| 35302 |
+
"learning_rate": 0.000450741854150175,
|
| 35303 |
+
"loss": 0.7188,
|
| 35304 |
+
"step": 50420
|
| 35305 |
+
},
|
| 35306 |
+
{
|
| 35307 |
+
"epoch": 0.20656180879823052,
|
| 35308 |
+
"grad_norm": 0.2693474590778351,
|
| 35309 |
+
"learning_rate": 0.0004507225995621735,
|
| 35310 |
+
"loss": 0.716,
|
| 35311 |
+
"step": 50430
|
| 35312 |
+
},
|
| 35313 |
+
{
|
| 35314 |
+
"epoch": 0.2066027689030884,
|
| 35315 |
+
"grad_norm": 0.2756979167461395,
|
| 35316 |
+
"learning_rate": 0.00045070334162310437,
|
| 35317 |
+
"loss": 0.7196,
|
| 35318 |
+
"step": 50440
|
| 35319 |
+
},
|
| 35320 |
+
{
|
| 35321 |
+
"epoch": 0.20664372900794625,
|
| 35322 |
+
"grad_norm": 0.31787070631980896,
|
| 35323 |
+
"learning_rate": 0.0004506840803332891,
|
| 35324 |
+
"loss": 0.72,
|
| 35325 |
+
"step": 50450
|
| 35326 |
+
},
|
| 35327 |
+
{
|
| 35328 |
+
"epoch": 0.20668468911280413,
|
| 35329 |
+
"grad_norm": 0.29845964908599854,
|
| 35330 |
+
"learning_rate": 0.0004506648156930494,
|
| 35331 |
+
"loss": 0.7171,
|
| 35332 |
+
"step": 50460
|
| 35333 |
+
},
|
| 35334 |
+
{
|
| 35335 |
+
"epoch": 0.206725649217662,
|
| 35336 |
+
"grad_norm": 0.2559990882873535,
|
| 35337 |
+
"learning_rate": 0.00045064554770270665,
|
| 35338 |
+
"loss": 0.7222,
|
| 35339 |
+
"step": 50470
|
| 35340 |
+
},
|
| 35341 |
+
{
|
| 35342 |
+
"epoch": 0.20676660932251986,
|
| 35343 |
+
"grad_norm": 0.30153951048851013,
|
| 35344 |
+
"learning_rate": 0.0004506262763625827,
|
| 35345 |
+
"loss": 0.7165,
|
| 35346 |
+
"step": 50480
|
| 35347 |
+
},
|
| 35348 |
+
{
|
| 35349 |
+
"epoch": 0.20680756942737774,
|
| 35350 |
+
"grad_norm": 0.3911590874195099,
|
| 35351 |
+
"learning_rate": 0.00045060700167299925,
|
| 35352 |
+
"loss": 0.7276,
|
| 35353 |
+
"step": 50490
|
| 35354 |
+
},
|
| 35355 |
+
{
|
| 35356 |
+
"epoch": 0.2068485295322356,
|
| 35357 |
+
"grad_norm": 0.27202489972114563,
|
| 35358 |
+
"learning_rate": 0.000450587723634278,
|
| 35359 |
+
"loss": 0.7302,
|
| 35360 |
+
"step": 50500
|
| 35361 |
+
},
|
| 35362 |
+
{
|
| 35363 |
+
"epoch": 0.20688948963709347,
|
| 35364 |
+
"grad_norm": 0.2691025137901306,
|
| 35365 |
+
"learning_rate": 0.000450568442246741,
|
| 35366 |
+
"loss": 0.7293,
|
| 35367 |
+
"step": 50510
|
| 35368 |
+
},
|
| 35369 |
+
{
|
| 35370 |
+
"epoch": 0.20693044974195135,
|
| 35371 |
+
"grad_norm": 0.28840580582618713,
|
| 35372 |
+
"learning_rate": 0.0004505491575107099,
|
| 35373 |
+
"loss": 0.7266,
|
| 35374 |
+
"step": 50520
|
| 35375 |
+
},
|
| 35376 |
+
{
|
| 35377 |
+
"epoch": 0.2069714098468092,
|
| 35378 |
+
"grad_norm": 0.30399638414382935,
|
| 35379 |
+
"learning_rate": 0.0004505298694265069,
|
| 35380 |
+
"loss": 0.7212,
|
| 35381 |
+
"step": 50530
|
| 35382 |
+
},
|
| 35383 |
+
{
|
| 35384 |
+
"epoch": 0.20701236995166708,
|
| 35385 |
+
"grad_norm": 0.26806214451789856,
|
| 35386 |
+
"learning_rate": 0.00045051057799445383,
|
| 35387 |
+
"loss": 0.7145,
|
| 35388 |
+
"step": 50540
|
| 35389 |
+
},
|
| 35390 |
+
{
|
| 35391 |
+
"epoch": 0.20705333005652493,
|
| 35392 |
+
"grad_norm": 0.26500576734542847,
|
| 35393 |
+
"learning_rate": 0.0004504912832148728,
|
| 35394 |
+
"loss": 0.7304,
|
| 35395 |
+
"step": 50550
|
| 35396 |
+
},
|
| 35397 |
+
{
|
| 35398 |
+
"epoch": 0.2070942901613828,
|
| 35399 |
+
"grad_norm": 0.25803065299987793,
|
| 35400 |
+
"learning_rate": 0.000450471985088086,
|
| 35401 |
+
"loss": 0.7173,
|
| 35402 |
+
"step": 50560
|
| 35403 |
+
},
|
| 35404 |
+
{
|
| 35405 |
+
"epoch": 0.2071352502662407,
|
| 35406 |
+
"grad_norm": 0.2964495122432709,
|
| 35407 |
+
"learning_rate": 0.0004504526836144156,
|
| 35408 |
+
"loss": 0.7268,
|
| 35409 |
+
"step": 50570
|
| 35410 |
+
},
|
| 35411 |
+
{
|
| 35412 |
+
"epoch": 0.20717621037109854,
|
| 35413 |
+
"grad_norm": 0.2871108651161194,
|
| 35414 |
+
"learning_rate": 0.0004504333787941838,
|
| 35415 |
+
"loss": 0.7239,
|
| 35416 |
+
"step": 50580
|
| 35417 |
+
},
|
| 35418 |
+
{
|
| 35419 |
+
"epoch": 0.20721717047595642,
|
| 35420 |
+
"grad_norm": 0.28984373807907104,
|
| 35421 |
+
"learning_rate": 0.00045041407062771293,
|
| 35422 |
+
"loss": 0.7236,
|
| 35423 |
+
"step": 50590
|
| 35424 |
+
},
|
| 35425 |
+
{
|
| 35426 |
+
"epoch": 0.2072581305808143,
|
| 35427 |
+
"grad_norm": 0.27330225706100464,
|
| 35428 |
+
"learning_rate": 0.00045039475911532535,
|
| 35429 |
+
"loss": 0.7285,
|
| 35430 |
+
"step": 50600
|
| 35431 |
+
},
|
| 35432 |
+
{
|
| 35433 |
+
"epoch": 0.20729909068567215,
|
| 35434 |
+
"grad_norm": 0.27506980299949646,
|
| 35435 |
+
"learning_rate": 0.00045037544425734346,
|
| 35436 |
+
"loss": 0.7182,
|
| 35437 |
+
"step": 50610
|
| 35438 |
+
},
|
| 35439 |
+
{
|
| 35440 |
+
"epoch": 0.20734005079053003,
|
| 35441 |
+
"grad_norm": 0.29893994331359863,
|
| 35442 |
+
"learning_rate": 0.0004503561260540896,
|
| 35443 |
+
"loss": 0.7277,
|
| 35444 |
+
"step": 50620
|
| 35445 |
+
},
|
| 35446 |
+
{
|
| 35447 |
+
"epoch": 0.20738101089538788,
|
| 35448 |
+
"grad_norm": 0.27006301283836365,
|
| 35449 |
+
"learning_rate": 0.0004503368045058864,
|
| 35450 |
+
"loss": 0.7292,
|
| 35451 |
+
"step": 50630
|
| 35452 |
+
},
|
| 35453 |
+
{
|
| 35454 |
+
"epoch": 0.20742197100024576,
|
| 35455 |
+
"grad_norm": 0.3087039887905121,
|
| 35456 |
+
"learning_rate": 0.0004503174796130565,
|
| 35457 |
+
"loss": 0.719,
|
| 35458 |
+
"step": 50640
|
| 35459 |
+
},
|
| 35460 |
+
{
|
| 35461 |
+
"epoch": 0.20746293110510364,
|
| 35462 |
+
"grad_norm": 0.2981952130794525,
|
| 35463 |
+
"learning_rate": 0.00045029815137592245,
|
| 35464 |
+
"loss": 0.719,
|
| 35465 |
+
"step": 50650
|
| 35466 |
+
},
|
| 35467 |
+
{
|
| 35468 |
+
"epoch": 0.2075038912099615,
|
| 35469 |
+
"grad_norm": 0.26963135600090027,
|
| 35470 |
+
"learning_rate": 0.0004502788197948069,
|
| 35471 |
+
"loss": 0.7174,
|
| 35472 |
+
"step": 50660
|
| 35473 |
+
},
|
| 35474 |
+
{
|
| 35475 |
+
"epoch": 0.20754485131481937,
|
| 35476 |
+
"grad_norm": 0.28951266407966614,
|
| 35477 |
+
"learning_rate": 0.00045025948487003263,
|
| 35478 |
+
"loss": 0.7267,
|
| 35479 |
+
"step": 50670
|
| 35480 |
+
},
|
| 35481 |
+
{
|
| 35482 |
+
"epoch": 0.20758581141967725,
|
| 35483 |
+
"grad_norm": 0.25498178601264954,
|
| 35484 |
+
"learning_rate": 0.00045024014660192247,
|
| 35485 |
+
"loss": 0.718,
|
| 35486 |
+
"step": 50680
|
| 35487 |
+
},
|
| 35488 |
+
{
|
| 35489 |
+
"epoch": 0.2076267715245351,
|
| 35490 |
+
"grad_norm": 0.28607216477394104,
|
| 35491 |
+
"learning_rate": 0.0004502208049907992,
|
| 35492 |
+
"loss": 0.7251,
|
| 35493 |
+
"step": 50690
|
| 35494 |
+
},
|
| 35495 |
+
{
|
| 35496 |
+
"epoch": 0.20766773162939298,
|
| 35497 |
+
"grad_norm": 0.2632274925708771,
|
| 35498 |
+
"learning_rate": 0.00045020146003698584,
|
| 35499 |
+
"loss": 0.7205,
|
| 35500 |
+
"step": 50700
|
| 35501 |
+
},
|
| 35502 |
+
{
|
| 35503 |
+
"epoch": 0.20770869173425083,
|
| 35504 |
+
"grad_norm": 0.2581040561199188,
|
| 35505 |
+
"learning_rate": 0.0004501821117408051,
|
| 35506 |
+
"loss": 0.7251,
|
| 35507 |
+
"step": 50710
|
| 35508 |
+
},
|
| 35509 |
+
{
|
| 35510 |
+
"epoch": 0.2077496518391087,
|
| 35511 |
+
"grad_norm": 0.2586017847061157,
|
| 35512 |
+
"learning_rate": 0.00045016276010258037,
|
| 35513 |
+
"loss": 0.7166,
|
| 35514 |
+
"step": 50720
|
| 35515 |
+
},
|
| 35516 |
+
{
|
| 35517 |
+
"epoch": 0.20779061194396659,
|
| 35518 |
+
"grad_norm": 0.2921902537345886,
|
| 35519 |
+
"learning_rate": 0.0004501434051226345,
|
| 35520 |
+
"loss": 0.7214,
|
| 35521 |
+
"step": 50730
|
| 35522 |
+
},
|
| 35523 |
+
{
|
| 35524 |
+
"epoch": 0.20783157204882444,
|
| 35525 |
+
"grad_norm": 0.319943368434906,
|
| 35526 |
+
"learning_rate": 0.0004501240468012906,
|
| 35527 |
+
"loss": 0.7188,
|
| 35528 |
+
"step": 50740
|
| 35529 |
+
},
|
| 35530 |
+
{
|
| 35531 |
+
"epoch": 0.20787253215368232,
|
| 35532 |
+
"grad_norm": 0.2887111306190491,
|
| 35533 |
+
"learning_rate": 0.00045010468513887196,
|
| 35534 |
+
"loss": 0.7231,
|
| 35535 |
+
"step": 50750
|
| 35536 |
+
},
|
| 35537 |
+
{
|
| 35538 |
+
"epoch": 0.2079134922585402,
|
| 35539 |
+
"grad_norm": 0.3027583062648773,
|
| 35540 |
+
"learning_rate": 0.0004500853201357018,
|
| 35541 |
+
"loss": 0.7146,
|
| 35542 |
+
"step": 50760
|
| 35543 |
+
},
|
| 35544 |
+
{
|
| 35545 |
+
"epoch": 0.20795445236339805,
|
| 35546 |
+
"grad_norm": 0.2659209370613098,
|
| 35547 |
+
"learning_rate": 0.0004500659517921034,
|
| 35548 |
+
"loss": 0.7134,
|
| 35549 |
+
"step": 50770
|
| 35550 |
+
},
|
| 35551 |
+
{
|
| 35552 |
+
"epoch": 0.20799541246825592,
|
| 35553 |
+
"grad_norm": 0.29458537697792053,
|
| 35554 |
+
"learning_rate": 0.00045004658010840007,
|
| 35555 |
+
"loss": 0.7209,
|
| 35556 |
+
"step": 50780
|
| 35557 |
+
},
|
| 35558 |
+
{
|
| 35559 |
+
"epoch": 0.20803637257311378,
|
| 35560 |
+
"grad_norm": 0.29622724652290344,
|
| 35561 |
+
"learning_rate": 0.00045002720508491523,
|
| 35562 |
+
"loss": 0.7108,
|
| 35563 |
+
"step": 50790
|
| 35564 |
+
},
|
| 35565 |
+
{
|
| 35566 |
+
"epoch": 0.20807733267797165,
|
| 35567 |
+
"grad_norm": 0.283232718706131,
|
| 35568 |
+
"learning_rate": 0.0004500078267219725,
|
| 35569 |
+
"loss": 0.7095,
|
| 35570 |
+
"step": 50800
|
| 35571 |
+
},
|
| 35572 |
+
{
|
| 35573 |
+
"epoch": 0.20811829278282953,
|
| 35574 |
+
"grad_norm": 0.25766533613204956,
|
| 35575 |
+
"learning_rate": 0.00044998844501989524,
|
| 35576 |
+
"loss": 0.7155,
|
| 35577 |
+
"step": 50810
|
| 35578 |
+
},
|
| 35579 |
+
{
|
| 35580 |
+
"epoch": 0.20815925288768738,
|
| 35581 |
+
"grad_norm": 0.2799261212348938,
|
| 35582 |
+
"learning_rate": 0.00044996905997900705,
|
| 35583 |
+
"loss": 0.7137,
|
| 35584 |
+
"step": 50820
|
| 35585 |
+
},
|
| 35586 |
+
{
|
| 35587 |
+
"epoch": 0.20820021299254526,
|
| 35588 |
+
"grad_norm": 0.2579410970211029,
|
| 35589 |
+
"learning_rate": 0.00044994967159963165,
|
| 35590 |
+
"loss": 0.7168,
|
| 35591 |
+
"step": 50830
|
| 35592 |
+
},
|
| 35593 |
+
{
|
| 35594 |
+
"epoch": 0.20824117309740314,
|
| 35595 |
+
"grad_norm": 0.282565176486969,
|
| 35596 |
+
"learning_rate": 0.00044993027988209257,
|
| 35597 |
+
"loss": 0.7119,
|
| 35598 |
+
"step": 50840
|
| 35599 |
+
},
|
| 35600 |
+
{
|
| 35601 |
+
"epoch": 0.208282133202261,
|
| 35602 |
+
"grad_norm": 0.27341437339782715,
|
| 35603 |
+
"learning_rate": 0.0004499108848267136,
|
| 35604 |
+
"loss": 0.7118,
|
| 35605 |
+
"step": 50850
|
| 35606 |
+
},
|
| 35607 |
+
{
|
| 35608 |
+
"epoch": 0.20832309330711887,
|
| 35609 |
+
"grad_norm": 0.25960370898246765,
|
| 35610 |
+
"learning_rate": 0.00044989148643381874,
|
| 35611 |
+
"loss": 0.7121,
|
| 35612 |
+
"step": 50860
|
| 35613 |
+
},
|
| 35614 |
+
{
|
| 35615 |
+
"epoch": 0.20836405341197672,
|
| 35616 |
+
"grad_norm": 0.27447187900543213,
|
| 35617 |
+
"learning_rate": 0.00044987208470373164,
|
| 35618 |
+
"loss": 0.7108,
|
| 35619 |
+
"step": 50870
|
| 35620 |
+
},
|
| 35621 |
+
{
|
| 35622 |
+
"epoch": 0.2084050135168346,
|
| 35623 |
+
"grad_norm": 0.2560919225215912,
|
| 35624 |
+
"learning_rate": 0.0004498526796367762,
|
| 35625 |
+
"loss": 0.7071,
|
| 35626 |
+
"step": 50880
|
| 35627 |
+
},
|
| 35628 |
+
{
|
| 35629 |
+
"epoch": 0.20844597362169248,
|
| 35630 |
+
"grad_norm": 0.2763515114784241,
|
| 35631 |
+
"learning_rate": 0.0004498332712332766,
|
| 35632 |
+
"loss": 0.7146,
|
| 35633 |
+
"step": 50890
|
| 35634 |
+
},
|
| 35635 |
+
{
|
| 35636 |
+
"epoch": 0.20848693372655033,
|
| 35637 |
+
"grad_norm": 0.2826942503452301,
|
| 35638 |
+
"learning_rate": 0.0004498138594935566,
|
| 35639 |
+
"loss": 0.7025,
|
| 35640 |
+
"step": 50900
|
| 35641 |
+
},
|
| 35642 |
+
{
|
| 35643 |
+
"epoch": 0.2085278938314082,
|
| 35644 |
+
"grad_norm": 0.277884304523468,
|
| 35645 |
+
"learning_rate": 0.00044979444441794043,
|
| 35646 |
+
"loss": 0.7207,
|
| 35647 |
+
"step": 50910
|
| 35648 |
+
},
|
| 35649 |
+
{
|
| 35650 |
+
"epoch": 0.20856885393626606,
|
| 35651 |
+
"grad_norm": 0.3166668713092804,
|
| 35652 |
+
"learning_rate": 0.0004497750260067522,
|
| 35653 |
+
"loss": 0.709,
|
| 35654 |
+
"step": 50920
|
| 35655 |
+
},
|
| 35656 |
+
{
|
| 35657 |
+
"epoch": 0.20860981404112394,
|
| 35658 |
+
"grad_norm": 0.2710054814815521,
|
| 35659 |
+
"learning_rate": 0.00044975560426031613,
|
| 35660 |
+
"loss": 0.7081,
|
| 35661 |
+
"step": 50930
|
| 35662 |
+
},
|
| 35663 |
+
{
|
| 35664 |
+
"epoch": 0.20865077414598182,
|
| 35665 |
+
"grad_norm": 0.2779574990272522,
|
| 35666 |
+
"learning_rate": 0.00044973617917895645,
|
| 35667 |
+
"loss": 0.7115,
|
| 35668 |
+
"step": 50940
|
| 35669 |
+
},
|
| 35670 |
+
{
|
| 35671 |
+
"epoch": 0.20869173425083967,
|
| 35672 |
+
"grad_norm": 0.2670971155166626,
|
| 35673 |
+
"learning_rate": 0.00044971675076299736,
|
| 35674 |
+
"loss": 0.7085,
|
| 35675 |
+
"step": 50950
|
| 35676 |
+
},
|
| 35677 |
+
{
|
| 35678 |
+
"epoch": 0.20873269435569755,
|
| 35679 |
+
"grad_norm": 0.25000834465026855,
|
| 35680 |
+
"learning_rate": 0.00044969731901276335,
|
| 35681 |
+
"loss": 0.7134,
|
| 35682 |
+
"step": 50960
|
| 35683 |
+
},
|
| 35684 |
+
{
|
| 35685 |
+
"epoch": 0.20877365446055543,
|
| 35686 |
+
"grad_norm": 0.30456289649009705,
|
| 35687 |
+
"learning_rate": 0.00044967788392857877,
|
| 35688 |
+
"loss": 0.7104,
|
| 35689 |
+
"step": 50970
|
| 35690 |
+
},
|
| 35691 |
+
{
|
| 35692 |
+
"epoch": 0.20881461456541328,
|
| 35693 |
+
"grad_norm": 0.2588064670562744,
|
| 35694 |
+
"learning_rate": 0.0004496584455107681,
|
| 35695 |
+
"loss": 0.7037,
|
| 35696 |
+
"step": 50980
|
| 35697 |
+
},
|
| 35698 |
+
{
|
| 35699 |
+
"epoch": 0.20885557467027116,
|
| 35700 |
+
"grad_norm": 0.26450926065444946,
|
| 35701 |
+
"learning_rate": 0.00044963900375965594,
|
| 35702 |
+
"loss": 0.7064,
|
| 35703 |
+
"step": 50990
|
| 35704 |
+
},
|
| 35705 |
+
{
|
| 35706 |
+
"epoch": 0.208896534775129,
|
| 35707 |
+
"grad_norm": 0.2816455066204071,
|
| 35708 |
+
"learning_rate": 0.0004496195586755667,
|
| 35709 |
+
"loss": 0.7077,
|
| 35710 |
+
"step": 51000
|
| 35711 |
}
|
| 35712 |
],
|
| 35713 |
"logging_steps": 10,
|
|
|
|
| 35727 |
"attributes": {}
|
| 35728 |
}
|
| 35729 |
},
|
| 35730 |
+
"total_flos": 5.061118972133376e+18,
|
| 35731 |
"train_batch_size": 2048,
|
| 35732 |
"trial_name": null,
|
| 35733 |
"trial_params": null
|
last-checkpoint/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4b71a3f84ffeb83af792485702e64fd44ffac28210d09ba5bfee8116a29eee3
|
| 3 |
+
size 5969
|