Training in progress, step 7000
Browse files- adapter_model.safetensors +1 -1
- last-checkpoint/adapter_config.json +7 -7
- last-checkpoint/adapter_model.safetensors +2 -2
- last-checkpoint/optimizer.pt +2 -2
- last-checkpoint/rng_state_0.pth +1 -1
- last-checkpoint/rng_state_1.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +6 -3138
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88891320
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e62960b69826660dec7e43ee28d0b6065d22575071d9d4f842e3e41111a24ff
|
| 3 |
size 88891320
|
last-checkpoint/adapter_config.json
CHANGED
|
@@ -30,15 +30,15 @@
|
|
| 30 |
"rank_pattern": {},
|
| 31 |
"revision": null,
|
| 32 |
"target_modules": [
|
| 33 |
-
"k_proj",
|
| 34 |
-
"in_proj",
|
| 35 |
-
"v_proj",
|
| 36 |
-
"out_proj",
|
| 37 |
-
"w2",
|
| 38 |
-
"w3",
|
| 39 |
"o_proj",
|
|
|
|
|
|
|
| 40 |
"w1",
|
| 41 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
],
|
| 43 |
"target_parameters": null,
|
| 44 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 30 |
"rank_pattern": {},
|
| 31 |
"revision": null,
|
| 32 |
"target_modules": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
"o_proj",
|
| 34 |
+
"w3",
|
| 35 |
+
"v_proj",
|
| 36 |
"w1",
|
| 37 |
+
"w2",
|
| 38 |
+
"q_proj",
|
| 39 |
+
"k_proj",
|
| 40 |
+
"in_proj",
|
| 41 |
+
"out_proj"
|
| 42 |
],
|
| 43 |
"target_parameters": null,
|
| 44 |
"task_type": "CAUSAL_LM",
|
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e4de68cc4a27d132fc06cff173852cfe5bfcb085c0504d028f81ef223e60d79
|
| 3 |
+
size 44458040
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d110f19bc9ff2b7c2d3cba4527735838df492ca5c989d63b2197e5b5a83a1be
|
| 3 |
+
size 89022475
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14917
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b4c2cedb8b1a315523827aa4607bcb75c00b9402e6760b44f99a9078b817150
|
| 3 |
size 14917
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14917
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:51b5aa279ae1bb3a6481daa322a25dd97361e385419089b20cd7988c11789409
|
| 3 |
size 14917
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0f8412c59e3898496baf69de3251cb166157d59299f2778285bb5cba6025112
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 0.
|
| 4 |
-
"best_model_checkpoint": "./lfm25-kodcode-lora/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 250,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -3923,3138 +3923,6 @@
|
|
| 3923 |
"eval_samples_per_second": 5.87,
|
| 3924 |
"eval_steps_per_second": 1.469,
|
| 3925 |
"step": 3750
|
| 3926 |
-
},
|
| 3927 |
-
{
|
| 3928 |
-
"entropy": 0.3068324435502291,
|
| 3929 |
-
"epoch": 0.2848538798083297,
|
| 3930 |
-
"grad_norm": 0.4066140353679657,
|
| 3931 |
-
"learning_rate": 8.392246647096156e-05,
|
| 3932 |
-
"loss": 0.34855701923370364,
|
| 3933 |
-
"mean_token_accuracy": 0.9019667461514473,
|
| 3934 |
-
"num_tokens": 101388.0,
|
| 3935 |
-
"step": 3760
|
| 3936 |
-
},
|
| 3937 |
-
{
|
| 3938 |
-
"entropy": 0.24433709159493447,
|
| 3939 |
-
"epoch": 0.28561146991420294,
|
| 3940 |
-
"grad_norm": 0.5887599587440491,
|
| 3941 |
-
"learning_rate": 8.38322378414151e-05,
|
| 3942 |
-
"loss": 0.2643440246582031,
|
| 3943 |
-
"mean_token_accuracy": 0.9256282597780228,
|
| 3944 |
-
"num_tokens": 195786.0,
|
| 3945 |
-
"step": 3770
|
| 3946 |
-
},
|
| 3947 |
-
{
|
| 3948 |
-
"entropy": 0.304387878254056,
|
| 3949 |
-
"epoch": 0.2863690600200761,
|
| 3950 |
-
"grad_norm": 0.4766692817211151,
|
| 3951 |
-
"learning_rate": 8.374180553630735e-05,
|
| 3952 |
-
"loss": 0.3629739284515381,
|
| 3953 |
-
"mean_token_accuracy": 0.8997384190559388,
|
| 3954 |
-
"num_tokens": 294546.0,
|
| 3955 |
-
"step": 3780
|
| 3956 |
-
},
|
| 3957 |
-
{
|
| 3958 |
-
"entropy": 0.26069591380655766,
|
| 3959 |
-
"epoch": 0.28712665012594935,
|
| 3960 |
-
"grad_norm": 0.4924900233745575,
|
| 3961 |
-
"learning_rate": 8.365117010005546e-05,
|
| 3962 |
-
"loss": 0.29716243743896487,
|
| 3963 |
-
"mean_token_accuracy": 0.9197860583662987,
|
| 3964 |
-
"num_tokens": 382234.0,
|
| 3965 |
-
"step": 3790
|
| 3966 |
-
},
|
| 3967 |
-
{
|
| 3968 |
-
"entropy": 0.3133056215941906,
|
| 3969 |
-
"epoch": 0.2878842402318226,
|
| 3970 |
-
"grad_norm": 0.6270611882209778,
|
| 3971 |
-
"learning_rate": 8.356033207829948e-05,
|
| 3972 |
-
"loss": 0.373241925239563,
|
| 3973 |
-
"mean_token_accuracy": 0.899902306497097,
|
| 3974 |
-
"num_tokens": 487216.0,
|
| 3975 |
-
"step": 3800
|
| 3976 |
-
},
|
| 3977 |
-
{
|
| 3978 |
-
"entropy": 0.29544789865612986,
|
| 3979 |
-
"epoch": 0.2886418303376958,
|
| 3980 |
-
"grad_norm": 0.39994874596595764,
|
| 3981 |
-
"learning_rate": 8.3469292017899e-05,
|
| 3982 |
-
"loss": 0.3260920286178589,
|
| 3983 |
-
"mean_token_accuracy": 0.9073387667536735,
|
| 3984 |
-
"num_tokens": 589454.0,
|
| 3985 |
-
"step": 3810
|
| 3986 |
-
},
|
| 3987 |
-
{
|
| 3988 |
-
"entropy": 0.3384067412465811,
|
| 3989 |
-
"epoch": 0.28939942044356903,
|
| 3990 |
-
"grad_norm": 0.4351018965244293,
|
| 3991 |
-
"learning_rate": 8.337805046692999e-05,
|
| 3992 |
-
"loss": 0.37573843002319335,
|
| 3993 |
-
"mean_token_accuracy": 0.8918576836585999,
|
| 3994 |
-
"num_tokens": 699999.0,
|
| 3995 |
-
"step": 3820
|
| 3996 |
-
},
|
| 3997 |
-
{
|
| 3998 |
-
"entropy": 0.28844635635614396,
|
| 3999 |
-
"epoch": 0.2901570105494422,
|
| 4000 |
-
"grad_norm": 0.39798203110694885,
|
| 4001 |
-
"learning_rate": 8.328660797468136e-05,
|
| 4002 |
-
"loss": 0.3440295696258545,
|
| 4003 |
-
"mean_token_accuracy": 0.9074199810624123,
|
| 4004 |
-
"num_tokens": 792965.0,
|
| 4005 |
-
"step": 3830
|
| 4006 |
-
},
|
| 4007 |
-
{
|
| 4008 |
-
"entropy": 0.29644420053809883,
|
| 4009 |
-
"epoch": 0.29091460065531544,
|
| 4010 |
-
"grad_norm": 0.5635829567909241,
|
| 4011 |
-
"learning_rate": 8.31949650916518e-05,
|
| 4012 |
-
"loss": 0.33669238090515136,
|
| 4013 |
-
"mean_token_accuracy": 0.9057722166180611,
|
| 4014 |
-
"num_tokens": 885671.0,
|
| 4015 |
-
"step": 3840
|
| 4016 |
-
},
|
| 4017 |
-
{
|
| 4018 |
-
"entropy": 0.2981845339760184,
|
| 4019 |
-
"epoch": 0.29167219076118867,
|
| 4020 |
-
"grad_norm": 0.5180003046989441,
|
| 4021 |
-
"learning_rate": 8.310312236954627e-05,
|
| 4022 |
-
"loss": 0.34060158729553225,
|
| 4023 |
-
"mean_token_accuracy": 0.9069652289152146,
|
| 4024 |
-
"num_tokens": 986080.0,
|
| 4025 |
-
"step": 3850
|
| 4026 |
-
},
|
| 4027 |
-
{
|
| 4028 |
-
"entropy": 0.3225794995203614,
|
| 4029 |
-
"epoch": 0.2924297808670619,
|
| 4030 |
-
"grad_norm": 0.3230516016483307,
|
| 4031 |
-
"learning_rate": 8.30110803612729e-05,
|
| 4032 |
-
"loss": 0.35950822830200196,
|
| 4033 |
-
"mean_token_accuracy": 0.8959019765257835,
|
| 4034 |
-
"num_tokens": 1096595.0,
|
| 4035 |
-
"step": 3860
|
| 4036 |
-
},
|
| 4037 |
-
{
|
| 4038 |
-
"entropy": 0.27073352932929995,
|
| 4039 |
-
"epoch": 0.29318737097293507,
|
| 4040 |
-
"grad_norm": 0.5060323476791382,
|
| 4041 |
-
"learning_rate": 8.291883962093951e-05,
|
| 4042 |
-
"loss": 0.3100705146789551,
|
| 4043 |
-
"mean_token_accuracy": 0.9122787326574325,
|
| 4044 |
-
"num_tokens": 1190705.0,
|
| 4045 |
-
"step": 3870
|
| 4046 |
-
},
|
| 4047 |
-
{
|
| 4048 |
-
"entropy": 0.3063302764669061,
|
| 4049 |
-
"epoch": 0.2939449610788083,
|
| 4050 |
-
"grad_norm": 0.4379747211933136,
|
| 4051 |
-
"learning_rate": 8.282640070385033e-05,
|
| 4052 |
-
"loss": 0.3423487901687622,
|
| 4053 |
-
"mean_token_accuracy": 0.9022525697946548,
|
| 4054 |
-
"num_tokens": 1290107.0,
|
| 4055 |
-
"step": 3880
|
| 4056 |
-
},
|
| 4057 |
-
{
|
| 4058 |
-
"entropy": 0.3316929185763001,
|
| 4059 |
-
"epoch": 0.2947025511846815,
|
| 4060 |
-
"grad_norm": 0.3801766037940979,
|
| 4061 |
-
"learning_rate": 8.273376416650263e-05,
|
| 4062 |
-
"loss": 0.37284531593322756,
|
| 4063 |
-
"mean_token_accuracy": 0.895954342186451,
|
| 4064 |
-
"num_tokens": 1395753.0,
|
| 4065 |
-
"step": 3890
|
| 4066 |
-
},
|
| 4067 |
-
{
|
| 4068 |
-
"entropy": 0.31571994014084337,
|
| 4069 |
-
"epoch": 0.29546014129055476,
|
| 4070 |
-
"grad_norm": 0.43525663018226624,
|
| 4071 |
-
"learning_rate": 8.26409305665834e-05,
|
| 4072 |
-
"loss": 0.36900758743286133,
|
| 4073 |
-
"mean_token_accuracy": 0.8997189715504647,
|
| 4074 |
-
"num_tokens": 1500536.0,
|
| 4075 |
-
"step": 3900
|
| 4076 |
-
},
|
| 4077 |
-
{
|
| 4078 |
-
"entropy": 0.32693297136574984,
|
| 4079 |
-
"epoch": 0.296217731396428,
|
| 4080 |
-
"grad_norm": 0.43510177731513977,
|
| 4081 |
-
"learning_rate": 8.254790046296595e-05,
|
| 4082 |
-
"loss": 0.37863659858703613,
|
| 4083 |
-
"mean_token_accuracy": 0.8956527307629585,
|
| 4084 |
-
"num_tokens": 1603202.0,
|
| 4085 |
-
"step": 3910
|
| 4086 |
-
},
|
| 4087 |
-
{
|
| 4088 |
-
"entropy": 0.2829545237123966,
|
| 4089 |
-
"epoch": 0.29697532150230116,
|
| 4090 |
-
"grad_norm": 0.40382707118988037,
|
| 4091 |
-
"learning_rate": 8.245467441570662e-05,
|
| 4092 |
-
"loss": 0.33570911884307864,
|
| 4093 |
-
"mean_token_accuracy": 0.9099944919347763,
|
| 4094 |
-
"num_tokens": 1702281.0,
|
| 4095 |
-
"step": 3920
|
| 4096 |
-
},
|
| 4097 |
-
{
|
| 4098 |
-
"entropy": 0.3078343799337745,
|
| 4099 |
-
"epoch": 0.2977329116081744,
|
| 4100 |
-
"grad_norm": 0.5139282941818237,
|
| 4101 |
-
"learning_rate": 8.236125298604136e-05,
|
| 4102 |
-
"loss": 0.345384407043457,
|
| 4103 |
-
"mean_token_accuracy": 0.9006545335054398,
|
| 4104 |
-
"num_tokens": 1794500.0,
|
| 4105 |
-
"step": 3930
|
| 4106 |
-
},
|
| 4107 |
-
{
|
| 4108 |
-
"entropy": 0.33859921526163816,
|
| 4109 |
-
"epoch": 0.2984905017140476,
|
| 4110 |
-
"grad_norm": 0.5043326020240784,
|
| 4111 |
-
"learning_rate": 8.22676367363823e-05,
|
| 4112 |
-
"loss": 0.36715083122253417,
|
| 4113 |
-
"mean_token_accuracy": 0.8948409363627434,
|
| 4114 |
-
"num_tokens": 1899621.0,
|
| 4115 |
-
"step": 3940
|
| 4116 |
-
},
|
| 4117 |
-
{
|
| 4118 |
-
"entropy": 0.34319984167814255,
|
| 4119 |
-
"epoch": 0.29924809181992085,
|
| 4120 |
-
"grad_norm": 0.5146629214286804,
|
| 4121 |
-
"learning_rate": 8.217382623031446e-05,
|
| 4122 |
-
"loss": 0.42072219848632814,
|
| 4123 |
-
"mean_token_accuracy": 0.8903807610273361,
|
| 4124 |
-
"num_tokens": 2007584.0,
|
| 4125 |
-
"step": 3950
|
| 4126 |
-
},
|
| 4127 |
-
{
|
| 4128 |
-
"entropy": 0.3177859207615256,
|
| 4129 |
-
"epoch": 0.3000056819257941,
|
| 4130 |
-
"grad_norm": 0.5011279582977295,
|
| 4131 |
-
"learning_rate": 8.207982203259234e-05,
|
| 4132 |
-
"loss": 0.36820187568664553,
|
| 4133 |
-
"mean_token_accuracy": 0.9007480621337891,
|
| 4134 |
-
"num_tokens": 2103608.0,
|
| 4135 |
-
"step": 3960
|
| 4136 |
-
},
|
| 4137 |
-
{
|
| 4138 |
-
"entropy": 0.2695161834359169,
|
| 4139 |
-
"epoch": 0.30076327203166725,
|
| 4140 |
-
"grad_norm": 0.4234781861305237,
|
| 4141 |
-
"learning_rate": 8.198562470913643e-05,
|
| 4142 |
-
"loss": 0.31445822715759275,
|
| 4143 |
-
"mean_token_accuracy": 0.9147355154156684,
|
| 4144 |
-
"num_tokens": 2206435.0,
|
| 4145 |
-
"step": 3970
|
| 4146 |
-
},
|
| 4147 |
-
{
|
| 4148 |
-
"entropy": 0.30907482504844663,
|
| 4149 |
-
"epoch": 0.3015208621375405,
|
| 4150 |
-
"grad_norm": 0.5427946448326111,
|
| 4151 |
-
"learning_rate": 8.189123482702991e-05,
|
| 4152 |
-
"loss": 0.359407639503479,
|
| 4153 |
-
"mean_token_accuracy": 0.8986603081226349,
|
| 4154 |
-
"num_tokens": 2303425.0,
|
| 4155 |
-
"step": 3980
|
| 4156 |
-
},
|
| 4157 |
-
{
|
| 4158 |
-
"entropy": 0.2811237759888172,
|
| 4159 |
-
"epoch": 0.3022784522434137,
|
| 4160 |
-
"grad_norm": 0.4475630223751068,
|
| 4161 |
-
"learning_rate": 8.179665295451518e-05,
|
| 4162 |
-
"loss": 0.3343693733215332,
|
| 4163 |
-
"mean_token_accuracy": 0.9098231837153434,
|
| 4164 |
-
"num_tokens": 2399105.0,
|
| 4165 |
-
"step": 3990
|
| 4166 |
-
},
|
| 4167 |
-
{
|
| 4168 |
-
"entropy": 0.30583363566547633,
|
| 4169 |
-
"epoch": 0.30303604234928694,
|
| 4170 |
-
"grad_norm": 0.4408576190471649,
|
| 4171 |
-
"learning_rate": 8.170187966099045e-05,
|
| 4172 |
-
"loss": 0.3527429819107056,
|
| 4173 |
-
"mean_token_accuracy": 0.9045899525284767,
|
| 4174 |
-
"num_tokens": 2491862.0,
|
| 4175 |
-
"step": 4000
|
| 4176 |
-
},
|
| 4177 |
-
{
|
| 4178 |
-
"epoch": 0.30303604234928694,
|
| 4179 |
-
"eval_entropy": 0.31009396722310045,
|
| 4180 |
-
"eval_loss": 0.3019750714302063,
|
| 4181 |
-
"eval_mean_token_accuracy": 0.9028535516521011,
|
| 4182 |
-
"eval_num_tokens": 2491862.0,
|
| 4183 |
-
"eval_runtime": 357.3942,
|
| 4184 |
-
"eval_samples_per_second": 5.971,
|
| 4185 |
-
"eval_steps_per_second": 1.494,
|
| 4186 |
-
"step": 4000
|
| 4187 |
-
},
|
| 4188 |
-
{
|
| 4189 |
-
"entropy": 0.3382018800824881,
|
| 4190 |
-
"epoch": 0.3037936324551601,
|
| 4191 |
-
"grad_norm": 0.42774832248687744,
|
| 4192 |
-
"learning_rate": 8.160691551700637e-05,
|
| 4193 |
-
"loss": 0.37759079933166506,
|
| 4194 |
-
"mean_token_accuracy": 0.8950443550944328,
|
| 4195 |
-
"num_tokens": 2600410.0,
|
| 4196 |
-
"step": 4010
|
| 4197 |
-
},
|
| 4198 |
-
{
|
| 4199 |
-
"entropy": 0.28815502356737854,
|
| 4200 |
-
"epoch": 0.30455122256103334,
|
| 4201 |
-
"grad_norm": 0.4374275207519531,
|
| 4202 |
-
"learning_rate": 8.151176109426244e-05,
|
| 4203 |
-
"loss": 0.3317250728607178,
|
| 4204 |
-
"mean_token_accuracy": 0.9076664626598359,
|
| 4205 |
-
"num_tokens": 2694551.0,
|
| 4206 |
-
"step": 4020
|
| 4207 |
-
},
|
| 4208 |
-
{
|
| 4209 |
-
"entropy": 0.27126045860350134,
|
| 4210 |
-
"epoch": 0.30530881266690657,
|
| 4211 |
-
"grad_norm": 0.3163757920265198,
|
| 4212 |
-
"learning_rate": 8.141641696560376e-05,
|
| 4213 |
-
"loss": 0.30475554466247556,
|
| 4214 |
-
"mean_token_accuracy": 0.9125639736652374,
|
| 4215 |
-
"num_tokens": 2786280.0,
|
| 4216 |
-
"step": 4030
|
| 4217 |
-
},
|
| 4218 |
-
{
|
| 4219 |
-
"entropy": 0.31229337155818937,
|
| 4220 |
-
"epoch": 0.3060664027727798,
|
| 4221 |
-
"grad_norm": 0.36219096183776855,
|
| 4222 |
-
"learning_rate": 8.132088370501743e-05,
|
| 4223 |
-
"loss": 0.355154824256897,
|
| 4224 |
-
"mean_token_accuracy": 0.8994580417871475,
|
| 4225 |
-
"num_tokens": 2892654.0,
|
| 4226 |
-
"step": 4040
|
| 4227 |
-
},
|
| 4228 |
-
{
|
| 4229 |
-
"entropy": 0.3099927732720971,
|
| 4230 |
-
"epoch": 0.30682399287865303,
|
| 4231 |
-
"grad_norm": 0.4749106168746948,
|
| 4232 |
-
"learning_rate": 8.12251618876292e-05,
|
| 4233 |
-
"loss": 0.3676262617111206,
|
| 4234 |
-
"mean_token_accuracy": 0.9003804937005043,
|
| 4235 |
-
"num_tokens": 2991264.0,
|
| 4236 |
-
"step": 4050
|
| 4237 |
-
},
|
| 4238 |
-
{
|
| 4239 |
-
"entropy": 0.2964092491194606,
|
| 4240 |
-
"epoch": 0.3075815829845262,
|
| 4241 |
-
"grad_norm": 0.4777383506298065,
|
| 4242 |
-
"learning_rate": 8.112925208969995e-05,
|
| 4243 |
-
"loss": 0.362530517578125,
|
| 4244 |
-
"mean_token_accuracy": 0.9084259837865829,
|
| 4245 |
-
"num_tokens": 3076708.0,
|
| 4246 |
-
"step": 4060
|
| 4247 |
-
},
|
| 4248 |
-
{
|
| 4249 |
-
"entropy": 0.3478623412549496,
|
| 4250 |
-
"epoch": 0.30833917309039943,
|
| 4251 |
-
"grad_norm": 0.4825819134712219,
|
| 4252 |
-
"learning_rate": 8.10331548886222e-05,
|
| 4253 |
-
"loss": 0.38155560493469237,
|
| 4254 |
-
"mean_token_accuracy": 0.8893183171749115,
|
| 4255 |
-
"num_tokens": 3181679.0,
|
| 4256 |
-
"step": 4070
|
| 4257 |
-
},
|
| 4258 |
-
{
|
| 4259 |
-
"entropy": 0.2972083490341902,
|
| 4260 |
-
"epoch": 0.30909676319627266,
|
| 4261 |
-
"grad_norm": 0.3799072802066803,
|
| 4262 |
-
"learning_rate": 8.09368708629167e-05,
|
| 4263 |
-
"loss": 0.3379563331604004,
|
| 4264 |
-
"mean_token_accuracy": 0.9061253800988197,
|
| 4265 |
-
"num_tokens": 3289178.0,
|
| 4266 |
-
"step": 4080
|
| 4267 |
-
},
|
| 4268 |
-
{
|
| 4269 |
-
"entropy": 0.279171503521502,
|
| 4270 |
-
"epoch": 0.3098543533021459,
|
| 4271 |
-
"grad_norm": 0.4669226109981537,
|
| 4272 |
-
"learning_rate": 8.084040059222891e-05,
|
| 4273 |
-
"loss": 0.32326531410217285,
|
| 4274 |
-
"mean_token_accuracy": 0.9118024915456772,
|
| 4275 |
-
"num_tokens": 3380600.0,
|
| 4276 |
-
"step": 4090
|
| 4277 |
-
},
|
| 4278 |
-
{
|
| 4279 |
-
"entropy": 0.27288217693567274,
|
| 4280 |
-
"epoch": 0.31061194340801906,
|
| 4281 |
-
"grad_norm": 0.4976431429386139,
|
| 4282 |
-
"learning_rate": 8.07437446573255e-05,
|
| 4283 |
-
"loss": 0.302996301651001,
|
| 4284 |
-
"mean_token_accuracy": 0.914215661585331,
|
| 4285 |
-
"num_tokens": 3478814.0,
|
| 4286 |
-
"step": 4100
|
| 4287 |
-
},
|
| 4288 |
-
{
|
| 4289 |
-
"entropy": 0.29566774647682903,
|
| 4290 |
-
"epoch": 0.3113695335138923,
|
| 4291 |
-
"grad_norm": 0.49561217427253723,
|
| 4292 |
-
"learning_rate": 8.064690364009091e-05,
|
| 4293 |
-
"loss": 0.3245049715042114,
|
| 4294 |
-
"mean_token_accuracy": 0.9081421956419945,
|
| 4295 |
-
"num_tokens": 3576216.0,
|
| 4296 |
-
"step": 4110
|
| 4297 |
-
},
|
| 4298 |
-
{
|
| 4299 |
-
"entropy": 0.3129298511892557,
|
| 4300 |
-
"epoch": 0.3121271236197655,
|
| 4301 |
-
"grad_norm": 0.3924306035041809,
|
| 4302 |
-
"learning_rate": 8.054987812352377e-05,
|
| 4303 |
-
"loss": 0.37849423885345457,
|
| 4304 |
-
"mean_token_accuracy": 0.8973022267222405,
|
| 4305 |
-
"num_tokens": 3684317.0,
|
| 4306 |
-
"step": 4120
|
| 4307 |
-
},
|
| 4308 |
-
{
|
| 4309 |
-
"entropy": 0.29853209499269723,
|
| 4310 |
-
"epoch": 0.31288471372563875,
|
| 4311 |
-
"grad_norm": 0.36907029151916504,
|
| 4312 |
-
"learning_rate": 8.045266869173341e-05,
|
| 4313 |
-
"loss": 0.3402961015701294,
|
| 4314 |
-
"mean_token_accuracy": 0.9055958077311516,
|
| 4315 |
-
"num_tokens": 3786052.0,
|
| 4316 |
-
"step": 4130
|
| 4317 |
-
},
|
| 4318 |
-
{
|
| 4319 |
-
"entropy": 0.3380041945725679,
|
| 4320 |
-
"epoch": 0.313642303831512,
|
| 4321 |
-
"grad_norm": 0.5555438995361328,
|
| 4322 |
-
"learning_rate": 8.03552759299364e-05,
|
| 4323 |
-
"loss": 0.3785285949707031,
|
| 4324 |
-
"mean_token_accuracy": 0.8910282999277115,
|
| 4325 |
-
"num_tokens": 3896833.0,
|
| 4326 |
-
"step": 4140
|
| 4327 |
-
},
|
| 4328 |
-
{
|
| 4329 |
-
"entropy": 0.30070192124694584,
|
| 4330 |
-
"epoch": 0.31439989393738516,
|
| 4331 |
-
"grad_norm": 0.415605753660202,
|
| 4332 |
-
"learning_rate": 8.025770042445295e-05,
|
| 4333 |
-
"loss": 0.3519805908203125,
|
| 4334 |
-
"mean_token_accuracy": 0.905980184674263,
|
| 4335 |
-
"num_tokens": 3997864.0,
|
| 4336 |
-
"step": 4150
|
| 4337 |
-
},
|
| 4338 |
-
{
|
| 4339 |
-
"entropy": 0.2683778068050742,
|
| 4340 |
-
"epoch": 0.3151574840432584,
|
| 4341 |
-
"grad_norm": 0.49308788776397705,
|
| 4342 |
-
"learning_rate": 8.015994276270346e-05,
|
| 4343 |
-
"loss": 0.32012176513671875,
|
| 4344 |
-
"mean_token_accuracy": 0.9152865901589393,
|
| 4345 |
-
"num_tokens": 4095534.0,
|
| 4346 |
-
"step": 4160
|
| 4347 |
-
},
|
| 4348 |
-
{
|
| 4349 |
-
"entropy": 0.32295512650161984,
|
| 4350 |
-
"epoch": 0.3159150741491316,
|
| 4351 |
-
"grad_norm": 0.4559709131717682,
|
| 4352 |
-
"learning_rate": 8.006200353320492e-05,
|
| 4353 |
-
"loss": 0.35291566848754885,
|
| 4354 |
-
"mean_token_accuracy": 0.8978117927908897,
|
| 4355 |
-
"num_tokens": 4195235.0,
|
| 4356 |
-
"step": 4170
|
| 4357 |
-
},
|
| 4358 |
-
{
|
| 4359 |
-
"entropy": 0.2599181361496449,
|
| 4360 |
-
"epoch": 0.31667266425500484,
|
| 4361 |
-
"grad_norm": 0.43297865986824036,
|
| 4362 |
-
"learning_rate": 7.996388332556735e-05,
|
| 4363 |
-
"loss": 0.30230867862701416,
|
| 4364 |
-
"mean_token_accuracy": 0.9172627389431,
|
| 4365 |
-
"num_tokens": 4280769.0,
|
| 4366 |
-
"step": 4180
|
| 4367 |
-
},
|
| 4368 |
-
{
|
| 4369 |
-
"entropy": 0.3053940389305353,
|
| 4370 |
-
"epoch": 0.3174302543608781,
|
| 4371 |
-
"grad_norm": 0.5943090319633484,
|
| 4372 |
-
"learning_rate": 7.986558273049034e-05,
|
| 4373 |
-
"loss": 0.35530800819396974,
|
| 4374 |
-
"mean_token_accuracy": 0.9012864857912064,
|
| 4375 |
-
"num_tokens": 4375588.0,
|
| 4376 |
-
"step": 4190
|
| 4377 |
-
},
|
| 4378 |
-
{
|
| 4379 |
-
"entropy": 0.2867847694084048,
|
| 4380 |
-
"epoch": 0.31818784446675125,
|
| 4381 |
-
"grad_norm": 0.48145633935928345,
|
| 4382 |
-
"learning_rate": 7.976710233975944e-05,
|
| 4383 |
-
"loss": 0.3357656002044678,
|
| 4384 |
-
"mean_token_accuracy": 0.9119586288928986,
|
| 4385 |
-
"num_tokens": 4474675.0,
|
| 4386 |
-
"step": 4200
|
| 4387 |
-
},
|
| 4388 |
-
{
|
| 4389 |
-
"entropy": 0.27073222529143093,
|
| 4390 |
-
"epoch": 0.3189454345726245,
|
| 4391 |
-
"grad_norm": 0.3421247601509094,
|
| 4392 |
-
"learning_rate": 7.966844274624254e-05,
|
| 4393 |
-
"loss": 0.31576545238494874,
|
| 4394 |
-
"mean_token_accuracy": 0.9127877086400986,
|
| 4395 |
-
"num_tokens": 4573163.0,
|
| 4396 |
-
"step": 4210
|
| 4397 |
-
},
|
| 4398 |
-
{
|
| 4399 |
-
"entropy": 0.3082149105146527,
|
| 4400 |
-
"epoch": 0.3197030246784977,
|
| 4401 |
-
"grad_norm": 0.3288472890853882,
|
| 4402 |
-
"learning_rate": 7.956960454388642e-05,
|
| 4403 |
-
"loss": 0.3599945068359375,
|
| 4404 |
-
"mean_token_accuracy": 0.9014466717839241,
|
| 4405 |
-
"num_tokens": 4675077.0,
|
| 4406 |
-
"step": 4220
|
| 4407 |
-
},
|
| 4408 |
-
{
|
| 4409 |
-
"entropy": 0.29703692328184844,
|
| 4410 |
-
"epoch": 0.32046061478437093,
|
| 4411 |
-
"grad_norm": 0.4654316306114197,
|
| 4412 |
-
"learning_rate": 7.947058832771311e-05,
|
| 4413 |
-
"loss": 0.3311642646789551,
|
| 4414 |
-
"mean_token_accuracy": 0.9072018593549729,
|
| 4415 |
-
"num_tokens": 4775595.0,
|
| 4416 |
-
"step": 4230
|
| 4417 |
-
},
|
| 4418 |
-
{
|
| 4419 |
-
"entropy": 0.33092996161431076,
|
| 4420 |
-
"epoch": 0.3212182048902441,
|
| 4421 |
-
"grad_norm": 0.4268597364425659,
|
| 4422 |
-
"learning_rate": 7.93713946938163e-05,
|
| 4423 |
-
"loss": 0.39023890495300295,
|
| 4424 |
-
"mean_token_accuracy": 0.8954888239502907,
|
| 4425 |
-
"num_tokens": 4874324.0,
|
| 4426 |
-
"step": 4240
|
| 4427 |
-
},
|
| 4428 |
-
{
|
| 4429 |
-
"entropy": 0.30921045057475566,
|
| 4430 |
-
"epoch": 0.32197579499611734,
|
| 4431 |
-
"grad_norm": 0.44206345081329346,
|
| 4432 |
-
"learning_rate": 7.927202423935771e-05,
|
| 4433 |
-
"loss": 0.36301393508911134,
|
| 4434 |
-
"mean_token_accuracy": 0.9011395126581192,
|
| 4435 |
-
"num_tokens": 4977513.0,
|
| 4436 |
-
"step": 4250
|
| 4437 |
-
},
|
| 4438 |
-
{
|
| 4439 |
-
"epoch": 0.32197579499611734,
|
| 4440 |
-
"eval_entropy": 0.30250602451324016,
|
| 4441 |
-
"eval_loss": 0.3017820715904236,
|
| 4442 |
-
"eval_mean_token_accuracy": 0.9029686442028717,
|
| 4443 |
-
"eval_num_tokens": 4977513.0,
|
| 4444 |
-
"eval_runtime": 357.3743,
|
| 4445 |
-
"eval_samples_per_second": 5.971,
|
| 4446 |
-
"eval_steps_per_second": 1.494,
|
| 4447 |
-
"step": 4250
|
| 4448 |
-
},
|
| 4449 |
-
{
|
| 4450 |
-
"entropy": 0.2946730049327016,
|
| 4451 |
-
"epoch": 0.32273338510199057,
|
| 4452 |
-
"grad_norm": 0.5596773624420166,
|
| 4453 |
-
"learning_rate": 7.917247756256365e-05,
|
| 4454 |
-
"loss": 0.34336128234863283,
|
| 4455 |
-
"mean_token_accuracy": 0.9069753110408783,
|
| 4456 |
-
"num_tokens": 5072409.0,
|
| 4457 |
-
"step": 4260
|
| 4458 |
-
},
|
| 4459 |
-
{
|
| 4460 |
-
"entropy": 0.3777470774948597,
|
| 4461 |
-
"epoch": 0.3234909752078638,
|
| 4462 |
-
"grad_norm": 0.43069401383399963,
|
| 4463 |
-
"learning_rate": 7.907275526272126e-05,
|
| 4464 |
-
"loss": 0.4284689903259277,
|
| 4465 |
-
"mean_token_accuracy": 0.8780335173010826,
|
| 4466 |
-
"num_tokens": 5177223.0,
|
| 4467 |
-
"step": 4270
|
| 4468 |
-
},
|
| 4469 |
-
{
|
| 4470 |
-
"entropy": 0.29496462028473613,
|
| 4471 |
-
"epoch": 0.324248565313737,
|
| 4472 |
-
"grad_norm": 0.4127286672592163,
|
| 4473 |
-
"learning_rate": 7.897285794017496e-05,
|
| 4474 |
-
"loss": 0.33785984516143797,
|
| 4475 |
-
"mean_token_accuracy": 0.9069562211632729,
|
| 4476 |
-
"num_tokens": 5272846.0,
|
| 4477 |
-
"step": 4280
|
| 4478 |
-
},
|
| 4479 |
-
{
|
| 4480 |
-
"entropy": 0.3130815394222736,
|
| 4481 |
-
"epoch": 0.3250061554196102,
|
| 4482 |
-
"grad_norm": 0.40738624334335327,
|
| 4483 |
-
"learning_rate": 7.887278619632283e-05,
|
| 4484 |
-
"loss": 0.3703619480133057,
|
| 4485 |
-
"mean_token_accuracy": 0.9007998913526535,
|
| 4486 |
-
"num_tokens": 5375828.0,
|
| 4487 |
-
"step": 4290
|
| 4488 |
-
},
|
| 4489 |
-
{
|
| 4490 |
-
"entropy": 0.29860169291496275,
|
| 4491 |
-
"epoch": 0.32576374552548343,
|
| 4492 |
-
"grad_norm": 0.37904271483421326,
|
| 4493 |
-
"learning_rate": 7.877254063361302e-05,
|
| 4494 |
-
"loss": 0.3511554718017578,
|
| 4495 |
-
"mean_token_accuracy": 0.9063147380948067,
|
| 4496 |
-
"num_tokens": 5482999.0,
|
| 4497 |
-
"step": 4300
|
| 4498 |
-
},
|
| 4499 |
-
{
|
| 4500 |
-
"entropy": 0.3422315252944827,
|
| 4501 |
-
"epoch": 0.32652133563135666,
|
| 4502 |
-
"grad_norm": 0.4242798388004303,
|
| 4503 |
-
"learning_rate": 7.867212185554007e-05,
|
| 4504 |
-
"loss": 0.40230045318603513,
|
| 4505 |
-
"mean_token_accuracy": 0.8903420940041542,
|
| 4506 |
-
"num_tokens": 5584694.0,
|
| 4507 |
-
"step": 4310
|
| 4508 |
-
},
|
| 4509 |
-
{
|
| 4510 |
-
"entropy": 0.3319382440298796,
|
| 4511 |
-
"epoch": 0.3272789257372299,
|
| 4512 |
-
"grad_norm": 0.5317144393920898,
|
| 4513 |
-
"learning_rate": 7.857153046664135e-05,
|
| 4514 |
-
"loss": 0.384430980682373,
|
| 4515 |
-
"mean_token_accuracy": 0.8938660219311714,
|
| 4516 |
-
"num_tokens": 5685401.0,
|
| 4517 |
-
"step": 4320
|
| 4518 |
-
},
|
| 4519 |
-
{
|
| 4520 |
-
"entropy": 0.3203262139111757,
|
| 4521 |
-
"epoch": 0.3280365158431031,
|
| 4522 |
-
"grad_norm": 0.44656360149383545,
|
| 4523 |
-
"learning_rate": 7.847076707249328e-05,
|
| 4524 |
-
"loss": 0.3503974437713623,
|
| 4525 |
-
"mean_token_accuracy": 0.9000782266259193,
|
| 4526 |
-
"num_tokens": 5789407.0,
|
| 4527 |
-
"step": 4330
|
| 4528 |
-
},
|
| 4529 |
-
{
|
| 4530 |
-
"entropy": 0.3162030352279544,
|
| 4531 |
-
"epoch": 0.3287941059489763,
|
| 4532 |
-
"grad_norm": 0.33549773693084717,
|
| 4533 |
-
"learning_rate": 7.836983227970789e-05,
|
| 4534 |
-
"loss": 0.35093986988067627,
|
| 4535 |
-
"mean_token_accuracy": 0.9014839589595794,
|
| 4536 |
-
"num_tokens": 5886985.0,
|
| 4537 |
-
"step": 4340
|
| 4538 |
-
},
|
| 4539 |
-
{
|
| 4540 |
-
"entropy": 0.29142586383968594,
|
| 4541 |
-
"epoch": 0.3295516960548495,
|
| 4542 |
-
"grad_norm": 0.539871871471405,
|
| 4543 |
-
"learning_rate": 7.826872669592901e-05,
|
| 4544 |
-
"loss": 0.3235773086547852,
|
| 4545 |
-
"mean_token_accuracy": 0.9067176148295403,
|
| 4546 |
-
"num_tokens": 5990347.0,
|
| 4547 |
-
"step": 4350
|
| 4548 |
-
},
|
| 4549 |
-
{
|
| 4550 |
-
"entropy": 0.29147457219660283,
|
| 4551 |
-
"epoch": 0.33030928616072275,
|
| 4552 |
-
"grad_norm": 0.8736031651496887,
|
| 4553 |
-
"learning_rate": 7.816745092982863e-05,
|
| 4554 |
-
"loss": 0.3602747917175293,
|
| 4555 |
-
"mean_token_accuracy": 0.9077178493142128,
|
| 4556 |
-
"num_tokens": 6082652.0,
|
| 4557 |
-
"step": 4360
|
| 4558 |
-
},
|
| 4559 |
-
{
|
| 4560 |
-
"entropy": 0.3061704048886895,
|
| 4561 |
-
"epoch": 0.331066876266596,
|
| 4562 |
-
"grad_norm": 0.456581175327301,
|
| 4563 |
-
"learning_rate": 7.806600559110333e-05,
|
| 4564 |
-
"loss": 0.34753856658935545,
|
| 4565 |
-
"mean_token_accuracy": 0.9054469063878059,
|
| 4566 |
-
"num_tokens": 6181691.0,
|
| 4567 |
-
"step": 4370
|
| 4568 |
-
},
|
| 4569 |
-
{
|
| 4570 |
-
"entropy": 0.30597354415804145,
|
| 4571 |
-
"epoch": 0.33182446637246915,
|
| 4572 |
-
"grad_norm": 0.48806032538414,
|
| 4573 |
-
"learning_rate": 7.796439129047048e-05,
|
| 4574 |
-
"loss": 0.33057436943054197,
|
| 4575 |
-
"mean_token_accuracy": 0.9024015262722969,
|
| 4576 |
-
"num_tokens": 6280742.0,
|
| 4577 |
-
"step": 4380
|
| 4578 |
-
},
|
| 4579 |
-
{
|
| 4580 |
-
"entropy": 0.32283892631530764,
|
| 4581 |
-
"epoch": 0.3325820564783424,
|
| 4582 |
-
"grad_norm": 0.3113273084163666,
|
| 4583 |
-
"learning_rate": 7.786260863966468e-05,
|
| 4584 |
-
"loss": 0.3692842960357666,
|
| 4585 |
-
"mean_token_accuracy": 0.8962025761604309,
|
| 4586 |
-
"num_tokens": 6389745.0,
|
| 4587 |
-
"step": 4390
|
| 4588 |
-
},
|
| 4589 |
-
{
|
| 4590 |
-
"entropy": 0.31780129205435514,
|
| 4591 |
-
"epoch": 0.3333396465842156,
|
| 4592 |
-
"grad_norm": 0.41104018688201904,
|
| 4593 |
-
"learning_rate": 7.776065825143399e-05,
|
| 4594 |
-
"loss": 0.3699489593505859,
|
| 4595 |
-
"mean_token_accuracy": 0.9006003260612487,
|
| 4596 |
-
"num_tokens": 6481252.0,
|
| 4597 |
-
"step": 4400
|
| 4598 |
-
},
|
| 4599 |
-
{
|
| 4600 |
-
"entropy": 0.3111474813893437,
|
| 4601 |
-
"epoch": 0.33409723669008884,
|
| 4602 |
-
"grad_norm": 0.4216645061969757,
|
| 4603 |
-
"learning_rate": 7.765854073953629e-05,
|
| 4604 |
-
"loss": 0.37152347564697263,
|
| 4605 |
-
"mean_token_accuracy": 0.901688639819622,
|
| 4606 |
-
"num_tokens": 6578986.0,
|
| 4607 |
-
"step": 4410
|
| 4608 |
-
},
|
| 4609 |
-
{
|
| 4610 |
-
"entropy": 0.3134287675842643,
|
| 4611 |
-
"epoch": 0.33485482679596207,
|
| 4612 |
-
"grad_norm": 0.44646093249320984,
|
| 4613 |
-
"learning_rate": 7.755625671873557e-05,
|
| 4614 |
-
"loss": 0.3566113471984863,
|
| 4615 |
-
"mean_token_accuracy": 0.8988679885864258,
|
| 4616 |
-
"num_tokens": 6675690.0,
|
| 4617 |
-
"step": 4420
|
| 4618 |
-
},
|
| 4619 |
-
{
|
| 4620 |
-
"entropy": 0.2818496130406857,
|
| 4621 |
-
"epoch": 0.33561241690183524,
|
| 4622 |
-
"grad_norm": 0.4837479591369629,
|
| 4623 |
-
"learning_rate": 7.745380680479824e-05,
|
| 4624 |
-
"loss": 0.3096198558807373,
|
| 4625 |
-
"mean_token_accuracy": 0.9128451719880104,
|
| 4626 |
-
"num_tokens": 6772435.0,
|
| 4627 |
-
"step": 4430
|
| 4628 |
-
},
|
| 4629 |
-
{
|
| 4630 |
-
"entropy": 0.2891602473333478,
|
| 4631 |
-
"epoch": 0.33637000700770847,
|
| 4632 |
-
"grad_norm": 0.46294134855270386,
|
| 4633 |
-
"learning_rate": 7.735119161448938e-05,
|
| 4634 |
-
"loss": 0.3227571725845337,
|
| 4635 |
-
"mean_token_accuracy": 0.9115611374378204,
|
| 4636 |
-
"num_tokens": 6870462.0,
|
| 4637 |
-
"step": 4440
|
| 4638 |
-
},
|
| 4639 |
-
{
|
| 4640 |
-
"entropy": 0.3321307893842459,
|
| 4641 |
-
"epoch": 0.3371275971135817,
|
| 4642 |
-
"grad_norm": 0.3993695378303528,
|
| 4643 |
-
"learning_rate": 7.724841176556911e-05,
|
| 4644 |
-
"loss": 0.37855651378631594,
|
| 4645 |
-
"mean_token_accuracy": 0.8910983562469482,
|
| 4646 |
-
"num_tokens": 6982005.0,
|
| 4647 |
-
"step": 4450
|
| 4648 |
-
},
|
| 4649 |
-
{
|
| 4650 |
-
"entropy": 0.3143839996308088,
|
| 4651 |
-
"epoch": 0.33788518721945493,
|
| 4652 |
-
"grad_norm": 0.44136250019073486,
|
| 4653 |
-
"learning_rate": 7.71454678767888e-05,
|
| 4654 |
-
"loss": 0.3535434722900391,
|
| 4655 |
-
"mean_token_accuracy": 0.8990781590342521,
|
| 4656 |
-
"num_tokens": 7074787.0,
|
| 4657 |
-
"step": 4460
|
| 4658 |
-
},
|
| 4659 |
-
{
|
| 4660 |
-
"entropy": 0.27286356603726747,
|
| 4661 |
-
"epoch": 0.3386427773253281,
|
| 4662 |
-
"grad_norm": 0.45516812801361084,
|
| 4663 |
-
"learning_rate": 7.704236056788737e-05,
|
| 4664 |
-
"loss": 0.3159470796585083,
|
| 4665 |
-
"mean_token_accuracy": 0.9138412743806839,
|
| 4666 |
-
"num_tokens": 7169295.0,
|
| 4667 |
-
"step": 4470
|
| 4668 |
-
},
|
| 4669 |
-
{
|
| 4670 |
-
"entropy": 0.31879308708012105,
|
| 4671 |
-
"epoch": 0.33940036743120133,
|
| 4672 |
-
"grad_norm": 0.562834620475769,
|
| 4673 |
-
"learning_rate": 7.693909045958751e-05,
|
| 4674 |
-
"loss": 0.375746488571167,
|
| 4675 |
-
"mean_token_accuracy": 0.8995465457439422,
|
| 4676 |
-
"num_tokens": 7263322.0,
|
| 4677 |
-
"step": 4480
|
| 4678 |
-
},
|
| 4679 |
-
{
|
| 4680 |
-
"entropy": 0.29084276044741275,
|
| 4681 |
-
"epoch": 0.34015795753707456,
|
| 4682 |
-
"grad_norm": 0.38388723134994507,
|
| 4683 |
-
"learning_rate": 7.683565817359206e-05,
|
| 4684 |
-
"loss": 0.35288500785827637,
|
| 4685 |
-
"mean_token_accuracy": 0.9082666411995888,
|
| 4686 |
-
"num_tokens": 7361968.0,
|
| 4687 |
-
"step": 4490
|
| 4688 |
-
},
|
| 4689 |
-
{
|
| 4690 |
-
"entropy": 0.29372019320726395,
|
| 4691 |
-
"epoch": 0.3409155476429478,
|
| 4692 |
-
"grad_norm": 0.652385950088501,
|
| 4693 |
-
"learning_rate": 7.673206433258015e-05,
|
| 4694 |
-
"loss": 0.3489070415496826,
|
| 4695 |
-
"mean_token_accuracy": 0.9063216969370842,
|
| 4696 |
-
"num_tokens": 7453884.0,
|
| 4697 |
-
"step": 4500
|
| 4698 |
-
},
|
| 4699 |
-
{
|
| 4700 |
-
"epoch": 0.3409155476429478,
|
| 4701 |
-
"eval_entropy": 0.30296776135371867,
|
| 4702 |
-
"eval_loss": 0.3005329668521881,
|
| 4703 |
-
"eval_mean_token_accuracy": 0.9036700241128157,
|
| 4704 |
-
"eval_num_tokens": 7453884.0,
|
| 4705 |
-
"eval_runtime": 357.6477,
|
| 4706 |
-
"eval_samples_per_second": 5.967,
|
| 4707 |
-
"eval_steps_per_second": 1.493,
|
| 4708 |
-
"step": 4500
|
| 4709 |
-
},
|
| 4710 |
-
{
|
| 4711 |
-
"entropy": 0.25267982203513384,
|
| 4712 |
-
"epoch": 0.341673137748821,
|
| 4713 |
-
"grad_norm": 0.6513199210166931,
|
| 4714 |
-
"learning_rate": 7.662830956020355e-05,
|
| 4715 |
-
"loss": 0.2851665258407593,
|
| 4716 |
-
"mean_token_accuracy": 0.9203153058886528,
|
| 4717 |
-
"num_tokens": 7546992.0,
|
| 4718 |
-
"step": 4510
|
| 4719 |
-
},
|
| 4720 |
-
{
|
| 4721 |
-
"entropy": 0.3188372856937349,
|
| 4722 |
-
"epoch": 0.3424307278546942,
|
| 4723 |
-
"grad_norm": 0.4398665428161621,
|
| 4724 |
-
"learning_rate": 7.652439448108276e-05,
|
| 4725 |
-
"loss": 0.3743882656097412,
|
| 4726 |
-
"mean_token_accuracy": 0.8984434530138969,
|
| 4727 |
-
"num_tokens": 7642665.0,
|
| 4728 |
-
"step": 4520
|
| 4729 |
-
},
|
| 4730 |
-
{
|
| 4731 |
-
"entropy": 0.3123008541762829,
|
| 4732 |
-
"epoch": 0.3431883179605674,
|
| 4733 |
-
"grad_norm": 0.3273858428001404,
|
| 4734 |
-
"learning_rate": 7.642031972080345e-05,
|
| 4735 |
-
"loss": 0.37012293338775637,
|
| 4736 |
-
"mean_token_accuracy": 0.9011475190520286,
|
| 4737 |
-
"num_tokens": 7742417.0,
|
| 4738 |
-
"step": 4530
|
| 4739 |
-
},
|
| 4740 |
-
{
|
| 4741 |
-
"entropy": 0.31543516404926775,
|
| 4742 |
-
"epoch": 0.34394590806644065,
|
| 4743 |
-
"grad_norm": 0.3998027443885803,
|
| 4744 |
-
"learning_rate": 7.63160859059126e-05,
|
| 4745 |
-
"loss": 0.37990047931671145,
|
| 4746 |
-
"mean_token_accuracy": 0.8965097144246101,
|
| 4747 |
-
"num_tokens": 7853372.0,
|
| 4748 |
-
"step": 4540
|
| 4749 |
-
},
|
| 4750 |
-
{
|
| 4751 |
-
"entropy": 0.3232473568990827,
|
| 4752 |
-
"epoch": 0.3447034981723139,
|
| 4753 |
-
"grad_norm": 0.5831270813941956,
|
| 4754 |
-
"learning_rate": 7.62116936639146e-05,
|
| 4755 |
-
"loss": 0.3985744476318359,
|
| 4756 |
-
"mean_token_accuracy": 0.8975970014929772,
|
| 4757 |
-
"num_tokens": 7965638.0,
|
| 4758 |
-
"step": 4550
|
| 4759 |
-
},
|
| 4760 |
-
{
|
| 4761 |
-
"entropy": 0.3064836641773582,
|
| 4762 |
-
"epoch": 0.3454610882781871,
|
| 4763 |
-
"grad_norm": 0.32403406500816345,
|
| 4764 |
-
"learning_rate": 7.610714362326777e-05,
|
| 4765 |
-
"loss": 0.34210906028747556,
|
| 4766 |
-
"mean_token_accuracy": 0.9028223529458046,
|
| 4767 |
-
"num_tokens": 8061769.0,
|
| 4768 |
-
"step": 4560
|
| 4769 |
-
},
|
| 4770 |
-
{
|
| 4771 |
-
"entropy": 0.2993687367066741,
|
| 4772 |
-
"epoch": 0.3462186783840603,
|
| 4773 |
-
"grad_norm": 0.3851373791694641,
|
| 4774 |
-
"learning_rate": 7.600243641338027e-05,
|
| 4775 |
-
"loss": 0.3269450902938843,
|
| 4776 |
-
"mean_token_accuracy": 0.9052052691578865,
|
| 4777 |
-
"num_tokens": 8158536.0,
|
| 4778 |
-
"step": 4570
|
| 4779 |
-
},
|
| 4780 |
-
{
|
| 4781 |
-
"entropy": 0.3500557569786906,
|
| 4782 |
-
"epoch": 0.3469762684899335,
|
| 4783 |
-
"grad_norm": 0.4885443449020386,
|
| 4784 |
-
"learning_rate": 7.589757266460653e-05,
|
| 4785 |
-
"loss": 0.41555023193359375,
|
| 4786 |
-
"mean_token_accuracy": 0.8889480486512185,
|
| 4787 |
-
"num_tokens": 8264141.0,
|
| 4788 |
-
"step": 4580
|
| 4789 |
-
},
|
| 4790 |
-
{
|
| 4791 |
-
"entropy": 0.32052452899515627,
|
| 4792 |
-
"epoch": 0.34773385859580674,
|
| 4793 |
-
"grad_norm": 0.5277999043464661,
|
| 4794 |
-
"learning_rate": 7.57925530082433e-05,
|
| 4795 |
-
"loss": 0.36511707305908203,
|
| 4796 |
-
"mean_token_accuracy": 0.8961958944797516,
|
| 4797 |
-
"num_tokens": 8373169.0,
|
| 4798 |
-
"step": 4590
|
| 4799 |
-
},
|
| 4800 |
-
{
|
| 4801 |
-
"entropy": 0.28373611588031056,
|
| 4802 |
-
"epoch": 0.34849144870168,
|
| 4803 |
-
"grad_norm": 0.49028876423835754,
|
| 4804 |
-
"learning_rate": 7.568737807652596e-05,
|
| 4805 |
-
"loss": 0.30989961624145507,
|
| 4806 |
-
"mean_token_accuracy": 0.9118182867765426,
|
| 4807 |
-
"num_tokens": 8466566.0,
|
| 4808 |
-
"step": 4600
|
| 4809 |
-
},
|
| 4810 |
-
{
|
| 4811 |
-
"entropy": 0.30789675302803515,
|
| 4812 |
-
"epoch": 0.34924903880755315,
|
| 4813 |
-
"grad_norm": 0.48537036776542664,
|
| 4814 |
-
"learning_rate": 7.558204850262467e-05,
|
| 4815 |
-
"loss": 0.36036148071289065,
|
| 4816 |
-
"mean_token_accuracy": 0.9010463237762452,
|
| 4817 |
-
"num_tokens": 8568165.0,
|
| 4818 |
-
"step": 4610
|
| 4819 |
-
},
|
| 4820 |
-
{
|
| 4821 |
-
"entropy": 0.26539887543767693,
|
| 4822 |
-
"epoch": 0.3500066289134264,
|
| 4823 |
-
"grad_norm": 0.435004860162735,
|
| 4824 |
-
"learning_rate": 7.547656492064055e-05,
|
| 4825 |
-
"loss": 0.3213709115982056,
|
| 4826 |
-
"mean_token_accuracy": 0.915293487906456,
|
| 4827 |
-
"num_tokens": 8667162.0,
|
| 4828 |
-
"step": 4620
|
| 4829 |
-
},
|
| 4830 |
-
{
|
| 4831 |
-
"entropy": 0.2999891651794314,
|
| 4832 |
-
"epoch": 0.3507642190192996,
|
| 4833 |
-
"grad_norm": 0.4458155035972595,
|
| 4834 |
-
"learning_rate": 7.537092796560188e-05,
|
| 4835 |
-
"loss": 0.352059006690979,
|
| 4836 |
-
"mean_token_accuracy": 0.9006678983569145,
|
| 4837 |
-
"num_tokens": 8767757.0,
|
| 4838 |
-
"step": 4630
|
| 4839 |
-
},
|
| 4840 |
-
{
|
| 4841 |
-
"entropy": 0.3175335772335529,
|
| 4842 |
-
"epoch": 0.35152180912517283,
|
| 4843 |
-
"grad_norm": 0.4449421167373657,
|
| 4844 |
-
"learning_rate": 7.526513827346027e-05,
|
| 4845 |
-
"loss": 0.34688520431518555,
|
| 4846 |
-
"mean_token_accuracy": 0.9023878127336502,
|
| 4847 |
-
"num_tokens": 8870892.0,
|
| 4848 |
-
"step": 4640
|
| 4849 |
-
},
|
| 4850 |
-
{
|
| 4851 |
-
"entropy": 0.2684692405164242,
|
| 4852 |
-
"epoch": 0.35227939923104606,
|
| 4853 |
-
"grad_norm": 0.48179900646209717,
|
| 4854 |
-
"learning_rate": 7.515919648108683e-05,
|
| 4855 |
-
"loss": 0.2930025100708008,
|
| 4856 |
-
"mean_token_accuracy": 0.9159289449453354,
|
| 4857 |
-
"num_tokens": 8961329.0,
|
| 4858 |
-
"step": 4650
|
| 4859 |
-
},
|
| 4860 |
-
{
|
| 4861 |
-
"entropy": 0.3012943722307682,
|
| 4862 |
-
"epoch": 0.35303698933691924,
|
| 4863 |
-
"grad_norm": 0.5411931872367859,
|
| 4864 |
-
"learning_rate": 7.505310322626835e-05,
|
| 4865 |
-
"loss": 0.32681946754455565,
|
| 4866 |
-
"mean_token_accuracy": 0.9047896936535835,
|
| 4867 |
-
"num_tokens": 9053322.0,
|
| 4868 |
-
"step": 4660
|
| 4869 |
-
},
|
| 4870 |
-
{
|
| 4871 |
-
"entropy": 0.2929389262571931,
|
| 4872 |
-
"epoch": 0.35379457944279247,
|
| 4873 |
-
"grad_norm": 0.6351791620254517,
|
| 4874 |
-
"learning_rate": 7.494685914770342e-05,
|
| 4875 |
-
"loss": 0.35527982711791994,
|
| 4876 |
-
"mean_token_accuracy": 0.9080116629600525,
|
| 4877 |
-
"num_tokens": 9150933.0,
|
| 4878 |
-
"step": 4670
|
| 4879 |
-
},
|
| 4880 |
-
{
|
| 4881 |
-
"entropy": 0.2740900555625558,
|
| 4882 |
-
"epoch": 0.3545521695486657,
|
| 4883 |
-
"grad_norm": 0.40528643131256104,
|
| 4884 |
-
"learning_rate": 7.484046488499864e-05,
|
| 4885 |
-
"loss": 0.326665472984314,
|
| 4886 |
-
"mean_token_accuracy": 0.9146007567644119,
|
| 4887 |
-
"num_tokens": 9253832.0,
|
| 4888 |
-
"step": 4680
|
| 4889 |
-
},
|
| 4890 |
-
{
|
| 4891 |
-
"entropy": 0.2925885286182165,
|
| 4892 |
-
"epoch": 0.3553097596545389,
|
| 4893 |
-
"grad_norm": 0.6988641619682312,
|
| 4894 |
-
"learning_rate": 7.473392107866473e-05,
|
| 4895 |
-
"loss": 0.32947583198547364,
|
| 4896 |
-
"mean_token_accuracy": 0.9095833480358124,
|
| 4897 |
-
"num_tokens": 9342476.0,
|
| 4898 |
-
"step": 4690
|
| 4899 |
-
},
|
| 4900 |
-
{
|
| 4901 |
-
"entropy": 0.328192562609911,
|
| 4902 |
-
"epoch": 0.35606734976041216,
|
| 4903 |
-
"grad_norm": 0.3606797158718109,
|
| 4904 |
-
"learning_rate": 7.462722837011273e-05,
|
| 4905 |
-
"loss": 0.37802944183349607,
|
| 4906 |
-
"mean_token_accuracy": 0.8960362210869789,
|
| 4907 |
-
"num_tokens": 9448219.0,
|
| 4908 |
-
"step": 4700
|
| 4909 |
-
},
|
| 4910 |
-
{
|
| 4911 |
-
"entropy": 0.29102398827672005,
|
| 4912 |
-
"epoch": 0.35682493986628533,
|
| 4913 |
-
"grad_norm": 0.37759271264076233,
|
| 4914 |
-
"learning_rate": 7.452038740165e-05,
|
| 4915 |
-
"loss": 0.32128334045410156,
|
| 4916 |
-
"mean_token_accuracy": 0.9089973151683808,
|
| 4917 |
-
"num_tokens": 9540752.0,
|
| 4918 |
-
"step": 4710
|
| 4919 |
-
},
|
| 4920 |
-
{
|
| 4921 |
-
"entropy": 0.31482851076871154,
|
| 4922 |
-
"epoch": 0.35758252997215856,
|
| 4923 |
-
"grad_norm": 0.542698085308075,
|
| 4924 |
-
"learning_rate": 7.441339881647653e-05,
|
| 4925 |
-
"loss": 0.35300025939941404,
|
| 4926 |
-
"mean_token_accuracy": 0.9003022000193596,
|
| 4927 |
-
"num_tokens": 9641938.0,
|
| 4928 |
-
"step": 4720
|
| 4929 |
-
},
|
| 4930 |
-
{
|
| 4931 |
-
"entropy": 0.2924330163747072,
|
| 4932 |
-
"epoch": 0.3583401200780318,
|
| 4933 |
-
"grad_norm": 0.3568849265575409,
|
| 4934 |
-
"learning_rate": 7.430626325868097e-05,
|
| 4935 |
-
"loss": 0.34119129180908203,
|
| 4936 |
-
"mean_token_accuracy": 0.906133034825325,
|
| 4937 |
-
"num_tokens": 9743671.0,
|
| 4938 |
-
"step": 4730
|
| 4939 |
-
},
|
| 4940 |
-
{
|
| 4941 |
-
"entropy": 0.3062052955850959,
|
| 4942 |
-
"epoch": 0.359097710183905,
|
| 4943 |
-
"grad_norm": 0.502115786075592,
|
| 4944 |
-
"learning_rate": 7.419898137323675e-05,
|
| 4945 |
-
"loss": 0.34509029388427737,
|
| 4946 |
-
"mean_token_accuracy": 0.9022952586412429,
|
| 4947 |
-
"num_tokens": 9844345.0,
|
| 4948 |
-
"step": 4740
|
| 4949 |
-
},
|
| 4950 |
-
{
|
| 4951 |
-
"entropy": 0.3179619304835796,
|
| 4952 |
-
"epoch": 0.3598553002897782,
|
| 4953 |
-
"grad_norm": 0.5590654015541077,
|
| 4954 |
-
"learning_rate": 7.409155380599823e-05,
|
| 4955 |
-
"loss": 0.3489116668701172,
|
| 4956 |
-
"mean_token_accuracy": 0.8987106159329414,
|
| 4957 |
-
"num_tokens": 9947842.0,
|
| 4958 |
-
"step": 4750
|
| 4959 |
-
},
|
| 4960 |
-
{
|
| 4961 |
-
"epoch": 0.3598553002897782,
|
| 4962 |
-
"eval_entropy": 0.30675893575511176,
|
| 4963 |
-
"eval_loss": 0.30102235078811646,
|
| 4964 |
-
"eval_mean_token_accuracy": 0.9033761302406869,
|
| 4965 |
-
"eval_num_tokens": 9947842.0,
|
| 4966 |
-
"eval_runtime": 357.4102,
|
| 4967 |
-
"eval_samples_per_second": 5.971,
|
| 4968 |
-
"eval_steps_per_second": 1.494,
|
| 4969 |
-
"step": 4750
|
| 4970 |
-
},
|
| 4971 |
-
{
|
| 4972 |
-
"entropy": 0.27557093296200036,
|
| 4973 |
-
"epoch": 0.3606128903956514,
|
| 4974 |
-
"grad_norm": 0.35161638259887695,
|
| 4975 |
-
"learning_rate": 7.398398120369681e-05,
|
| 4976 |
-
"loss": 0.3087491512298584,
|
| 4977 |
-
"mean_token_accuracy": 0.9143075257539749,
|
| 4978 |
-
"num_tokens": 10047101.0,
|
| 4979 |
-
"step": 4760
|
| 4980 |
-
},
|
| 4981 |
-
{
|
| 4982 |
-
"entropy": 0.2880682935938239,
|
| 4983 |
-
"epoch": 0.36137048050152465,
|
| 4984 |
-
"grad_norm": 0.36225664615631104,
|
| 4985 |
-
"learning_rate": 7.387626421393698e-05,
|
| 4986 |
-
"loss": 0.313517165184021,
|
| 4987 |
-
"mean_token_accuracy": 0.9082889899611473,
|
| 4988 |
-
"num_tokens": 10147418.0,
|
| 4989 |
-
"step": 4770
|
| 4990 |
-
},
|
| 4991 |
-
{
|
| 4992 |
-
"entropy": 0.2912915203720331,
|
| 4993 |
-
"epoch": 0.3621280706073979,
|
| 4994 |
-
"grad_norm": 0.37559425830841064,
|
| 4995 |
-
"learning_rate": 7.376840348519253e-05,
|
| 4996 |
-
"loss": 0.3523188591003418,
|
| 4997 |
-
"mean_token_accuracy": 0.9088112473487854,
|
| 4998 |
-
"num_tokens": 10243583.0,
|
| 4999 |
-
"step": 4780
|
| 5000 |
-
},
|
| 5001 |
-
{
|
| 5002 |
-
"entropy": 0.29392408989369867,
|
| 5003 |
-
"epoch": 0.3628856607132711,
|
| 5004 |
-
"grad_norm": 0.34611624479293823,
|
| 5005 |
-
"learning_rate": 7.366039966680251e-05,
|
| 5006 |
-
"loss": 0.3264188289642334,
|
| 5007 |
-
"mean_token_accuracy": 0.9094091519713402,
|
| 5008 |
-
"num_tokens": 10341956.0,
|
| 5009 |
-
"step": 4790
|
| 5010 |
-
},
|
| 5011 |
-
{
|
| 5012 |
-
"entropy": 0.27624342907220123,
|
| 5013 |
-
"epoch": 0.3636432508191443,
|
| 5014 |
-
"grad_norm": 0.49688178300857544,
|
| 5015 |
-
"learning_rate": 7.355225340896748e-05,
|
| 5016 |
-
"loss": 0.32373270988464353,
|
| 5017 |
-
"mean_token_accuracy": 0.9136532321572304,
|
| 5018 |
-
"num_tokens": 10433217.0,
|
| 5019 |
-
"step": 4800
|
| 5020 |
-
},
|
| 5021 |
-
{
|
| 5022 |
-
"entropy": 0.32521933866664765,
|
| 5023 |
-
"epoch": 0.3644008409250175,
|
| 5024 |
-
"grad_norm": 0.32955390214920044,
|
| 5025 |
-
"learning_rate": 7.344396536274543e-05,
|
| 5026 |
-
"loss": 0.38016834259033205,
|
| 5027 |
-
"mean_token_accuracy": 0.895936033129692,
|
| 5028 |
-
"num_tokens": 10540053.0,
|
| 5029 |
-
"step": 4810
|
| 5030 |
-
},
|
| 5031 |
-
{
|
| 5032 |
-
"entropy": 0.33542325869202616,
|
| 5033 |
-
"epoch": 0.36515843103089074,
|
| 5034 |
-
"grad_norm": 0.48000800609588623,
|
| 5035 |
-
"learning_rate": 7.333553618004795e-05,
|
| 5036 |
-
"loss": 0.3799370527267456,
|
| 5037 |
-
"mean_token_accuracy": 0.8949392944574356,
|
| 5038 |
-
"num_tokens": 10640441.0,
|
| 5039 |
-
"step": 4820
|
| 5040 |
-
},
|
| 5041 |
-
{
|
| 5042 |
-
"entropy": 0.2541424440219998,
|
| 5043 |
-
"epoch": 0.36591602113676397,
|
| 5044 |
-
"grad_norm": 0.48736461997032166,
|
| 5045 |
-
"learning_rate": 7.322696651363634e-05,
|
| 5046 |
-
"loss": 0.30034053325653076,
|
| 5047 |
-
"mean_token_accuracy": 0.9218615472316742,
|
| 5048 |
-
"num_tokens": 10725720.0,
|
| 5049 |
-
"step": 4830
|
| 5050 |
-
},
|
| 5051 |
-
{
|
| 5052 |
-
"entropy": 0.28171106036752463,
|
| 5053 |
-
"epoch": 0.3666736112426372,
|
| 5054 |
-
"grad_norm": 0.5695344805717468,
|
| 5055 |
-
"learning_rate": 7.31182570171176e-05,
|
| 5056 |
-
"loss": 0.3393592834472656,
|
| 5057 |
-
"mean_token_accuracy": 0.9125670120120049,
|
| 5058 |
-
"num_tokens": 10831987.0,
|
| 5059 |
-
"step": 4840
|
| 5060 |
-
},
|
| 5061 |
-
{
|
| 5062 |
-
"entropy": 0.26796618979424236,
|
| 5063 |
-
"epoch": 0.3674312013485104,
|
| 5064 |
-
"grad_norm": 0.46774476766586304,
|
| 5065 |
-
"learning_rate": 7.300940834494055e-05,
|
| 5066 |
-
"loss": 0.31605896949768064,
|
| 5067 |
-
"mean_token_accuracy": 0.9121967896819114,
|
| 5068 |
-
"num_tokens": 10930561.0,
|
| 5069 |
-
"step": 4850
|
| 5070 |
-
},
|
| 5071 |
-
{
|
| 5072 |
-
"entropy": 0.3210433704778552,
|
| 5073 |
-
"epoch": 0.3681887914543836,
|
| 5074 |
-
"grad_norm": 0.3978744447231293,
|
| 5075 |
-
"learning_rate": 7.290042115239185e-05,
|
| 5076 |
-
"loss": 0.38169336318969727,
|
| 5077 |
-
"mean_token_accuracy": 0.8966830179095269,
|
| 5078 |
-
"num_tokens": 11039180.0,
|
| 5079 |
-
"step": 4860
|
| 5080 |
-
},
|
| 5081 |
-
{
|
| 5082 |
-
"entropy": 0.27403727918863297,
|
| 5083 |
-
"epoch": 0.36894638156025683,
|
| 5084 |
-
"grad_norm": 0.51429682970047,
|
| 5085 |
-
"learning_rate": 7.279129609559207e-05,
|
| 5086 |
-
"loss": 0.31013762950897217,
|
| 5087 |
-
"mean_token_accuracy": 0.9130677282810211,
|
| 5088 |
-
"num_tokens": 11135489.0,
|
| 5089 |
-
"step": 4870
|
| 5090 |
-
},
|
| 5091 |
-
{
|
| 5092 |
-
"entropy": 0.31362601201981305,
|
| 5093 |
-
"epoch": 0.36970397166613006,
|
| 5094 |
-
"grad_norm": 0.42733001708984375,
|
| 5095 |
-
"learning_rate": 7.268203383149179e-05,
|
| 5096 |
-
"loss": 0.3729254722595215,
|
| 5097 |
-
"mean_token_accuracy": 0.9008683413267136,
|
| 5098 |
-
"num_tokens": 11232364.0,
|
| 5099 |
-
"step": 4880
|
| 5100 |
-
},
|
| 5101 |
-
{
|
| 5102 |
-
"entropy": 0.2981378760188818,
|
| 5103 |
-
"epoch": 0.37046156177200323,
|
| 5104 |
-
"grad_norm": 0.5828710198402405,
|
| 5105 |
-
"learning_rate": 7.257263501786752e-05,
|
| 5106 |
-
"loss": 0.3386087894439697,
|
| 5107 |
-
"mean_token_accuracy": 0.9076459318399429,
|
| 5108 |
-
"num_tokens": 11324730.0,
|
| 5109 |
-
"step": 4890
|
| 5110 |
-
},
|
| 5111 |
-
{
|
| 5112 |
-
"entropy": 0.3232640836387873,
|
| 5113 |
-
"epoch": 0.37121915187787646,
|
| 5114 |
-
"grad_norm": 0.36474403738975525,
|
| 5115 |
-
"learning_rate": 7.24631003133179e-05,
|
| 5116 |
-
"loss": 0.3743882179260254,
|
| 5117 |
-
"mean_token_accuracy": 0.8979454353451729,
|
| 5118 |
-
"num_tokens": 11417447.0,
|
| 5119 |
-
"step": 4900
|
| 5120 |
-
},
|
| 5121 |
-
{
|
| 5122 |
-
"entropy": 0.3150969460606575,
|
| 5123 |
-
"epoch": 0.3719767419837497,
|
| 5124 |
-
"grad_norm": 0.530674934387207,
|
| 5125 |
-
"learning_rate": 7.235343037725962e-05,
|
| 5126 |
-
"loss": 0.3686549663543701,
|
| 5127 |
-
"mean_token_accuracy": 0.9020062759518623,
|
| 5128 |
-
"num_tokens": 11517636.0,
|
| 5129 |
-
"step": 4910
|
| 5130 |
-
},
|
| 5131 |
-
{
|
| 5132 |
-
"entropy": 0.3168086024001241,
|
| 5133 |
-
"epoch": 0.3727343320896229,
|
| 5134 |
-
"grad_norm": 0.5276638269424438,
|
| 5135 |
-
"learning_rate": 7.224362586992351e-05,
|
| 5136 |
-
"loss": 0.35547144412994386,
|
| 5137 |
-
"mean_token_accuracy": 0.8995771422982216,
|
| 5138 |
-
"num_tokens": 11615276.0,
|
| 5139 |
-
"step": 4920
|
| 5140 |
-
},
|
| 5141 |
-
{
|
| 5142 |
-
"entropy": 0.2907005790621042,
|
| 5143 |
-
"epoch": 0.37349192219549615,
|
| 5144 |
-
"grad_norm": 0.5746142268180847,
|
| 5145 |
-
"learning_rate": 7.213368745235048e-05,
|
| 5146 |
-
"loss": 0.32363438606262207,
|
| 5147 |
-
"mean_token_accuracy": 0.9064985603094101,
|
| 5148 |
-
"num_tokens": 11710465.0,
|
| 5149 |
-
"step": 4930
|
| 5150 |
-
},
|
| 5151 |
-
{
|
| 5152 |
-
"entropy": 0.30047825146466495,
|
| 5153 |
-
"epoch": 0.3742495123013693,
|
| 5154 |
-
"grad_norm": 0.40970680117607117,
|
| 5155 |
-
"learning_rate": 7.202361578638766e-05,
|
| 5156 |
-
"loss": 0.3484736680984497,
|
| 5157 |
-
"mean_token_accuracy": 0.9060021191835403,
|
| 5158 |
-
"num_tokens": 11808573.0,
|
| 5159 |
-
"step": 4940
|
| 5160 |
-
},
|
| 5161 |
-
{
|
| 5162 |
-
"entropy": 0.3003014411777258,
|
| 5163 |
-
"epoch": 0.37500710240724255,
|
| 5164 |
-
"grad_norm": 0.4596264958381653,
|
| 5165 |
-
"learning_rate": 7.191341153468434e-05,
|
| 5166 |
-
"loss": 0.3368295907974243,
|
| 5167 |
-
"mean_token_accuracy": 0.9048133313655853,
|
| 5168 |
-
"num_tokens": 11899946.0,
|
| 5169 |
-
"step": 4950
|
| 5170 |
-
},
|
| 5171 |
-
{
|
| 5172 |
-
"entropy": 0.2725072126835585,
|
| 5173 |
-
"epoch": 0.3757646925131158,
|
| 5174 |
-
"grad_norm": 0.5989105105400085,
|
| 5175 |
-
"learning_rate": 7.180307536068802e-05,
|
| 5176 |
-
"loss": 0.333084774017334,
|
| 5177 |
-
"mean_token_accuracy": 0.9113850623369217,
|
| 5178 |
-
"num_tokens": 11997869.0,
|
| 5179 |
-
"step": 4960
|
| 5180 |
-
},
|
| 5181 |
-
{
|
| 5182 |
-
"entropy": 0.29812303017824887,
|
| 5183 |
-
"epoch": 0.376522282618989,
|
| 5184 |
-
"grad_norm": 0.4399600327014923,
|
| 5185 |
-
"learning_rate": 7.169260792864032e-05,
|
| 5186 |
-
"loss": 0.34678874015808103,
|
| 5187 |
-
"mean_token_accuracy": 0.9052843302488327,
|
| 5188 |
-
"num_tokens": 12102835.0,
|
| 5189 |
-
"step": 4970
|
| 5190 |
-
},
|
| 5191 |
-
{
|
| 5192 |
-
"entropy": 0.26962209567427636,
|
| 5193 |
-
"epoch": 0.3772798727248622,
|
| 5194 |
-
"grad_norm": 0.4352383017539978,
|
| 5195 |
-
"learning_rate": 7.158200990357315e-05,
|
| 5196 |
-
"loss": 0.30444366931915284,
|
| 5197 |
-
"mean_token_accuracy": 0.9157704383134841,
|
| 5198 |
-
"num_tokens": 12191798.0,
|
| 5199 |
-
"step": 4980
|
| 5200 |
-
},
|
| 5201 |
-
{
|
| 5202 |
-
"entropy": 0.3402407165616751,
|
| 5203 |
-
"epoch": 0.3780374628307354,
|
| 5204 |
-
"grad_norm": 0.39899390935897827,
|
| 5205 |
-
"learning_rate": 7.147128195130454e-05,
|
| 5206 |
-
"loss": 0.38022818565368655,
|
| 5207 |
-
"mean_token_accuracy": 0.8923196330666542,
|
| 5208 |
-
"num_tokens": 12296405.0,
|
| 5209 |
-
"step": 4990
|
| 5210 |
-
},
|
| 5211 |
-
{
|
| 5212 |
-
"entropy": 0.276414324529469,
|
| 5213 |
-
"epoch": 0.37879505293660864,
|
| 5214 |
-
"grad_norm": 0.46220678091049194,
|
| 5215 |
-
"learning_rate": 7.136042473843474e-05,
|
| 5216 |
-
"loss": 0.32870604991912844,
|
| 5217 |
-
"mean_token_accuracy": 0.9129181683063508,
|
| 5218 |
-
"num_tokens": 12382701.0,
|
| 5219 |
-
"step": 5000
|
| 5220 |
-
},
|
| 5221 |
-
{
|
| 5222 |
-
"epoch": 0.37879505293660864,
|
| 5223 |
-
"eval_entropy": 0.3011974955068322,
|
| 5224 |
-
"eval_loss": 0.3008781373500824,
|
| 5225 |
-
"eval_mean_token_accuracy": 0.9031777002391744,
|
| 5226 |
-
"eval_num_tokens": 12382701.0,
|
| 5227 |
-
"eval_runtime": 357.5727,
|
| 5228 |
-
"eval_samples_per_second": 5.968,
|
| 5229 |
-
"eval_steps_per_second": 1.493,
|
| 5230 |
-
"step": 5000
|
| 5231 |
-
},
|
| 5232 |
-
{
|
| 5233 |
-
"entropy": 0.27525581177324054,
|
| 5234 |
-
"epoch": 0.3795526430424819,
|
| 5235 |
-
"grad_norm": 0.4783889353275299,
|
| 5236 |
-
"learning_rate": 7.124943893234215e-05,
|
| 5237 |
-
"loss": 0.32194807529449465,
|
| 5238 |
-
"mean_token_accuracy": 0.9114383578300476,
|
| 5239 |
-
"num_tokens": 12474804.0,
|
| 5240 |
-
"step": 5010
|
| 5241 |
-
},
|
| 5242 |
-
{
|
| 5243 |
-
"entropy": 0.301926470734179,
|
| 5244 |
-
"epoch": 0.3803102331483551,
|
| 5245 |
-
"grad_norm": 0.35501521825790405,
|
| 5246 |
-
"learning_rate": 7.113832520117935e-05,
|
| 5247 |
-
"loss": 0.35866732597351075,
|
| 5248 |
-
"mean_token_accuracy": 0.9026682913303375,
|
| 5249 |
-
"num_tokens": 12574525.0,
|
| 5250 |
-
"step": 5020
|
| 5251 |
-
},
|
| 5252 |
-
{
|
| 5253 |
-
"entropy": 0.3118064312264323,
|
| 5254 |
-
"epoch": 0.3810678232542283,
|
| 5255 |
-
"grad_norm": 0.49730929732322693,
|
| 5256 |
-
"learning_rate": 7.102708421386902e-05,
|
| 5257 |
-
"loss": 0.3602226734161377,
|
| 5258 |
-
"mean_token_accuracy": 0.8994961470365525,
|
| 5259 |
-
"num_tokens": 12667636.0,
|
| 5260 |
-
"step": 5030
|
| 5261 |
-
},
|
| 5262 |
-
{
|
| 5263 |
-
"entropy": 0.3229465847834945,
|
| 5264 |
-
"epoch": 0.3818254133601015,
|
| 5265 |
-
"grad_norm": 0.45787057280540466,
|
| 5266 |
-
"learning_rate": 7.091571664009993e-05,
|
| 5267 |
-
"loss": 0.35032095909118655,
|
| 5268 |
-
"mean_token_accuracy": 0.9014500126242637,
|
| 5269 |
-
"num_tokens": 12768678.0,
|
| 5270 |
-
"step": 5040
|
| 5271 |
-
},
|
| 5272 |
-
{
|
| 5273 |
-
"entropy": 0.2986622307449579,
|
| 5274 |
-
"epoch": 0.38258300346597474,
|
| 5275 |
-
"grad_norm": 0.3898067772388458,
|
| 5276 |
-
"learning_rate": 7.080422315032298e-05,
|
| 5277 |
-
"loss": 0.33089771270751955,
|
| 5278 |
-
"mean_token_accuracy": 0.9028651803731919,
|
| 5279 |
-
"num_tokens": 12870507.0,
|
| 5280 |
-
"step": 5050
|
| 5281 |
-
},
|
| 5282 |
-
{
|
| 5283 |
-
"entropy": 0.3021549839526415,
|
| 5284 |
-
"epoch": 0.38334059357184797,
|
| 5285 |
-
"grad_norm": 0.42782846093177795,
|
| 5286 |
-
"learning_rate": 7.069260441574706e-05,
|
| 5287 |
-
"loss": 0.3436141014099121,
|
| 5288 |
-
"mean_token_accuracy": 0.9041045948863029,
|
| 5289 |
-
"num_tokens": 12975024.0,
|
| 5290 |
-
"step": 5060
|
| 5291 |
-
},
|
| 5292 |
-
{
|
| 5293 |
-
"entropy": 0.2647577865049243,
|
| 5294 |
-
"epoch": 0.3840981836777212,
|
| 5295 |
-
"grad_norm": 0.45617273449897766,
|
| 5296 |
-
"learning_rate": 7.058086110833503e-05,
|
| 5297 |
-
"loss": 0.3188798904418945,
|
| 5298 |
-
"mean_token_accuracy": 0.9151462346315384,
|
| 5299 |
-
"num_tokens": 13068056.0,
|
| 5300 |
-
"step": 5070
|
| 5301 |
-
},
|
| 5302 |
-
{
|
| 5303 |
-
"entropy": 0.2972287181764841,
|
| 5304 |
-
"epoch": 0.38485577378359437,
|
| 5305 |
-
"grad_norm": 0.4732530117034912,
|
| 5306 |
-
"learning_rate": 7.046899390079979e-05,
|
| 5307 |
-
"loss": 0.3440873384475708,
|
| 5308 |
-
"mean_token_accuracy": 0.9067432582378387,
|
| 5309 |
-
"num_tokens": 13163903.0,
|
| 5310 |
-
"step": 5080
|
| 5311 |
-
},
|
| 5312 |
-
{
|
| 5313 |
-
"entropy": 0.34867528155446054,
|
| 5314 |
-
"epoch": 0.3856133638894676,
|
| 5315 |
-
"grad_norm": 0.3450987637042999,
|
| 5316 |
-
"learning_rate": 7.035700346660003e-05,
|
| 5317 |
-
"loss": 0.4140889644622803,
|
| 5318 |
-
"mean_token_accuracy": 0.8880186006426811,
|
| 5319 |
-
"num_tokens": 13272073.0,
|
| 5320 |
-
"step": 5090
|
| 5321 |
-
},
|
| 5322 |
-
{
|
| 5323 |
-
"entropy": 0.32138296142220496,
|
| 5324 |
-
"epoch": 0.3863709539953408,
|
| 5325 |
-
"grad_norm": 0.41912323236465454,
|
| 5326 |
-
"learning_rate": 7.024489047993636e-05,
|
| 5327 |
-
"loss": 0.37928009033203125,
|
| 5328 |
-
"mean_token_accuracy": 0.8977195471525192,
|
| 5329 |
-
"num_tokens": 13367042.0,
|
| 5330 |
-
"step": 5100
|
| 5331 |
-
},
|
| 5332 |
-
{
|
| 5333 |
-
"entropy": 0.29931807778775693,
|
| 5334 |
-
"epoch": 0.38712854410121406,
|
| 5335 |
-
"grad_norm": 0.44405844807624817,
|
| 5336 |
-
"learning_rate": 7.013265561574716e-05,
|
| 5337 |
-
"loss": 0.3626858711242676,
|
| 5338 |
-
"mean_token_accuracy": 0.9048500895500183,
|
| 5339 |
-
"num_tokens": 13465139.0,
|
| 5340 |
-
"step": 5110
|
| 5341 |
-
},
|
| 5342 |
-
{
|
| 5343 |
-
"entropy": 0.3437439644709229,
|
| 5344 |
-
"epoch": 0.38788613420708723,
|
| 5345 |
-
"grad_norm": 0.5075298547744751,
|
| 5346 |
-
"learning_rate": 7.002029954970448e-05,
|
| 5347 |
-
"loss": 0.3893740177154541,
|
| 5348 |
-
"mean_token_accuracy": 0.8930304124951363,
|
| 5349 |
-
"num_tokens": 13566049.0,
|
| 5350 |
-
"step": 5120
|
| 5351 |
-
},
|
| 5352 |
-
{
|
| 5353 |
-
"entropy": 0.281341490522027,
|
| 5354 |
-
"epoch": 0.38864372431296046,
|
| 5355 |
-
"grad_norm": 0.4543285369873047,
|
| 5356 |
-
"learning_rate": 6.99078229582101e-05,
|
| 5357 |
-
"loss": 0.3350512981414795,
|
| 5358 |
-
"mean_token_accuracy": 0.9074875012040138,
|
| 5359 |
-
"num_tokens": 13657117.0,
|
| 5360 |
-
"step": 5130
|
| 5361 |
-
},
|
| 5362 |
-
{
|
| 5363 |
-
"entropy": 0.2882433099672198,
|
| 5364 |
-
"epoch": 0.3894013144188337,
|
| 5365 |
-
"grad_norm": 0.5046371221542358,
|
| 5366 |
-
"learning_rate": 6.979522651839136e-05,
|
| 5367 |
-
"loss": 0.336344575881958,
|
| 5368 |
-
"mean_token_accuracy": 0.9081276401877403,
|
| 5369 |
-
"num_tokens": 13748643.0,
|
| 5370 |
-
"step": 5140
|
| 5371 |
-
},
|
| 5372 |
-
{
|
| 5373 |
-
"entropy": 0.2940206663683057,
|
| 5374 |
-
"epoch": 0.3901589045247069,
|
| 5375 |
-
"grad_norm": 0.45559731125831604,
|
| 5376 |
-
"learning_rate": 6.968251090809708e-05,
|
| 5377 |
-
"loss": 0.341286826133728,
|
| 5378 |
-
"mean_token_accuracy": 0.9080539286136627,
|
| 5379 |
-
"num_tokens": 13848277.0,
|
| 5380 |
-
"step": 5150
|
| 5381 |
-
},
|
| 5382 |
-
{
|
| 5383 |
-
"entropy": 0.34208077993243935,
|
| 5384 |
-
"epoch": 0.39091649463058015,
|
| 5385 |
-
"grad_norm": 0.36759278178215027,
|
| 5386 |
-
"learning_rate": 6.956967680589353e-05,
|
| 5387 |
-
"loss": 0.3881381034851074,
|
| 5388 |
-
"mean_token_accuracy": 0.8900120064616204,
|
| 5389 |
-
"num_tokens": 13953548.0,
|
| 5390 |
-
"step": 5160
|
| 5391 |
-
},
|
| 5392 |
-
{
|
| 5393 |
-
"entropy": 0.2917953571304679,
|
| 5394 |
-
"epoch": 0.3916740847364533,
|
| 5395 |
-
"grad_norm": 0.6413111686706543,
|
| 5396 |
-
"learning_rate": 6.945672489106028e-05,
|
| 5397 |
-
"loss": 0.3312584400177002,
|
| 5398 |
-
"mean_token_accuracy": 0.9041773527860641,
|
| 5399 |
-
"num_tokens": 14050050.0,
|
| 5400 |
-
"step": 5170
|
| 5401 |
-
},
|
| 5402 |
-
{
|
| 5403 |
-
"entropy": 0.3064444277435541,
|
| 5404 |
-
"epoch": 0.39243167484232655,
|
| 5405 |
-
"grad_norm": 0.4423501491546631,
|
| 5406 |
-
"learning_rate": 6.934365584358624e-05,
|
| 5407 |
-
"loss": 0.3687903881072998,
|
| 5408 |
-
"mean_token_accuracy": 0.9040727481245995,
|
| 5409 |
-
"num_tokens": 14146367.0,
|
| 5410 |
-
"step": 5180
|
| 5411 |
-
},
|
| 5412 |
-
{
|
| 5413 |
-
"entropy": 0.32541023269295694,
|
| 5414 |
-
"epoch": 0.3931892649481998,
|
| 5415 |
-
"grad_norm": 0.452756404876709,
|
| 5416 |
-
"learning_rate": 6.923047034416542e-05,
|
| 5417 |
-
"loss": 0.3667147636413574,
|
| 5418 |
-
"mean_token_accuracy": 0.8978587627410889,
|
| 5419 |
-
"num_tokens": 14241995.0,
|
| 5420 |
-
"step": 5190
|
| 5421 |
-
},
|
| 5422 |
-
{
|
| 5423 |
-
"entropy": 0.313952512294054,
|
| 5424 |
-
"epoch": 0.393946855054073,
|
| 5425 |
-
"grad_norm": 0.515623152256012,
|
| 5426 |
-
"learning_rate": 6.911716907419283e-05,
|
| 5427 |
-
"loss": 0.3508143424987793,
|
| 5428 |
-
"mean_token_accuracy": 0.9012334987521171,
|
| 5429 |
-
"num_tokens": 14338325.0,
|
| 5430 |
-
"step": 5200
|
| 5431 |
-
},
|
| 5432 |
-
{
|
| 5433 |
-
"entropy": 0.2895424833521247,
|
| 5434 |
-
"epoch": 0.39470444515994624,
|
| 5435 |
-
"grad_norm": 0.4226093292236328,
|
| 5436 |
-
"learning_rate": 6.900375271576055e-05,
|
| 5437 |
-
"loss": 0.3347956657409668,
|
| 5438 |
-
"mean_token_accuracy": 0.9094123750925064,
|
| 5439 |
-
"num_tokens": 14444097.0,
|
| 5440 |
-
"step": 5210
|
| 5441 |
-
},
|
| 5442 |
-
{
|
| 5443 |
-
"entropy": 0.32342710457742213,
|
| 5444 |
-
"epoch": 0.3954620352658194,
|
| 5445 |
-
"grad_norm": 0.42203420400619507,
|
| 5446 |
-
"learning_rate": 6.889022195165344e-05,
|
| 5447 |
-
"loss": 0.3710180759429932,
|
| 5448 |
-
"mean_token_accuracy": 0.8959487065672874,
|
| 5449 |
-
"num_tokens": 14546666.0,
|
| 5450 |
-
"step": 5220
|
| 5451 |
-
},
|
| 5452 |
-
{
|
| 5453 |
-
"entropy": 0.2814556578174233,
|
| 5454 |
-
"epoch": 0.39621962537169264,
|
| 5455 |
-
"grad_norm": 0.3739827573299408,
|
| 5456 |
-
"learning_rate": 6.877657746534513e-05,
|
| 5457 |
-
"loss": 0.32498607635498045,
|
| 5458 |
-
"mean_token_accuracy": 0.9107230916619301,
|
| 5459 |
-
"num_tokens": 14642321.0,
|
| 5460 |
-
"step": 5230
|
| 5461 |
-
},
|
| 5462 |
-
{
|
| 5463 |
-
"entropy": 0.32770489919930695,
|
| 5464 |
-
"epoch": 0.39697721547756587,
|
| 5465 |
-
"grad_norm": 0.42540836334228516,
|
| 5466 |
-
"learning_rate": 6.866281994099389e-05,
|
| 5467 |
-
"loss": 0.3772111415863037,
|
| 5468 |
-
"mean_token_accuracy": 0.8976331815123558,
|
| 5469 |
-
"num_tokens": 14754272.0,
|
| 5470 |
-
"step": 5240
|
| 5471 |
-
},
|
| 5472 |
-
{
|
| 5473 |
-
"entropy": 0.29973213113844394,
|
| 5474 |
-
"epoch": 0.3977348055834391,
|
| 5475 |
-
"grad_norm": 0.41516080498695374,
|
| 5476 |
-
"learning_rate": 6.854895006343842e-05,
|
| 5477 |
-
"loss": 0.33910579681396485,
|
| 5478 |
-
"mean_token_accuracy": 0.9034910127520561,
|
| 5479 |
-
"num_tokens": 14862455.0,
|
| 5480 |
-
"step": 5250
|
| 5481 |
-
},
|
| 5482 |
-
{
|
| 5483 |
-
"epoch": 0.3977348055834391,
|
| 5484 |
-
"eval_entropy": 0.3002741236923339,
|
| 5485 |
-
"eval_loss": 0.29961150884628296,
|
| 5486 |
-
"eval_mean_token_accuracy": 0.9034789078467794,
|
| 5487 |
-
"eval_num_tokens": 14862455.0,
|
| 5488 |
-
"eval_runtime": 357.7363,
|
| 5489 |
-
"eval_samples_per_second": 5.965,
|
| 5490 |
-
"eval_steps_per_second": 1.493,
|
| 5491 |
-
"step": 5250
|
| 5492 |
-
},
|
| 5493 |
-
{
|
| 5494 |
-
"entropy": 0.3370519518852234,
|
| 5495 |
-
"epoch": 0.3984923956893123,
|
| 5496 |
-
"grad_norm": 0.4814501404762268,
|
| 5497 |
-
"learning_rate": 6.843496851819388e-05,
|
| 5498 |
-
"loss": 0.39408583641052247,
|
| 5499 |
-
"mean_token_accuracy": 0.8927295506000519,
|
| 5500 |
-
"num_tokens": 14974981.0,
|
| 5501 |
-
"step": 5260
|
| 5502 |
-
},
|
| 5503 |
-
{
|
| 5504 |
-
"entropy": 0.3009413816034794,
|
| 5505 |
-
"epoch": 0.3992499857951855,
|
| 5506 |
-
"grad_norm": 0.5458259582519531,
|
| 5507 |
-
"learning_rate": 6.832087599144771e-05,
|
| 5508 |
-
"loss": 0.3641390323638916,
|
| 5509 |
-
"mean_token_accuracy": 0.9049391418695449,
|
| 5510 |
-
"num_tokens": 15062830.0,
|
| 5511 |
-
"step": 5270
|
| 5512 |
-
},
|
| 5513 |
-
{
|
| 5514 |
-
"entropy": 0.2642664210870862,
|
| 5515 |
-
"epoch": 0.40000757590105873,
|
| 5516 |
-
"grad_norm": 0.696412205696106,
|
| 5517 |
-
"learning_rate": 6.820667317005537e-05,
|
| 5518 |
-
"loss": 0.3026557922363281,
|
| 5519 |
-
"mean_token_accuracy": 0.9156915605068207,
|
| 5520 |
-
"num_tokens": 15160965.0,
|
| 5521 |
-
"step": 5280
|
| 5522 |
-
},
|
| 5523 |
-
{
|
| 5524 |
-
"entropy": 0.29925984516739845,
|
| 5525 |
-
"epoch": 0.40076516600693196,
|
| 5526 |
-
"grad_norm": 0.3674147427082062,
|
| 5527 |
-
"learning_rate": 6.809236074153643e-05,
|
| 5528 |
-
"loss": 0.35786943435668944,
|
| 5529 |
-
"mean_token_accuracy": 0.9058224901556968,
|
| 5530 |
-
"num_tokens": 15263494.0,
|
| 5531 |
-
"step": 5290
|
| 5532 |
-
},
|
| 5533 |
-
{
|
| 5534 |
-
"entropy": 0.2937476862221956,
|
| 5535 |
-
"epoch": 0.4015227561128052,
|
| 5536 |
-
"grad_norm": 0.44328176975250244,
|
| 5537 |
-
"learning_rate": 6.797793939407021e-05,
|
| 5538 |
-
"loss": 0.35303363800048826,
|
| 5539 |
-
"mean_token_accuracy": 0.9050178065896034,
|
| 5540 |
-
"num_tokens": 15364361.0,
|
| 5541 |
-
"step": 5300
|
| 5542 |
-
},
|
| 5543 |
-
{
|
| 5544 |
-
"entropy": 0.27982344552874566,
|
| 5545 |
-
"epoch": 0.40228034621867836,
|
| 5546 |
-
"grad_norm": 0.43815651535987854,
|
| 5547 |
-
"learning_rate": 6.786340981649182e-05,
|
| 5548 |
-
"loss": 0.32604780197143557,
|
| 5549 |
-
"mean_token_accuracy": 0.9095379173755646,
|
| 5550 |
-
"num_tokens": 15464753.0,
|
| 5551 |
-
"step": 5310
|
| 5552 |
-
},
|
| 5553 |
-
{
|
| 5554 |
-
"entropy": 0.2889172719791532,
|
| 5555 |
-
"epoch": 0.4030379363245516,
|
| 5556 |
-
"grad_norm": 0.46756795048713684,
|
| 5557 |
-
"learning_rate": 6.774877269828787e-05,
|
| 5558 |
-
"loss": 0.31214098930358886,
|
| 5559 |
-
"mean_token_accuracy": 0.9092742562294006,
|
| 5560 |
-
"num_tokens": 15555172.0,
|
| 5561 |
-
"step": 5320
|
| 5562 |
-
},
|
| 5563 |
-
{
|
| 5564 |
-
"entropy": 0.30592968575656415,
|
| 5565 |
-
"epoch": 0.4037955264304248,
|
| 5566 |
-
"grad_norm": 0.5958737730979919,
|
| 5567 |
-
"learning_rate": 6.763402872959245e-05,
|
| 5568 |
-
"loss": 0.34911766052246096,
|
| 5569 |
-
"mean_token_accuracy": 0.9045114248991013,
|
| 5570 |
-
"num_tokens": 15655737.0,
|
| 5571 |
-
"step": 5330
|
| 5572 |
-
},
|
| 5573 |
-
{
|
| 5574 |
-
"entropy": 0.3149020399898291,
|
| 5575 |
-
"epoch": 0.40455311653629805,
|
| 5576 |
-
"grad_norm": 0.6274760365486145,
|
| 5577 |
-
"learning_rate": 6.751917860118284e-05,
|
| 5578 |
-
"loss": 0.3545043468475342,
|
| 5579 |
-
"mean_token_accuracy": 0.9007835671305656,
|
| 5580 |
-
"num_tokens": 15753377.0,
|
| 5581 |
-
"step": 5340
|
| 5582 |
-
},
|
| 5583 |
-
{
|
| 5584 |
-
"entropy": 0.27794329486787317,
|
| 5585 |
-
"epoch": 0.4053107066421712,
|
| 5586 |
-
"grad_norm": 0.40906354784965515,
|
| 5587 |
-
"learning_rate": 6.740422300447546e-05,
|
| 5588 |
-
"loss": 0.30841042995452883,
|
| 5589 |
-
"mean_token_accuracy": 0.9137280076742172,
|
| 5590 |
-
"num_tokens": 15843846.0,
|
| 5591 |
-
"step": 5350
|
| 5592 |
-
},
|
| 5593 |
-
{
|
| 5594 |
-
"entropy": 0.29877801686525346,
|
| 5595 |
-
"epoch": 0.40606829674804445,
|
| 5596 |
-
"grad_norm": 0.43842074275016785,
|
| 5597 |
-
"learning_rate": 6.728916263152165e-05,
|
| 5598 |
-
"loss": 0.3355414390563965,
|
| 5599 |
-
"mean_token_accuracy": 0.9042882114648819,
|
| 5600 |
-
"num_tokens": 15940264.0,
|
| 5601 |
-
"step": 5360
|
| 5602 |
-
},
|
| 5603 |
-
{
|
| 5604 |
-
"entropy": 0.3168094277381897,
|
| 5605 |
-
"epoch": 0.4068258868539177,
|
| 5606 |
-
"grad_norm": 0.3877590596675873,
|
| 5607 |
-
"learning_rate": 6.717399817500352e-05,
|
| 5608 |
-
"loss": 0.36755588054656985,
|
| 5609 |
-
"mean_token_accuracy": 0.8993088215589523,
|
| 5610 |
-
"num_tokens": 16044378.0,
|
| 5611 |
-
"step": 5370
|
| 5612 |
-
},
|
| 5613 |
-
{
|
| 5614 |
-
"entropy": 0.28583061657845976,
|
| 5615 |
-
"epoch": 0.4075834769597909,
|
| 5616 |
-
"grad_norm": 0.43321725726127625,
|
| 5617 |
-
"learning_rate": 6.70587303282298e-05,
|
| 5618 |
-
"loss": 0.3152596473693848,
|
| 5619 |
-
"mean_token_accuracy": 0.9106203690171242,
|
| 5620 |
-
"num_tokens": 16136663.0,
|
| 5621 |
-
"step": 5380
|
| 5622 |
-
},
|
| 5623 |
-
{
|
| 5624 |
-
"entropy": 0.3030781650915742,
|
| 5625 |
-
"epoch": 0.40834106706566414,
|
| 5626 |
-
"grad_norm": 0.46747297048568726,
|
| 5627 |
-
"learning_rate": 6.694335978513165e-05,
|
| 5628 |
-
"loss": 0.3706023693084717,
|
| 5629 |
-
"mean_token_accuracy": 0.904845617711544,
|
| 5630 |
-
"num_tokens": 16230004.0,
|
| 5631 |
-
"step": 5390
|
| 5632 |
-
},
|
| 5633 |
-
{
|
| 5634 |
-
"entropy": 0.2962409736588597,
|
| 5635 |
-
"epoch": 0.4090986571715373,
|
| 5636 |
-
"grad_norm": 0.47175347805023193,
|
| 5637 |
-
"learning_rate": 6.68278872402584e-05,
|
| 5638 |
-
"loss": 0.3297926187515259,
|
| 5639 |
-
"mean_token_accuracy": 0.9071791052818299,
|
| 5640 |
-
"num_tokens": 16337774.0,
|
| 5641 |
-
"step": 5400
|
| 5642 |
-
},
|
| 5643 |
-
{
|
| 5644 |
-
"entropy": 0.2856398120522499,
|
| 5645 |
-
"epoch": 0.40985624727741055,
|
| 5646 |
-
"grad_norm": 0.5184034109115601,
|
| 5647 |
-
"learning_rate": 6.671231338877353e-05,
|
| 5648 |
-
"loss": 0.33323845863342283,
|
| 5649 |
-
"mean_token_accuracy": 0.9066996708512306,
|
| 5650 |
-
"num_tokens": 16434958.0,
|
| 5651 |
-
"step": 5410
|
| 5652 |
-
},
|
| 5653 |
-
{
|
| 5654 |
-
"entropy": 0.28190935757011176,
|
| 5655 |
-
"epoch": 0.4106138373832838,
|
| 5656 |
-
"grad_norm": 0.5188422203063965,
|
| 5657 |
-
"learning_rate": 6.65966389264504e-05,
|
| 5658 |
-
"loss": 0.36467928886413575,
|
| 5659 |
-
"mean_token_accuracy": 0.9099793508648872,
|
| 5660 |
-
"num_tokens": 16528627.0,
|
| 5661 |
-
"step": 5420
|
| 5662 |
-
},
|
| 5663 |
-
{
|
| 5664 |
-
"entropy": 0.3506340319290757,
|
| 5665 |
-
"epoch": 0.411371427489157,
|
| 5666 |
-
"grad_norm": 0.42907848954200745,
|
| 5667 |
-
"learning_rate": 6.648086454966802e-05,
|
| 5668 |
-
"loss": 0.39072270393371583,
|
| 5669 |
-
"mean_token_accuracy": 0.8875074088573456,
|
| 5670 |
-
"num_tokens": 16636661.0,
|
| 5671 |
-
"step": 5430
|
| 5672 |
-
},
|
| 5673 |
-
{
|
| 5674 |
-
"entropy": 0.29132668040692805,
|
| 5675 |
-
"epoch": 0.41212901759503023,
|
| 5676 |
-
"grad_norm": 0.39410361647605896,
|
| 5677 |
-
"learning_rate": 6.636499095540693e-05,
|
| 5678 |
-
"loss": 0.3283844470977783,
|
| 5679 |
-
"mean_token_accuracy": 0.9061837956309319,
|
| 5680 |
-
"num_tokens": 16736056.0,
|
| 5681 |
-
"step": 5440
|
| 5682 |
-
},
|
| 5683 |
-
{
|
| 5684 |
-
"entropy": 0.2893992530182004,
|
| 5685 |
-
"epoch": 0.4128866077009034,
|
| 5686 |
-
"grad_norm": 0.44495856761932373,
|
| 5687 |
-
"learning_rate": 6.624901884124497e-05,
|
| 5688 |
-
"loss": 0.3219173431396484,
|
| 5689 |
-
"mean_token_accuracy": 0.9095578044652939,
|
| 5690 |
-
"num_tokens": 16831788.0,
|
| 5691 |
-
"step": 5450
|
| 5692 |
-
},
|
| 5693 |
-
{
|
| 5694 |
-
"entropy": 0.2820703413337469,
|
| 5695 |
-
"epoch": 0.41364419780677664,
|
| 5696 |
-
"grad_norm": 0.437040776014328,
|
| 5697 |
-
"learning_rate": 6.613294890535309e-05,
|
| 5698 |
-
"loss": 0.33107748031616213,
|
| 5699 |
-
"mean_token_accuracy": 0.9077925100922585,
|
| 5700 |
-
"num_tokens": 16930665.0,
|
| 5701 |
-
"step": 5460
|
| 5702 |
-
},
|
| 5703 |
-
{
|
| 5704 |
-
"entropy": 0.24273199774324894,
|
| 5705 |
-
"epoch": 0.41440178791264987,
|
| 5706 |
-
"grad_norm": 0.33514174818992615,
|
| 5707 |
-
"learning_rate": 6.601678184649111e-05,
|
| 5708 |
-
"loss": 0.28858370780944825,
|
| 5709 |
-
"mean_token_accuracy": 0.9210346817970276,
|
| 5710 |
-
"num_tokens": 17019821.0,
|
| 5711 |
-
"step": 5470
|
| 5712 |
-
},
|
| 5713 |
-
{
|
| 5714 |
-
"entropy": 0.2930982571095228,
|
| 5715 |
-
"epoch": 0.4151593780185231,
|
| 5716 |
-
"grad_norm": 0.38038352131843567,
|
| 5717 |
-
"learning_rate": 6.590051836400363e-05,
|
| 5718 |
-
"loss": 0.3299015998840332,
|
| 5719 |
-
"mean_token_accuracy": 0.9075604423880577,
|
| 5720 |
-
"num_tokens": 17109270.0,
|
| 5721 |
-
"step": 5480
|
| 5722 |
-
},
|
| 5723 |
-
{
|
| 5724 |
-
"entropy": 0.29958692379295826,
|
| 5725 |
-
"epoch": 0.41591696812439627,
|
| 5726 |
-
"grad_norm": 0.42856743931770325,
|
| 5727 |
-
"learning_rate": 6.578415915781562e-05,
|
| 5728 |
-
"loss": 0.3183560848236084,
|
| 5729 |
-
"mean_token_accuracy": 0.9050814941525459,
|
| 5730 |
-
"num_tokens": 17203521.0,
|
| 5731 |
-
"step": 5490
|
| 5732 |
-
},
|
| 5733 |
-
{
|
| 5734 |
-
"entropy": 0.31369390320032836,
|
| 5735 |
-
"epoch": 0.4166745582302695,
|
| 5736 |
-
"grad_norm": 0.42200779914855957,
|
| 5737 |
-
"learning_rate": 6.566770492842842e-05,
|
| 5738 |
-
"loss": 0.3765551090240479,
|
| 5739 |
-
"mean_token_accuracy": 0.8998699098825454,
|
| 5740 |
-
"num_tokens": 17300889.0,
|
| 5741 |
-
"step": 5500
|
| 5742 |
-
},
|
| 5743 |
-
{
|
| 5744 |
-
"epoch": 0.4166745582302695,
|
| 5745 |
-
"eval_entropy": 0.3062387081037523,
|
| 5746 |
-
"eval_loss": 0.29997920989990234,
|
| 5747 |
-
"eval_mean_token_accuracy": 0.9034306161412585,
|
| 5748 |
-
"eval_num_tokens": 17300889.0,
|
| 5749 |
-
"eval_runtime": 357.6857,
|
| 5750 |
-
"eval_samples_per_second": 5.966,
|
| 5751 |
-
"eval_steps_per_second": 1.493,
|
| 5752 |
-
"step": 5500
|
| 5753 |
-
},
|
| 5754 |
-
{
|
| 5755 |
-
"entropy": 0.3275546273216605,
|
| 5756 |
-
"epoch": 0.4174321483361427,
|
| 5757 |
-
"grad_norm": 0.4728966951370239,
|
| 5758 |
-
"learning_rate": 6.555115637691536e-05,
|
| 5759 |
-
"loss": 0.37491455078125,
|
| 5760 |
-
"mean_token_accuracy": 0.8985271066427231,
|
| 5761 |
-
"num_tokens": 17402212.0,
|
| 5762 |
-
"step": 5510
|
| 5763 |
-
},
|
| 5764 |
-
{
|
| 5765 |
-
"entropy": 0.32291659861803057,
|
| 5766 |
-
"epoch": 0.41818973844201596,
|
| 5767 |
-
"grad_norm": 0.41598349809646606,
|
| 5768 |
-
"learning_rate": 6.543451420491764e-05,
|
| 5769 |
-
"loss": 0.36802659034729,
|
| 5770 |
-
"mean_token_accuracy": 0.8929702371358872,
|
| 5771 |
-
"num_tokens": 17511141.0,
|
| 5772 |
-
"step": 5520
|
| 5773 |
-
},
|
| 5774 |
-
{
|
| 5775 |
-
"entropy": 0.32626404892653227,
|
| 5776 |
-
"epoch": 0.4189473285478892,
|
| 5777 |
-
"grad_norm": 0.47906798124313354,
|
| 5778 |
-
"learning_rate": 6.531777911464007e-05,
|
| 5779 |
-
"loss": 0.3838531017303467,
|
| 5780 |
-
"mean_token_accuracy": 0.894839309155941,
|
| 5781 |
-
"num_tokens": 17612806.0,
|
| 5782 |
-
"step": 5530
|
| 5783 |
-
},
|
| 5784 |
-
{
|
| 5785 |
-
"entropy": 0.2872452598065138,
|
| 5786 |
-
"epoch": 0.41970491865376236,
|
| 5787 |
-
"grad_norm": 0.5578105449676514,
|
| 5788 |
-
"learning_rate": 6.520095180884683e-05,
|
| 5789 |
-
"loss": 0.32535998821258544,
|
| 5790 |
-
"mean_token_accuracy": 0.9079094067215919,
|
| 5791 |
-
"num_tokens": 17706491.0,
|
| 5792 |
-
"step": 5540
|
| 5793 |
-
},
|
| 5794 |
-
{
|
| 5795 |
-
"entropy": 0.3136607936583459,
|
| 5796 |
-
"epoch": 0.4204625087596356,
|
| 5797 |
-
"grad_norm": 0.47384339570999146,
|
| 5798 |
-
"learning_rate": 6.508403299085724e-05,
|
| 5799 |
-
"loss": 0.36521048545837403,
|
| 5800 |
-
"mean_token_accuracy": 0.9005001828074455,
|
| 5801 |
-
"num_tokens": 17802170.0,
|
| 5802 |
-
"step": 5550
|
| 5803 |
-
},
|
| 5804 |
-
{
|
| 5805 |
-
"entropy": 0.29471229687333106,
|
| 5806 |
-
"epoch": 0.4212200988655088,
|
| 5807 |
-
"grad_norm": 0.6994742751121521,
|
| 5808 |
-
"learning_rate": 6.496702336454157e-05,
|
| 5809 |
-
"loss": 0.33774256706237793,
|
| 5810 |
-
"mean_token_accuracy": 0.9037168011069298,
|
| 5811 |
-
"num_tokens": 17893281.0,
|
| 5812 |
-
"step": 5560
|
| 5813 |
-
},
|
| 5814 |
-
{
|
| 5815 |
-
"entropy": 0.3260188512504101,
|
| 5816 |
-
"epoch": 0.42197768897138205,
|
| 5817 |
-
"grad_norm": 0.5837869048118591,
|
| 5818 |
-
"learning_rate": 6.484992363431676e-05,
|
| 5819 |
-
"loss": 0.3840935230255127,
|
| 5820 |
-
"mean_token_accuracy": 0.8950624346733094,
|
| 5821 |
-
"num_tokens": 17997803.0,
|
| 5822 |
-
"step": 5570
|
| 5823 |
-
},
|
| 5824 |
-
{
|
| 5825 |
-
"entropy": 0.32133214343339206,
|
| 5826 |
-
"epoch": 0.4227352790772553,
|
| 5827 |
-
"grad_norm": 0.37200942635536194,
|
| 5828 |
-
"learning_rate": 6.473273450514219e-05,
|
| 5829 |
-
"loss": 0.3590023279190063,
|
| 5830 |
-
"mean_token_accuracy": 0.9000173777341842,
|
| 5831 |
-
"num_tokens": 18090099.0,
|
| 5832 |
-
"step": 5580
|
| 5833 |
-
},
|
| 5834 |
-
{
|
| 5835 |
-
"entropy": 0.32563979979604485,
|
| 5836 |
-
"epoch": 0.42349286918312845,
|
| 5837 |
-
"grad_norm": 0.37389349937438965,
|
| 5838 |
-
"learning_rate": 6.461545668251542e-05,
|
| 5839 |
-
"loss": 0.3636253833770752,
|
| 5840 |
-
"mean_token_accuracy": 0.897948132455349,
|
| 5841 |
-
"num_tokens": 18189492.0,
|
| 5842 |
-
"step": 5590
|
| 5843 |
-
},
|
| 5844 |
-
{
|
| 5845 |
-
"entropy": 0.2901277866214514,
|
| 5846 |
-
"epoch": 0.4242504592890017,
|
| 5847 |
-
"grad_norm": 0.38119590282440186,
|
| 5848 |
-
"learning_rate": 6.449809087246796e-05,
|
| 5849 |
-
"loss": 0.32559292316436766,
|
| 5850 |
-
"mean_token_accuracy": 0.9096684545278549,
|
| 5851 |
-
"num_tokens": 18285317.0,
|
| 5852 |
-
"step": 5600
|
| 5853 |
-
},
|
| 5854 |
-
{
|
| 5855 |
-
"entropy": 0.31151607912033796,
|
| 5856 |
-
"epoch": 0.4250080493948749,
|
| 5857 |
-
"grad_norm": 0.34305769205093384,
|
| 5858 |
-
"learning_rate": 6.438063778156106e-05,
|
| 5859 |
-
"loss": 0.33219189643859864,
|
| 5860 |
-
"mean_token_accuracy": 0.9019489496946335,
|
| 5861 |
-
"num_tokens": 18380421.0,
|
| 5862 |
-
"step": 5610
|
| 5863 |
-
},
|
| 5864 |
-
{
|
| 5865 |
-
"entropy": 0.2918217882514,
|
| 5866 |
-
"epoch": 0.42576563950074814,
|
| 5867 |
-
"grad_norm": 0.42412933707237244,
|
| 5868 |
-
"learning_rate": 6.426309811688136e-05,
|
| 5869 |
-
"loss": 0.3444188117980957,
|
| 5870 |
-
"mean_token_accuracy": 0.9022915855050087,
|
| 5871 |
-
"num_tokens": 18468327.0,
|
| 5872 |
-
"step": 5620
|
| 5873 |
-
},
|
| 5874 |
-
{
|
| 5875 |
-
"entropy": 0.31709956247359516,
|
| 5876 |
-
"epoch": 0.4265232296066213,
|
| 5877 |
-
"grad_norm": 0.4144665002822876,
|
| 5878 |
-
"learning_rate": 6.414547258603674e-05,
|
| 5879 |
-
"loss": 0.3667241811752319,
|
| 5880 |
-
"mean_token_accuracy": 0.9032866105437278,
|
| 5881 |
-
"num_tokens": 18559727.0,
|
| 5882 |
-
"step": 5630
|
| 5883 |
-
},
|
| 5884 |
-
{
|
| 5885 |
-
"entropy": 0.2825597863644361,
|
| 5886 |
-
"epoch": 0.42728081971249454,
|
| 5887 |
-
"grad_norm": 0.4813626706600189,
|
| 5888 |
-
"learning_rate": 6.402776189715195e-05,
|
| 5889 |
-
"loss": 0.32198758125305177,
|
| 5890 |
-
"mean_token_accuracy": 0.9123972833156586,
|
| 5891 |
-
"num_tokens": 18652384.0,
|
| 5892 |
-
"step": 5640
|
| 5893 |
-
},
|
| 5894 |
-
{
|
| 5895 |
-
"entropy": 0.33043870665133,
|
| 5896 |
-
"epoch": 0.42803840981836777,
|
| 5897 |
-
"grad_norm": 0.4929293096065521,
|
| 5898 |
-
"learning_rate": 6.390996675886445e-05,
|
| 5899 |
-
"loss": 0.3855895042419434,
|
| 5900 |
-
"mean_token_accuracy": 0.8926237165927887,
|
| 5901 |
-
"num_tokens": 18762607.0,
|
| 5902 |
-
"step": 5650
|
| 5903 |
-
},
|
| 5904 |
-
{
|
| 5905 |
-
"entropy": 0.2616382507607341,
|
| 5906 |
-
"epoch": 0.428795999924241,
|
| 5907 |
-
"grad_norm": 0.5456161499023438,
|
| 5908 |
-
"learning_rate": 6.379208788032009e-05,
|
| 5909 |
-
"loss": 0.3109184980392456,
|
| 5910 |
-
"mean_token_accuracy": 0.9183192536234855,
|
| 5911 |
-
"num_tokens": 18851290.0,
|
| 5912 |
-
"step": 5660
|
| 5913 |
-
},
|
| 5914 |
-
{
|
| 5915 |
-
"entropy": 0.2915184019133449,
|
| 5916 |
-
"epoch": 0.42955359003011423,
|
| 5917 |
-
"grad_norm": 0.5472036004066467,
|
| 5918 |
-
"learning_rate": 6.367412597116884e-05,
|
| 5919 |
-
"loss": 0.33669614791870117,
|
| 5920 |
-
"mean_token_accuracy": 0.9075099468231201,
|
| 5921 |
-
"num_tokens": 18951844.0,
|
| 5922 |
-
"step": 5670
|
| 5923 |
-
},
|
| 5924 |
-
{
|
| 5925 |
-
"entropy": 0.3137003116309643,
|
| 5926 |
-
"epoch": 0.4303111801359874,
|
| 5927 |
-
"grad_norm": 0.3971874415874481,
|
| 5928 |
-
"learning_rate": 6.355608174156053e-05,
|
| 5929 |
-
"loss": 0.3647712230682373,
|
| 5930 |
-
"mean_token_accuracy": 0.8999422565102577,
|
| 5931 |
-
"num_tokens": 19057970.0,
|
| 5932 |
-
"step": 5680
|
| 5933 |
-
},
|
| 5934 |
-
{
|
| 5935 |
-
"entropy": 0.3097508229315281,
|
| 5936 |
-
"epoch": 0.43106877024186063,
|
| 5937 |
-
"grad_norm": 0.4564378559589386,
|
| 5938 |
-
"learning_rate": 6.343795590214056e-05,
|
| 5939 |
-
"loss": 0.36319799423217775,
|
| 5940 |
-
"mean_token_accuracy": 0.9011824861168861,
|
| 5941 |
-
"num_tokens": 19159331.0,
|
| 5942 |
-
"step": 5690
|
| 5943 |
-
},
|
| 5944 |
-
{
|
| 5945 |
-
"entropy": 0.24787705838680268,
|
| 5946 |
-
"epoch": 0.43182636034773386,
|
| 5947 |
-
"grad_norm": 0.3847849667072296,
|
| 5948 |
-
"learning_rate": 6.331974916404571e-05,
|
| 5949 |
-
"loss": 0.3177677631378174,
|
| 5950 |
-
"mean_token_accuracy": 0.918899592757225,
|
| 5951 |
-
"num_tokens": 19250927.0,
|
| 5952 |
-
"step": 5700
|
| 5953 |
-
},
|
| 5954 |
-
{
|
| 5955 |
-
"entropy": 0.303675995580852,
|
| 5956 |
-
"epoch": 0.4325839504536071,
|
| 5957 |
-
"grad_norm": 0.4380459785461426,
|
| 5958 |
-
"learning_rate": 6.320146223889966e-05,
|
| 5959 |
-
"loss": 0.3646137475967407,
|
| 5960 |
-
"mean_token_accuracy": 0.904253339767456,
|
| 5961 |
-
"num_tokens": 19348065.0,
|
| 5962 |
-
"step": 5710
|
| 5963 |
-
},
|
| 5964 |
-
{
|
| 5965 |
-
"entropy": 0.34324677288532257,
|
| 5966 |
-
"epoch": 0.4333415405594803,
|
| 5967 |
-
"grad_norm": 0.44166335463523865,
|
| 5968 |
-
"learning_rate": 6.308309583880891e-05,
|
| 5969 |
-
"loss": 0.37530155181884767,
|
| 5970 |
-
"mean_token_accuracy": 0.8911663755774498,
|
| 5971 |
-
"num_tokens": 19454025.0,
|
| 5972 |
-
"step": 5720
|
| 5973 |
-
},
|
| 5974 |
-
{
|
| 5975 |
-
"entropy": 0.3109349662438035,
|
| 5976 |
-
"epoch": 0.4340991306653535,
|
| 5977 |
-
"grad_norm": 0.33173251152038574,
|
| 5978 |
-
"learning_rate": 6.296465067635842e-05,
|
| 5979 |
-
"loss": 0.36859874725341796,
|
| 5980 |
-
"mean_token_accuracy": 0.8991603925824165,
|
| 5981 |
-
"num_tokens": 19553057.0,
|
| 5982 |
-
"step": 5730
|
| 5983 |
-
},
|
| 5984 |
-
{
|
| 5985 |
-
"entropy": 0.28704450093209743,
|
| 5986 |
-
"epoch": 0.4348567207712267,
|
| 5987 |
-
"grad_norm": 0.3608269989490509,
|
| 5988 |
-
"learning_rate": 6.284612746460724e-05,
|
| 5989 |
-
"loss": 0.3440445899963379,
|
| 5990 |
-
"mean_token_accuracy": 0.9064769595861435,
|
| 5991 |
-
"num_tokens": 19650594.0,
|
| 5992 |
-
"step": 5740
|
| 5993 |
-
},
|
| 5994 |
-
{
|
| 5995 |
-
"entropy": 0.30900111999362706,
|
| 5996 |
-
"epoch": 0.43561431087709995,
|
| 5997 |
-
"grad_norm": 0.44841620326042175,
|
| 5998 |
-
"learning_rate": 6.272752691708437e-05,
|
| 5999 |
-
"loss": 0.3561258316040039,
|
| 6000 |
-
"mean_token_accuracy": 0.9029030427336693,
|
| 6001 |
-
"num_tokens": 19749863.0,
|
| 6002 |
-
"step": 5750
|
| 6003 |
-
},
|
| 6004 |
-
{
|
| 6005 |
-
"epoch": 0.43561431087709995,
|
| 6006 |
-
"eval_entropy": 0.30515875275271215,
|
| 6007 |
-
"eval_loss": 0.2985568642616272,
|
| 6008 |
-
"eval_mean_token_accuracy": 0.9037165420778682,
|
| 6009 |
-
"eval_num_tokens": 19749863.0,
|
| 6010 |
-
"eval_runtime": 357.722,
|
| 6011 |
-
"eval_samples_per_second": 5.966,
|
| 6012 |
-
"eval_steps_per_second": 1.493,
|
| 6013 |
-
"step": 5750
|
| 6014 |
-
},
|
| 6015 |
-
{
|
| 6016 |
-
"entropy": 0.2830718349665403,
|
| 6017 |
-
"epoch": 0.4363719009829732,
|
| 6018 |
-
"grad_norm": 0.47940748929977417,
|
| 6019 |
-
"learning_rate": 6.260884974778435e-05,
|
| 6020 |
-
"loss": 0.3643668174743652,
|
| 6021 |
-
"mean_token_accuracy": 0.9123772129416465,
|
| 6022 |
-
"num_tokens": 19840370.0,
|
| 6023 |
-
"step": 5760
|
| 6024 |
-
},
|
| 6025 |
-
{
|
| 6026 |
-
"entropy": 0.2904076725244522,
|
| 6027 |
-
"epoch": 0.43712949108884636,
|
| 6028 |
-
"grad_norm": 0.4412832260131836,
|
| 6029 |
-
"learning_rate": 6.249009667116297e-05,
|
| 6030 |
-
"loss": 0.3307987928390503,
|
| 6031 |
-
"mean_token_accuracy": 0.9063455522060394,
|
| 6032 |
-
"num_tokens": 19937394.0,
|
| 6033 |
-
"step": 5770
|
| 6034 |
-
},
|
| 6035 |
-
{
|
| 6036 |
-
"entropy": 0.2838974526152015,
|
| 6037 |
-
"epoch": 0.4378870811947196,
|
| 6038 |
-
"grad_norm": 0.41971099376678467,
|
| 6039 |
-
"learning_rate": 6.237126840213304e-05,
|
| 6040 |
-
"loss": 0.33430943489074705,
|
| 6041 |
-
"mean_token_accuracy": 0.9091265559196472,
|
| 6042 |
-
"num_tokens": 20033458.0,
|
| 6043 |
-
"step": 5780
|
| 6044 |
-
},
|
| 6045 |
-
{
|
| 6046 |
-
"entropy": 0.32569307591766117,
|
| 6047 |
-
"epoch": 0.4386446713005928,
|
| 6048 |
-
"grad_norm": 0.4747779667377472,
|
| 6049 |
-
"learning_rate": 6.225236565606004e-05,
|
| 6050 |
-
"loss": 0.3790421724319458,
|
| 6051 |
-
"mean_token_accuracy": 0.8979799091815949,
|
| 6052 |
-
"num_tokens": 20133685.0,
|
| 6053 |
-
"step": 5790
|
| 6054 |
-
},
|
| 6055 |
-
{
|
| 6056 |
-
"entropy": 0.2692491073161364,
|
| 6057 |
-
"epoch": 0.43940226140646604,
|
| 6058 |
-
"grad_norm": 0.5367014408111572,
|
| 6059 |
-
"learning_rate": 6.213338914875774e-05,
|
| 6060 |
-
"loss": 0.3243349552154541,
|
| 6061 |
-
"mean_token_accuracy": 0.913755439221859,
|
| 6062 |
-
"num_tokens": 20224830.0,
|
| 6063 |
-
"step": 5800
|
| 6064 |
-
},
|
| 6065 |
-
{
|
| 6066 |
-
"entropy": 0.32781150061637165,
|
| 6067 |
-
"epoch": 0.4401598515123393,
|
| 6068 |
-
"grad_norm": 0.43051204085350037,
|
| 6069 |
-
"learning_rate": 6.201433959648406e-05,
|
| 6070 |
-
"loss": 0.3987837791442871,
|
| 6071 |
-
"mean_token_accuracy": 0.8932113081216813,
|
| 6072 |
-
"num_tokens": 20331959.0,
|
| 6073 |
-
"step": 5810
|
| 6074 |
-
},
|
| 6075 |
-
{
|
| 6076 |
-
"entropy": 0.30669678319245575,
|
| 6077 |
-
"epoch": 0.44091744161821245,
|
| 6078 |
-
"grad_norm": 0.4132750928401947,
|
| 6079 |
-
"learning_rate": 6.18952177159366e-05,
|
| 6080 |
-
"loss": 0.33748936653137207,
|
| 6081 |
-
"mean_token_accuracy": 0.9031162261962891,
|
| 6082 |
-
"num_tokens": 20424775.0,
|
| 6083 |
-
"step": 5820
|
| 6084 |
-
},
|
| 6085 |
-
{
|
| 6086 |
-
"entropy": 0.3255875803530216,
|
| 6087 |
-
"epoch": 0.4416750317240857,
|
| 6088 |
-
"grad_norm": 0.3934338092803955,
|
| 6089 |
-
"learning_rate": 6.177602422424845e-05,
|
| 6090 |
-
"loss": 0.37794721126556396,
|
| 6091 |
-
"mean_token_accuracy": 0.8958927169442177,
|
| 6092 |
-
"num_tokens": 20530300.0,
|
| 6093 |
-
"step": 5830
|
| 6094 |
-
},
|
| 6095 |
-
{
|
| 6096 |
-
"entropy": 0.31736835446208717,
|
| 6097 |
-
"epoch": 0.4424326218299589,
|
| 6098 |
-
"grad_norm": 0.5045745372772217,
|
| 6099 |
-
"learning_rate": 6.16567598389837e-05,
|
| 6100 |
-
"loss": 0.3667649269104004,
|
| 6101 |
-
"mean_token_accuracy": 0.8979859486222267,
|
| 6102 |
-
"num_tokens": 20634329.0,
|
| 6103 |
-
"step": 5840
|
| 6104 |
-
},
|
| 6105 |
-
{
|
| 6106 |
-
"entropy": 0.35155081637203695,
|
| 6107 |
-
"epoch": 0.44319021193583213,
|
| 6108 |
-
"grad_norm": 0.2996836006641388,
|
| 6109 |
-
"learning_rate": 6.153742527813335e-05,
|
| 6110 |
-
"loss": 0.39502758979797364,
|
| 6111 |
-
"mean_token_accuracy": 0.8876864716410637,
|
| 6112 |
-
"num_tokens": 20747706.0,
|
| 6113 |
-
"step": 5850
|
| 6114 |
-
},
|
| 6115 |
-
{
|
| 6116 |
-
"entropy": 0.2867135440930724,
|
| 6117 |
-
"epoch": 0.4439478020417053,
|
| 6118 |
-
"grad_norm": 0.48310214281082153,
|
| 6119 |
-
"learning_rate": 6.141802126011082e-05,
|
| 6120 |
-
"loss": 0.3136277198791504,
|
| 6121 |
-
"mean_token_accuracy": 0.9113105162978172,
|
| 6122 |
-
"num_tokens": 20849034.0,
|
| 6123 |
-
"step": 5860
|
| 6124 |
-
},
|
| 6125 |
-
{
|
| 6126 |
-
"entropy": 0.29101333748549224,
|
| 6127 |
-
"epoch": 0.44470539214757854,
|
| 6128 |
-
"grad_norm": 0.4323224723339081,
|
| 6129 |
-
"learning_rate": 6.129854850374767e-05,
|
| 6130 |
-
"loss": 0.34631197452545165,
|
| 6131 |
-
"mean_token_accuracy": 0.9075991079211235,
|
| 6132 |
-
"num_tokens": 20940476.0,
|
| 6133 |
-
"step": 5870
|
| 6134 |
-
},
|
| 6135 |
-
{
|
| 6136 |
-
"entropy": 0.29152622614055873,
|
| 6137 |
-
"epoch": 0.44546298225345177,
|
| 6138 |
-
"grad_norm": 0.47805655002593994,
|
| 6139 |
-
"learning_rate": 6.117900772828925e-05,
|
| 6140 |
-
"loss": 0.36275200843811034,
|
| 6141 |
-
"mean_token_accuracy": 0.9024003639817237,
|
| 6142 |
-
"num_tokens": 21039838.0,
|
| 6143 |
-
"step": 5880
|
| 6144 |
-
},
|
| 6145 |
-
{
|
| 6146 |
-
"entropy": 0.27908653263002636,
|
| 6147 |
-
"epoch": 0.446220572359325,
|
| 6148 |
-
"grad_norm": 0.47880277037620544,
|
| 6149 |
-
"learning_rate": 6.105939965339045e-05,
|
| 6150 |
-
"loss": 0.3114320755004883,
|
| 6151 |
-
"mean_token_accuracy": 0.9123646855354309,
|
| 6152 |
-
"num_tokens": 21134278.0,
|
| 6153 |
-
"step": 5890
|
| 6154 |
-
},
|
| 6155 |
-
{
|
| 6156 |
-
"entropy": 0.3011189514771104,
|
| 6157 |
-
"epoch": 0.4469781624651982,
|
| 6158 |
-
"grad_norm": 0.42654845118522644,
|
| 6159 |
-
"learning_rate": 6.093972499911127e-05,
|
| 6160 |
-
"loss": 0.3585020065307617,
|
| 6161 |
-
"mean_token_accuracy": 0.9066590443253517,
|
| 6162 |
-
"num_tokens": 21236960.0,
|
| 6163 |
-
"step": 5900
|
| 6164 |
-
},
|
| 6165 |
-
{
|
| 6166 |
-
"entropy": 0.2730740765109658,
|
| 6167 |
-
"epoch": 0.4477357525710714,
|
| 6168 |
-
"grad_norm": 0.5363500714302063,
|
| 6169 |
-
"learning_rate": 6.081998448591257e-05,
|
| 6170 |
-
"loss": 0.3068811893463135,
|
| 6171 |
-
"mean_token_accuracy": 0.9141506284475327,
|
| 6172 |
-
"num_tokens": 21335982.0,
|
| 6173 |
-
"step": 5910
|
| 6174 |
-
},
|
| 6175 |
-
{
|
| 6176 |
-
"entropy": 0.35876688826829195,
|
| 6177 |
-
"epoch": 0.44849334267694463,
|
| 6178 |
-
"grad_norm": 0.3467099368572235,
|
| 6179 |
-
"learning_rate": 6.0700178834651655e-05,
|
| 6180 |
-
"loss": 0.41603832244873046,
|
| 6181 |
-
"mean_token_accuracy": 0.8845574945211411,
|
| 6182 |
-
"num_tokens": 21441445.0,
|
| 6183 |
-
"step": 5920
|
| 6184 |
-
},
|
| 6185 |
-
{
|
| 6186 |
-
"entropy": 0.3085024130530655,
|
| 6187 |
-
"epoch": 0.44925093278281786,
|
| 6188 |
-
"grad_norm": 0.38694334030151367,
|
| 6189 |
-
"learning_rate": 6.0580308766577985e-05,
|
| 6190 |
-
"loss": 0.35320563316345216,
|
| 6191 |
-
"mean_token_accuracy": 0.9032577097415924,
|
| 6192 |
-
"num_tokens": 21549419.0,
|
| 6193 |
-
"step": 5930
|
| 6194 |
-
},
|
| 6195 |
-
{
|
| 6196 |
-
"entropy": 0.3079379880800843,
|
| 6197 |
-
"epoch": 0.4500085228886911,
|
| 6198 |
-
"grad_norm": 0.36321520805358887,
|
| 6199 |
-
"learning_rate": 6.046037500332882e-05,
|
| 6200 |
-
"loss": 0.3737701416015625,
|
| 6201 |
-
"mean_token_accuracy": 0.9004300713539124,
|
| 6202 |
-
"num_tokens": 21656513.0,
|
| 6203 |
-
"step": 5940
|
| 6204 |
-
},
|
| 6205 |
-
{
|
| 6206 |
-
"entropy": 0.29978050384670496,
|
| 6207 |
-
"epoch": 0.4507661129945643,
|
| 6208 |
-
"grad_norm": 0.5048180818557739,
|
| 6209 |
-
"learning_rate": 6.034037826692489e-05,
|
| 6210 |
-
"loss": 0.34405601024627686,
|
| 6211 |
-
"mean_token_accuracy": 0.9022786557674408,
|
| 6212 |
-
"num_tokens": 21759248.0,
|
| 6213 |
-
"step": 5950
|
| 6214 |
-
},
|
| 6215 |
-
{
|
| 6216 |
-
"entropy": 0.2774682804942131,
|
| 6217 |
-
"epoch": 0.4515237031004375,
|
| 6218 |
-
"grad_norm": 0.4303182363510132,
|
| 6219 |
-
"learning_rate": 6.022031927976599e-05,
|
| 6220 |
-
"loss": 0.31653585433959963,
|
| 6221 |
-
"mean_token_accuracy": 0.9109986230731011,
|
| 6222 |
-
"num_tokens": 21857203.0,
|
| 6223 |
-
"step": 5960
|
| 6224 |
-
},
|
| 6225 |
-
{
|
| 6226 |
-
"entropy": 0.28896033111959696,
|
| 6227 |
-
"epoch": 0.4522812932063107,
|
| 6228 |
-
"grad_norm": 0.40719154477119446,
|
| 6229 |
-
"learning_rate": 6.0100198764626723e-05,
|
| 6230 |
-
"loss": 0.3317974090576172,
|
| 6231 |
-
"mean_token_accuracy": 0.9067942887544632,
|
| 6232 |
-
"num_tokens": 21948662.0,
|
| 6233 |
-
"step": 5970
|
| 6234 |
-
},
|
| 6235 |
-
{
|
| 6236 |
-
"entropy": 0.2868704780936241,
|
| 6237 |
-
"epoch": 0.45303888331218395,
|
| 6238 |
-
"grad_norm": 0.36758166551589966,
|
| 6239 |
-
"learning_rate": 5.998001744465207e-05,
|
| 6240 |
-
"loss": 0.31374106407165525,
|
| 6241 |
-
"mean_token_accuracy": 0.9087432786822319,
|
| 6242 |
-
"num_tokens": 22045957.0,
|
| 6243 |
-
"step": 5980
|
| 6244 |
-
},
|
| 6245 |
-
{
|
| 6246 |
-
"entropy": 0.308706765063107,
|
| 6247 |
-
"epoch": 0.4537964734180572,
|
| 6248 |
-
"grad_norm": 0.4636876881122589,
|
| 6249 |
-
"learning_rate": 5.985977604335309e-05,
|
| 6250 |
-
"loss": 0.3620981454849243,
|
| 6251 |
-
"mean_token_accuracy": 0.8988838449120522,
|
| 6252 |
-
"num_tokens": 22151506.0,
|
| 6253 |
-
"step": 5990
|
| 6254 |
-
},
|
| 6255 |
-
{
|
| 6256 |
-
"entropy": 0.29360035676509144,
|
| 6257 |
-
"epoch": 0.45455406352393035,
|
| 6258 |
-
"grad_norm": 0.36570054292678833,
|
| 6259 |
-
"learning_rate": 5.9739475284602506e-05,
|
| 6260 |
-
"loss": 0.3208062171936035,
|
| 6261 |
-
"mean_token_accuracy": 0.9093364477157593,
|
| 6262 |
-
"num_tokens": 22244020.0,
|
| 6263 |
-
"step": 6000
|
| 6264 |
-
},
|
| 6265 |
-
{
|
| 6266 |
-
"epoch": 0.45455406352393035,
|
| 6267 |
-
"eval_entropy": 0.303972092223413,
|
| 6268 |
-
"eval_loss": 0.29795870184898376,
|
| 6269 |
-
"eval_mean_token_accuracy": 0.9040873720404807,
|
| 6270 |
-
"eval_num_tokens": 22244020.0,
|
| 6271 |
-
"eval_runtime": 357.2271,
|
| 6272 |
-
"eval_samples_per_second": 5.974,
|
| 6273 |
-
"eval_steps_per_second": 1.495,
|
| 6274 |
-
"step": 6000
|
| 6275 |
-
},
|
| 6276 |
-
{
|
| 6277 |
-
"entropy": 0.3139019813388586,
|
| 6278 |
-
"epoch": 0.4553116536298036,
|
| 6279 |
-
"grad_norm": 0.43573814630508423,
|
| 6280 |
-
"learning_rate": 5.9619115892630426e-05,
|
| 6281 |
-
"loss": 0.3711949110031128,
|
| 6282 |
-
"mean_token_accuracy": 0.8981863662600518,
|
| 6283 |
-
"num_tokens": 22352642.0,
|
| 6284 |
-
"step": 6010
|
| 6285 |
-
},
|
| 6286 |
-
{
|
| 6287 |
-
"entropy": 0.29842711314558984,
|
| 6288 |
-
"epoch": 0.4560692437356768,
|
| 6289 |
-
"grad_norm": 0.5387157201766968,
|
| 6290 |
-
"learning_rate": 5.949869859201993e-05,
|
| 6291 |
-
"loss": 0.34447512626647947,
|
| 6292 |
-
"mean_token_accuracy": 0.9069076806306839,
|
| 6293 |
-
"num_tokens": 22447438.0,
|
| 6294 |
-
"step": 6020
|
| 6295 |
-
},
|
| 6296 |
-
{
|
| 6297 |
-
"entropy": 0.27448117565363644,
|
| 6298 |
-
"epoch": 0.45682683384155004,
|
| 6299 |
-
"grad_norm": 0.40428847074508667,
|
| 6300 |
-
"learning_rate": 5.937822410770267e-05,
|
| 6301 |
-
"loss": 0.31616170406341554,
|
| 6302 |
-
"mean_token_accuracy": 0.9149715229868889,
|
| 6303 |
-
"num_tokens": 22543080.0,
|
| 6304 |
-
"step": 6030
|
| 6305 |
-
},
|
| 6306 |
-
{
|
| 6307 |
-
"entropy": 0.35422000046819446,
|
| 6308 |
-
"epoch": 0.45758442394742327,
|
| 6309 |
-
"grad_norm": 0.34128761291503906,
|
| 6310 |
-
"learning_rate": 5.9257693164954606e-05,
|
| 6311 |
-
"loss": 0.39284601211547854,
|
| 6312 |
-
"mean_token_accuracy": 0.886053703725338,
|
| 6313 |
-
"num_tokens": 22656243.0,
|
| 6314 |
-
"step": 6040
|
| 6315 |
-
},
|
| 6316 |
-
{
|
| 6317 |
-
"entropy": 0.3126713650301099,
|
| 6318 |
-
"epoch": 0.45834201405329644,
|
| 6319 |
-
"grad_norm": 0.48676598072052,
|
| 6320 |
-
"learning_rate": 5.913710648939158e-05,
|
| 6321 |
-
"loss": 0.3839789867401123,
|
| 6322 |
-
"mean_token_accuracy": 0.8973751902580261,
|
| 6323 |
-
"num_tokens": 22753384.0,
|
| 6324 |
-
"step": 6050
|
| 6325 |
-
},
|
| 6326 |
-
{
|
| 6327 |
-
"entropy": 0.34667299371212723,
|
| 6328 |
-
"epoch": 0.45909960415916967,
|
| 6329 |
-
"grad_norm": 0.43320995569229126,
|
| 6330 |
-
"learning_rate": 5.9016464806964936e-05,
|
| 6331 |
-
"loss": 0.41129021644592284,
|
| 6332 |
-
"mean_token_accuracy": 0.890424282848835,
|
| 6333 |
-
"num_tokens": 22865202.0,
|
| 6334 |
-
"step": 6060
|
| 6335 |
-
},
|
| 6336 |
-
{
|
| 6337 |
-
"entropy": 0.3049141285941005,
|
| 6338 |
-
"epoch": 0.4598571942650429,
|
| 6339 |
-
"grad_norm": 0.33471742272377014,
|
| 6340 |
-
"learning_rate": 5.889576884395718e-05,
|
| 6341 |
-
"loss": 0.35721607208251954,
|
| 6342 |
-
"mean_token_accuracy": 0.9060751304030419,
|
| 6343 |
-
"num_tokens": 22965482.0,
|
| 6344 |
-
"step": 6070
|
| 6345 |
-
},
|
| 6346 |
-
{
|
| 6347 |
-
"entropy": 0.27999355345964433,
|
| 6348 |
-
"epoch": 0.46061478437091613,
|
| 6349 |
-
"grad_norm": 0.46490412950515747,
|
| 6350 |
-
"learning_rate": 5.877501932697758e-05,
|
| 6351 |
-
"loss": 0.31168513298034667,
|
| 6352 |
-
"mean_token_accuracy": 0.9126041397452355,
|
| 6353 |
-
"num_tokens": 23053319.0,
|
| 6354 |
-
"step": 6080
|
| 6355 |
-
},
|
| 6356 |
-
{
|
| 6357 |
-
"entropy": 0.2967614172026515,
|
| 6358 |
-
"epoch": 0.46137237447678936,
|
| 6359 |
-
"grad_norm": 0.52817302942276,
|
| 6360 |
-
"learning_rate": 5.865421698295783e-05,
|
| 6361 |
-
"loss": 0.3302607536315918,
|
| 6362 |
-
"mean_token_accuracy": 0.9083400264382362,
|
| 6363 |
-
"num_tokens": 23159163.0,
|
| 6364 |
-
"step": 6090
|
| 6365 |
-
},
|
| 6366 |
-
{
|
| 6367 |
-
"entropy": 0.29451429694890974,
|
| 6368 |
-
"epoch": 0.46212996458266253,
|
| 6369 |
-
"grad_norm": 0.44573232531547546,
|
| 6370 |
-
"learning_rate": 5.853336253914767e-05,
|
| 6371 |
-
"loss": 0.3604841709136963,
|
| 6372 |
-
"mean_token_accuracy": 0.9084043517708779,
|
| 6373 |
-
"num_tokens": 23255157.0,
|
| 6374 |
-
"step": 6100
|
| 6375 |
-
},
|
| 6376 |
-
{
|
| 6377 |
-
"entropy": 0.3307742692530155,
|
| 6378 |
-
"epoch": 0.46288755468853576,
|
| 6379 |
-
"grad_norm": 0.40377649664878845,
|
| 6380 |
-
"learning_rate": 5.841245672311042e-05,
|
| 6381 |
-
"loss": 0.3890522480010986,
|
| 6382 |
-
"mean_token_accuracy": 0.8916374981403351,
|
| 6383 |
-
"num_tokens": 23366492.0,
|
| 6384 |
-
"step": 6110
|
| 6385 |
-
},
|
| 6386 |
-
{
|
| 6387 |
-
"entropy": 0.3169036017730832,
|
| 6388 |
-
"epoch": 0.463645144794409,
|
| 6389 |
-
"grad_norm": 0.4411364495754242,
|
| 6390 |
-
"learning_rate": 5.829150026271871e-05,
|
| 6391 |
-
"loss": 0.34163494110107423,
|
| 6392 |
-
"mean_token_accuracy": 0.9004677504301071,
|
| 6393 |
-
"num_tokens": 23467234.0,
|
| 6394 |
-
"step": 6120
|
| 6395 |
-
},
|
| 6396 |
-
{
|
| 6397 |
-
"entropy": 0.2998361073434353,
|
| 6398 |
-
"epoch": 0.4644027349002822,
|
| 6399 |
-
"grad_norm": 0.4793619215488434,
|
| 6400 |
-
"learning_rate": 5.817049388615009e-05,
|
| 6401 |
-
"loss": 0.3488142967224121,
|
| 6402 |
-
"mean_token_accuracy": 0.9044462025165558,
|
| 6403 |
-
"num_tokens": 23565694.0,
|
| 6404 |
-
"step": 6130
|
| 6405 |
-
},
|
| 6406 |
-
{
|
| 6407 |
-
"entropy": 0.27653252333402634,
|
| 6408 |
-
"epoch": 0.4651603250061554,
|
| 6409 |
-
"grad_norm": 0.4835834205150604,
|
| 6410 |
-
"learning_rate": 5.804943832188255e-05,
|
| 6411 |
-
"loss": 0.3205927848815918,
|
| 6412 |
-
"mean_token_accuracy": 0.9117123991250992,
|
| 6413 |
-
"num_tokens": 23656208.0,
|
| 6414 |
-
"step": 6140
|
| 6415 |
-
},
|
| 6416 |
-
{
|
| 6417 |
-
"entropy": 0.3110093142837286,
|
| 6418 |
-
"epoch": 0.4659179151120286,
|
| 6419 |
-
"grad_norm": 0.6018494367599487,
|
| 6420 |
-
"learning_rate": 5.792833429869026e-05,
|
| 6421 |
-
"loss": 0.3660226821899414,
|
| 6422 |
-
"mean_token_accuracy": 0.9011919766664505,
|
| 6423 |
-
"num_tokens": 23757247.0,
|
| 6424 |
-
"step": 6150
|
| 6425 |
-
},
|
| 6426 |
-
{
|
| 6427 |
-
"entropy": 0.2761795477941632,
|
| 6428 |
-
"epoch": 0.46667550521790185,
|
| 6429 |
-
"grad_norm": 0.4015946388244629,
|
| 6430 |
-
"learning_rate": 5.780718254563905e-05,
|
| 6431 |
-
"loss": 0.32297894954681394,
|
| 6432 |
-
"mean_token_accuracy": 0.9118266269564629,
|
| 6433 |
-
"num_tokens": 23852560.0,
|
| 6434 |
-
"step": 6160
|
| 6435 |
-
},
|
| 6436 |
-
{
|
| 6437 |
-
"entropy": 0.3156697480008006,
|
| 6438 |
-
"epoch": 0.4674330953237751,
|
| 6439 |
-
"grad_norm": 0.5316373705863953,
|
| 6440 |
-
"learning_rate": 5.7685983792082176e-05,
|
| 6441 |
-
"loss": 0.36042113304138185,
|
| 6442 |
-
"mean_token_accuracy": 0.900458799302578,
|
| 6443 |
-
"num_tokens": 23951295.0,
|
| 6444 |
-
"step": 6170
|
| 6445 |
-
},
|
| 6446 |
-
{
|
| 6447 |
-
"entropy": 0.2707783591002226,
|
| 6448 |
-
"epoch": 0.4681906854296483,
|
| 6449 |
-
"grad_norm": 0.48289015889167786,
|
| 6450 |
-
"learning_rate": 5.756473876765576e-05,
|
| 6451 |
-
"loss": 0.33272371292114256,
|
| 6452 |
-
"mean_token_accuracy": 0.9146774232387542,
|
| 6453 |
-
"num_tokens": 24039675.0,
|
| 6454 |
-
"step": 6180
|
| 6455 |
-
},
|
| 6456 |
-
{
|
| 6457 |
-
"entropy": 0.26940905805677173,
|
| 6458 |
-
"epoch": 0.4689482755355215,
|
| 6459 |
-
"grad_norm": 0.33979037404060364,
|
| 6460 |
-
"learning_rate": 5.744344820227454e-05,
|
| 6461 |
-
"loss": 0.29962873458862305,
|
| 6462 |
-
"mean_token_accuracy": 0.9140648379921913,
|
| 6463 |
-
"num_tokens": 24133037.0,
|
| 6464 |
-
"step": 6190
|
| 6465 |
-
},
|
| 6466 |
-
{
|
| 6467 |
-
"entropy": 0.3069524383172393,
|
| 6468 |
-
"epoch": 0.4697058656413947,
|
| 6469 |
-
"grad_norm": 0.48873454332351685,
|
| 6470 |
-
"learning_rate": 5.7322112826127406e-05,
|
| 6471 |
-
"loss": 0.3563365936279297,
|
| 6472 |
-
"mean_token_accuracy": 0.9023343905806541,
|
| 6473 |
-
"num_tokens": 24227006.0,
|
| 6474 |
-
"step": 6200
|
| 6475 |
-
},
|
| 6476 |
-
{
|
| 6477 |
-
"entropy": 0.29299616999924183,
|
| 6478 |
-
"epoch": 0.47046345574726794,
|
| 6479 |
-
"grad_norm": 0.3211856186389923,
|
| 6480 |
-
"learning_rate": 5.720073336967299e-05,
|
| 6481 |
-
"loss": 0.34385735988616944,
|
| 6482 |
-
"mean_token_accuracy": 0.9041966706514358,
|
| 6483 |
-
"num_tokens": 24327519.0,
|
| 6484 |
-
"step": 6210
|
| 6485 |
-
},
|
| 6486 |
-
{
|
| 6487 |
-
"entropy": 0.3108357986435294,
|
| 6488 |
-
"epoch": 0.4712210458531412,
|
| 6489 |
-
"grad_norm": 0.5044354796409607,
|
| 6490 |
-
"learning_rate": 5.7079310563635335e-05,
|
| 6491 |
-
"loss": 0.3258206367492676,
|
| 6492 |
-
"mean_token_accuracy": 0.9025010168552399,
|
| 6493 |
-
"num_tokens": 24426364.0,
|
| 6494 |
-
"step": 6220
|
| 6495 |
-
},
|
| 6496 |
-
{
|
| 6497 |
-
"entropy": 0.29107468016445637,
|
| 6498 |
-
"epoch": 0.47197863595901435,
|
| 6499 |
-
"grad_norm": 0.5551505088806152,
|
| 6500 |
-
"learning_rate": 5.695784513899942e-05,
|
| 6501 |
-
"loss": 0.3395212173461914,
|
| 6502 |
-
"mean_token_accuracy": 0.9079627841711044,
|
| 6503 |
-
"num_tokens": 24523309.0,
|
| 6504 |
-
"step": 6230
|
| 6505 |
-
},
|
| 6506 |
-
{
|
| 6507 |
-
"entropy": 0.3183888182044029,
|
| 6508 |
-
"epoch": 0.4727362260648876,
|
| 6509 |
-
"grad_norm": 0.40023085474967957,
|
| 6510 |
-
"learning_rate": 5.6836337827006816e-05,
|
| 6511 |
-
"loss": 0.4081729412078857,
|
| 6512 |
-
"mean_token_accuracy": 0.8955054402351379,
|
| 6513 |
-
"num_tokens": 24621479.0,
|
| 6514 |
-
"step": 6240
|
| 6515 |
-
},
|
| 6516 |
-
{
|
| 6517 |
-
"entropy": 0.31882689893245697,
|
| 6518 |
-
"epoch": 0.4734938161707608,
|
| 6519 |
-
"grad_norm": 0.40556657314300537,
|
| 6520 |
-
"learning_rate": 5.6714789359151264e-05,
|
| 6521 |
-
"loss": 0.36399688720703127,
|
| 6522 |
-
"mean_token_accuracy": 0.8994935199618339,
|
| 6523 |
-
"num_tokens": 24726832.0,
|
| 6524 |
-
"step": 6250
|
| 6525 |
-
},
|
| 6526 |
-
{
|
| 6527 |
-
"epoch": 0.4734938161707608,
|
| 6528 |
-
"eval_entropy": 0.303136145521639,
|
| 6529 |
-
"eval_loss": 0.29779738187789917,
|
| 6530 |
-
"eval_mean_token_accuracy": 0.904087961166539,
|
| 6531 |
-
"eval_num_tokens": 24726832.0,
|
| 6532 |
-
"eval_runtime": 357.2815,
|
| 6533 |
-
"eval_samples_per_second": 5.973,
|
| 6534 |
-
"eval_steps_per_second": 1.495,
|
| 6535 |
-
"step": 6250
|
| 6536 |
-
},
|
| 6537 |
-
{
|
| 6538 |
-
"entropy": 0.27081307470798494,
|
| 6539 |
-
"epoch": 0.47425140627663404,
|
| 6540 |
-
"grad_norm": 0.5495783090591431,
|
| 6541 |
-
"learning_rate": 5.6593200467174245e-05,
|
| 6542 |
-
"loss": 0.31279902458190917,
|
| 6543 |
-
"mean_token_accuracy": 0.9163315385580063,
|
| 6544 |
-
"num_tokens": 24824021.0,
|
| 6545 |
-
"step": 6260
|
| 6546 |
-
},
|
| 6547 |
-
{
|
| 6548 |
-
"entropy": 0.27001900840550663,
|
| 6549 |
-
"epoch": 0.47500899638250726,
|
| 6550 |
-
"grad_norm": 0.3735206425189972,
|
| 6551 |
-
"learning_rate": 5.647157188306064e-05,
|
| 6552 |
-
"loss": 0.3338909149169922,
|
| 6553 |
-
"mean_token_accuracy": 0.9135658860206604,
|
| 6554 |
-
"num_tokens": 24919833.0,
|
| 6555 |
-
"step": 6270
|
| 6556 |
-
},
|
| 6557 |
-
{
|
| 6558 |
-
"entropy": 0.32687586909160016,
|
| 6559 |
-
"epoch": 0.47576658648838044,
|
| 6560 |
-
"grad_norm": 0.3312423527240753,
|
| 6561 |
-
"learning_rate": 5.634990433903423e-05,
|
| 6562 |
-
"loss": 0.35171685218811033,
|
| 6563 |
-
"mean_token_accuracy": 0.8957670792937279,
|
| 6564 |
-
"num_tokens": 25032296.0,
|
| 6565 |
-
"step": 6280
|
| 6566 |
-
},
|
| 6567 |
-
{
|
| 6568 |
-
"entropy": 0.3096007126383483,
|
| 6569 |
-
"epoch": 0.47652417659425367,
|
| 6570 |
-
"grad_norm": 0.3826523721218109,
|
| 6571 |
-
"learning_rate": 5.622819856755339e-05,
|
| 6572 |
-
"loss": 0.3587393283843994,
|
| 6573 |
-
"mean_token_accuracy": 0.9020399078726768,
|
| 6574 |
-
"num_tokens": 25128729.0,
|
| 6575 |
-
"step": 6290
|
| 6576 |
-
},
|
| 6577 |
-
{
|
| 6578 |
-
"entropy": 0.2763217631727457,
|
| 6579 |
-
"epoch": 0.4772817667001269,
|
| 6580 |
-
"grad_norm": 0.48616793751716614,
|
| 6581 |
-
"learning_rate": 5.610645530130661e-05,
|
| 6582 |
-
"loss": 0.3158841609954834,
|
| 6583 |
-
"mean_token_accuracy": 0.9111469686031342,
|
| 6584 |
-
"num_tokens": 25221459.0,
|
| 6585 |
-
"step": 6300
|
| 6586 |
-
},
|
| 6587 |
-
{
|
| 6588 |
-
"entropy": 0.32194995488971473,
|
| 6589 |
-
"epoch": 0.4780393568060001,
|
| 6590 |
-
"grad_norm": 0.42415404319763184,
|
| 6591 |
-
"learning_rate": 5.5984675273208124e-05,
|
| 6592 |
-
"loss": 0.38160552978515627,
|
| 6593 |
-
"mean_token_accuracy": 0.8982667028903961,
|
| 6594 |
-
"num_tokens": 25321914.0,
|
| 6595 |
-
"step": 6310
|
| 6596 |
-
},
|
| 6597 |
-
{
|
| 6598 |
-
"entropy": 0.2592316344380379,
|
| 6599 |
-
"epoch": 0.47879694691187336,
|
| 6600 |
-
"grad_norm": 0.583716630935669,
|
| 6601 |
-
"learning_rate": 5.586285921639343e-05,
|
| 6602 |
-
"loss": 0.30498790740966797,
|
| 6603 |
-
"mean_token_accuracy": 0.9196190744638443,
|
| 6604 |
-
"num_tokens": 25408813.0,
|
| 6605 |
-
"step": 6320
|
| 6606 |
-
},
|
| 6607 |
-
{
|
| 6608 |
-
"entropy": 0.3198434682562947,
|
| 6609 |
-
"epoch": 0.47955453701774653,
|
| 6610 |
-
"grad_norm": 0.47471076250076294,
|
| 6611 |
-
"learning_rate": 5.5741007864214976e-05,
|
| 6612 |
-
"loss": 0.36321723461151123,
|
| 6613 |
-
"mean_token_accuracy": 0.8998541802167892,
|
| 6614 |
-
"num_tokens": 25510039.0,
|
| 6615 |
-
"step": 6330
|
| 6616 |
-
},
|
| 6617 |
-
{
|
| 6618 |
-
"entropy": 0.33730865847319363,
|
| 6619 |
-
"epoch": 0.48031212712361976,
|
| 6620 |
-
"grad_norm": 0.4704512357711792,
|
| 6621 |
-
"learning_rate": 5.5619121950237666e-05,
|
| 6622 |
-
"loss": 0.4000189781188965,
|
| 6623 |
-
"mean_token_accuracy": 0.8912207677960395,
|
| 6624 |
-
"num_tokens": 25612365.0,
|
| 6625 |
-
"step": 6340
|
| 6626 |
-
},
|
| 6627 |
-
{
|
| 6628 |
-
"entropy": 0.3006171802058816,
|
| 6629 |
-
"epoch": 0.481069717229493,
|
| 6630 |
-
"grad_norm": 0.41279882192611694,
|
| 6631 |
-
"learning_rate": 5.549720220823448e-05,
|
| 6632 |
-
"loss": 0.3377386569976807,
|
| 6633 |
-
"mean_token_accuracy": 0.9025795921683312,
|
| 6634 |
-
"num_tokens": 25719555.0,
|
| 6635 |
-
"step": 6350
|
| 6636 |
-
},
|
| 6637 |
-
{
|
| 6638 |
-
"entropy": 0.3255089933052659,
|
| 6639 |
-
"epoch": 0.4818273073353662,
|
| 6640 |
-
"grad_norm": 0.5086070895195007,
|
| 6641 |
-
"learning_rate": 5.5375249372182034e-05,
|
| 6642 |
-
"loss": 0.35865049362182616,
|
| 6643 |
-
"mean_token_accuracy": 0.8984927222132683,
|
| 6644 |
-
"num_tokens": 25823771.0,
|
| 6645 |
-
"step": 6360
|
| 6646 |
-
},
|
| 6647 |
-
{
|
| 6648 |
-
"entropy": 0.3123222546651959,
|
| 6649 |
-
"epoch": 0.4825848974412394,
|
| 6650 |
-
"grad_norm": 0.5222538709640503,
|
| 6651 |
-
"learning_rate": 5.52532641762562e-05,
|
| 6652 |
-
"loss": 0.37432212829589845,
|
| 6653 |
-
"mean_token_accuracy": 0.9010790154337883,
|
| 6654 |
-
"num_tokens": 25922754.0,
|
| 6655 |
-
"step": 6370
|
| 6656 |
-
},
|
| 6657 |
-
{
|
| 6658 |
-
"entropy": 0.3135323330760002,
|
| 6659 |
-
"epoch": 0.4833424875471126,
|
| 6660 |
-
"grad_norm": 0.45767465233802795,
|
| 6661 |
-
"learning_rate": 5.5131247354827665e-05,
|
| 6662 |
-
"loss": 0.3751972198486328,
|
| 6663 |
-
"mean_token_accuracy": 0.9013716772198677,
|
| 6664 |
-
"num_tokens": 26022646.0,
|
| 6665 |
-
"step": 6380
|
| 6666 |
-
},
|
| 6667 |
-
{
|
| 6668 |
-
"entropy": 0.32033412903547287,
|
| 6669 |
-
"epoch": 0.48410007765298585,
|
| 6670 |
-
"grad_norm": 0.40000393986701965,
|
| 6671 |
-
"learning_rate": 5.500919964245747e-05,
|
| 6672 |
-
"loss": 0.3764453887939453,
|
| 6673 |
-
"mean_token_accuracy": 0.8962203413248062,
|
| 6674 |
-
"num_tokens": 26130616.0,
|
| 6675 |
-
"step": 6390
|
| 6676 |
-
},
|
| 6677 |
-
{
|
| 6678 |
-
"entropy": 0.32832188382744787,
|
| 6679 |
-
"epoch": 0.4848576677588591,
|
| 6680 |
-
"grad_norm": 0.8431655764579773,
|
| 6681 |
-
"learning_rate": 5.488712177389267e-05,
|
| 6682 |
-
"loss": 0.3589762210845947,
|
| 6683 |
-
"mean_token_accuracy": 0.8948887467384339,
|
| 6684 |
-
"num_tokens": 26228494.0,
|
| 6685 |
-
"step": 6400
|
| 6686 |
-
},
|
| 6687 |
-
{
|
| 6688 |
-
"entropy": 0.3090921881608665,
|
| 6689 |
-
"epoch": 0.4856152578647323,
|
| 6690 |
-
"grad_norm": 0.5108557343482971,
|
| 6691 |
-
"learning_rate": 5.4765014484061806e-05,
|
| 6692 |
-
"loss": 0.36724700927734377,
|
| 6693 |
-
"mean_token_accuracy": 0.9016465470194817,
|
| 6694 |
-
"num_tokens": 26330015.0,
|
| 6695 |
-
"step": 6410
|
| 6696 |
-
},
|
| 6697 |
-
{
|
| 6698 |
-
"entropy": 0.2968562699854374,
|
| 6699 |
-
"epoch": 0.4863728479706055,
|
| 6700 |
-
"grad_norm": 0.3894047737121582,
|
| 6701 |
-
"learning_rate": 5.4642878508070594e-05,
|
| 6702 |
-
"loss": 0.35214571952819823,
|
| 6703 |
-
"mean_token_accuracy": 0.9037809073925018,
|
| 6704 |
-
"num_tokens": 26428167.0,
|
| 6705 |
-
"step": 6420
|
| 6706 |
-
},
|
| 6707 |
-
{
|
| 6708 |
-
"entropy": 0.31641832515597346,
|
| 6709 |
-
"epoch": 0.4871304380764787,
|
| 6710 |
-
"grad_norm": 0.38789114356040955,
|
| 6711 |
-
"learning_rate": 5.452071458119744e-05,
|
| 6712 |
-
"loss": 0.3768194913864136,
|
| 6713 |
-
"mean_token_accuracy": 0.9009707912802696,
|
| 6714 |
-
"num_tokens": 26530879.0,
|
| 6715 |
-
"step": 6430
|
| 6716 |
-
},
|
| 6717 |
-
{
|
| 6718 |
-
"entropy": 0.3176609221845865,
|
| 6719 |
-
"epoch": 0.48788802818235194,
|
| 6720 |
-
"grad_norm": 0.3938576579093933,
|
| 6721 |
-
"learning_rate": 5.439852343888899e-05,
|
| 6722 |
-
"loss": 0.3701298952102661,
|
| 6723 |
-
"mean_token_accuracy": 0.9001378238201141,
|
| 6724 |
-
"num_tokens": 26625399.0,
|
| 6725 |
-
"step": 6440
|
| 6726 |
-
},
|
| 6727 |
-
{
|
| 6728 |
-
"entropy": 0.34214024171233176,
|
| 6729 |
-
"epoch": 0.48864561828822517,
|
| 6730 |
-
"grad_norm": 0.4817556142807007,
|
| 6731 |
-
"learning_rate": 5.427630581675575e-05,
|
| 6732 |
-
"loss": 0.375986385345459,
|
| 6733 |
-
"mean_token_accuracy": 0.8895856380462647,
|
| 6734 |
-
"num_tokens": 26722870.0,
|
| 6735 |
-
"step": 6450
|
| 6736 |
-
},
|
| 6737 |
-
{
|
| 6738 |
-
"entropy": 0.2940156176686287,
|
| 6739 |
-
"epoch": 0.4894032083940984,
|
| 6740 |
-
"grad_norm": 0.412383109331131,
|
| 6741 |
-
"learning_rate": 5.4154062450567636e-05,
|
| 6742 |
-
"loss": 0.34354362487792967,
|
| 6743 |
-
"mean_token_accuracy": 0.9055873632431031,
|
| 6744 |
-
"num_tokens": 26815947.0,
|
| 6745 |
-
"step": 6460
|
| 6746 |
-
},
|
| 6747 |
-
{
|
| 6748 |
-
"entropy": 0.28387089930474757,
|
| 6749 |
-
"epoch": 0.4901607984999716,
|
| 6750 |
-
"grad_norm": 0.43847861886024475,
|
| 6751 |
-
"learning_rate": 5.4031794076249544e-05,
|
| 6752 |
-
"loss": 0.3144033908843994,
|
| 6753 |
-
"mean_token_accuracy": 0.9105543494224548,
|
| 6754 |
-
"num_tokens": 26916068.0,
|
| 6755 |
-
"step": 6470
|
| 6756 |
-
},
|
| 6757 |
-
{
|
| 6758 |
-
"entropy": 0.3208195770159364,
|
| 6759 |
-
"epoch": 0.4909183886058448,
|
| 6760 |
-
"grad_norm": 0.43610596656799316,
|
| 6761 |
-
"learning_rate": 5.390950142987694e-05,
|
| 6762 |
-
"loss": 0.37391247749328616,
|
| 6763 |
-
"mean_token_accuracy": 0.8961284562945366,
|
| 6764 |
-
"num_tokens": 27020540.0,
|
| 6765 |
-
"step": 6480
|
| 6766 |
-
},
|
| 6767 |
-
{
|
| 6768 |
-
"entropy": 0.2713425327092409,
|
| 6769 |
-
"epoch": 0.49167597871171803,
|
| 6770 |
-
"grad_norm": 0.6305099129676819,
|
| 6771 |
-
"learning_rate": 5.378718524767139e-05,
|
| 6772 |
-
"loss": 0.3020663261413574,
|
| 6773 |
-
"mean_token_accuracy": 0.9132449522614479,
|
| 6774 |
-
"num_tokens": 27117277.0,
|
| 6775 |
-
"step": 6490
|
| 6776 |
-
},
|
| 6777 |
-
{
|
| 6778 |
-
"entropy": 0.26218025628477337,
|
| 6779 |
-
"epoch": 0.49243356881759126,
|
| 6780 |
-
"grad_norm": 0.5174180269241333,
|
| 6781 |
-
"learning_rate": 5.366484626599615e-05,
|
| 6782 |
-
"loss": 0.30744128227233886,
|
| 6783 |
-
"mean_token_accuracy": 0.9163722321391106,
|
| 6784 |
-
"num_tokens": 27205259.0,
|
| 6785 |
-
"step": 6500
|
| 6786 |
-
},
|
| 6787 |
-
{
|
| 6788 |
-
"epoch": 0.49243356881759126,
|
| 6789 |
-
"eval_entropy": 0.30084027919206724,
|
| 6790 |
-
"eval_loss": 0.2969069182872772,
|
| 6791 |
-
"eval_mean_token_accuracy": 0.9041626769281952,
|
| 6792 |
-
"eval_num_tokens": 27205259.0,
|
| 6793 |
-
"eval_runtime": 357.728,
|
| 6794 |
-
"eval_samples_per_second": 5.965,
|
| 6795 |
-
"eval_steps_per_second": 1.493,
|
| 6796 |
-
"step": 6500
|
| 6797 |
-
},
|
| 6798 |
-
{
|
| 6799 |
-
"entropy": 0.2952243184670806,
|
| 6800 |
-
"epoch": 0.49319115892346443,
|
| 6801 |
-
"grad_norm": 0.5314688086509705,
|
| 6802 |
-
"learning_rate": 5.3542485221351735e-05,
|
| 6803 |
-
"loss": 0.3479542970657349,
|
| 6804 |
-
"mean_token_accuracy": 0.9054689884185791,
|
| 6805 |
-
"num_tokens": 27302832.0,
|
| 6806 |
-
"step": 6510
|
| 6807 |
-
},
|
| 6808 |
-
{
|
| 6809 |
-
"entropy": 0.27767247427254915,
|
| 6810 |
-
"epoch": 0.49394874902933766,
|
| 6811 |
-
"grad_norm": 0.45793017745018005,
|
| 6812 |
-
"learning_rate": 5.34201028503715e-05,
|
| 6813 |
-
"loss": 0.32643852233886717,
|
| 6814 |
-
"mean_token_accuracy": 0.911968544125557,
|
| 6815 |
-
"num_tokens": 27398344.0,
|
| 6816 |
-
"step": 6520
|
| 6817 |
-
},
|
| 6818 |
-
{
|
| 6819 |
-
"entropy": 0.2605927696451545,
|
| 6820 |
-
"epoch": 0.4947063391352109,
|
| 6821 |
-
"grad_norm": 0.4267292618751526,
|
| 6822 |
-
"learning_rate": 5.3297699889817185e-05,
|
| 6823 |
-
"loss": 0.3160386085510254,
|
| 6824 |
-
"mean_token_accuracy": 0.917172135412693,
|
| 6825 |
-
"num_tokens": 27494240.0,
|
| 6826 |
-
"step": 6530
|
| 6827 |
-
},
|
| 6828 |
-
{
|
| 6829 |
-
"entropy": 0.32263444885611536,
|
| 6830 |
-
"epoch": 0.4954639292410841,
|
| 6831 |
-
"grad_norm": 0.38362303376197815,
|
| 6832 |
-
"learning_rate": 5.317527707657446e-05,
|
| 6833 |
-
"loss": 0.3594916343688965,
|
| 6834 |
-
"mean_token_accuracy": 0.8953352093696594,
|
| 6835 |
-
"num_tokens": 27595887.0,
|
| 6836 |
-
"step": 6540
|
| 6837 |
-
},
|
| 6838 |
-
{
|
| 6839 |
-
"entropy": 0.3229256052523851,
|
| 6840 |
-
"epoch": 0.49622151934695735,
|
| 6841 |
-
"grad_norm": 0.3866725564002991,
|
| 6842 |
-
"learning_rate": 5.305283514764852e-05,
|
| 6843 |
-
"loss": 0.3599198579788208,
|
| 6844 |
-
"mean_token_accuracy": 0.8966040164232254,
|
| 6845 |
-
"num_tokens": 27701478.0,
|
| 6846 |
-
"step": 6550
|
| 6847 |
-
},
|
| 6848 |
-
{
|
| 6849 |
-
"entropy": 0.2938591906800866,
|
| 6850 |
-
"epoch": 0.4969791094528305,
|
| 6851 |
-
"grad_norm": 0.44755950570106506,
|
| 6852 |
-
"learning_rate": 5.293037484015966e-05,
|
| 6853 |
-
"loss": 0.3460268974304199,
|
| 6854 |
-
"mean_token_accuracy": 0.9055036634206772,
|
| 6855 |
-
"num_tokens": 27803617.0,
|
| 6856 |
-
"step": 6560
|
| 6857 |
-
},
|
| 6858 |
-
{
|
| 6859 |
-
"entropy": 0.2707622142508626,
|
| 6860 |
-
"epoch": 0.49773669955870375,
|
| 6861 |
-
"grad_norm": 0.3887022137641907,
|
| 6862 |
-
"learning_rate": 5.280789689133878e-05,
|
| 6863 |
-
"loss": 0.3057152509689331,
|
| 6864 |
-
"mean_token_accuracy": 0.9182133570313453,
|
| 6865 |
-
"num_tokens": 27902014.0,
|
| 6866 |
-
"step": 6570
|
| 6867 |
-
},
|
| 6868 |
-
{
|
| 6869 |
-
"entropy": 0.3023366378620267,
|
| 6870 |
-
"epoch": 0.498494289664577,
|
| 6871 |
-
"grad_norm": 0.43178898096084595,
|
| 6872 |
-
"learning_rate": 5.268540203852301e-05,
|
| 6873 |
-
"loss": 0.33755409717559814,
|
| 6874 |
-
"mean_token_accuracy": 0.9019709035754204,
|
| 6875 |
-
"num_tokens": 27997603.0,
|
| 6876 |
-
"step": 6580
|
| 6877 |
-
},
|
| 6878 |
-
{
|
| 6879 |
-
"entropy": 0.25826080329716206,
|
| 6880 |
-
"epoch": 0.4992518797704502,
|
| 6881 |
-
"grad_norm": 0.5452413558959961,
|
| 6882 |
-
"learning_rate": 5.256289101915126e-05,
|
| 6883 |
-
"loss": 0.2897059440612793,
|
| 6884 |
-
"mean_token_accuracy": 0.9177193090319633,
|
| 6885 |
-
"num_tokens": 28091115.0,
|
| 6886 |
-
"step": 6590
|
| 6887 |
-
},
|
| 6888 |
-
{
|
| 6889 |
-
"entropy": 0.25413902197033167,
|
| 6890 |
-
"epoch": 0.5000094698763234,
|
| 6891 |
-
"grad_norm": 0.4130087196826935,
|
| 6892 |
-
"learning_rate": 5.244036457075972e-05,
|
| 6893 |
-
"loss": 0.2878809928894043,
|
| 6894 |
-
"mean_token_accuracy": 0.9205578058958054,
|
| 6895 |
-
"num_tokens": 28188707.0,
|
| 6896 |
-
"step": 6600
|
| 6897 |
-
},
|
| 6898 |
-
{
|
| 6899 |
-
"entropy": 0.32136391047388313,
|
| 6900 |
-
"epoch": 0.5007670599821966,
|
| 6901 |
-
"grad_norm": 0.343299001455307,
|
| 6902 |
-
"learning_rate": 5.231782343097751e-05,
|
| 6903 |
-
"loss": 0.37444486618041994,
|
| 6904 |
-
"mean_token_accuracy": 0.8931363686919213,
|
| 6905 |
-
"num_tokens": 28298814.0,
|
| 6906 |
-
"step": 6610
|
| 6907 |
-
},
|
| 6908 |
-
{
|
| 6909 |
-
"entropy": 0.2971101008355618,
|
| 6910 |
-
"epoch": 0.5015246500880699,
|
| 6911 |
-
"grad_norm": 0.41217291355133057,
|
| 6912 |
-
"learning_rate": 5.219526833752215e-05,
|
| 6913 |
-
"loss": 0.34679837226867677,
|
| 6914 |
-
"mean_token_accuracy": 0.9051818385720253,
|
| 6915 |
-
"num_tokens": 28395681.0,
|
| 6916 |
-
"step": 6620
|
| 6917 |
-
},
|
| 6918 |
-
{
|
| 6919 |
-
"entropy": 0.32872276240959764,
|
| 6920 |
-
"epoch": 0.5022822401939431,
|
| 6921 |
-
"grad_norm": 0.3719559907913208,
|
| 6922 |
-
"learning_rate": 5.2072700028195206e-05,
|
| 6923 |
-
"loss": 0.36319718360900877,
|
| 6924 |
-
"mean_token_accuracy": 0.8953921139240265,
|
| 6925 |
-
"num_tokens": 28504482.0,
|
| 6926 |
-
"step": 6630
|
| 6927 |
-
},
|
| 6928 |
-
{
|
| 6929 |
-
"entropy": 0.31123167704790833,
|
| 6930 |
-
"epoch": 0.5030398302998162,
|
| 6931 |
-
"grad_norm": 0.5293285250663757,
|
| 6932 |
-
"learning_rate": 5.1950119240877794e-05,
|
| 6933 |
-
"loss": 0.37692461013793943,
|
| 6934 |
-
"mean_token_accuracy": 0.8991176769137382,
|
| 6935 |
-
"num_tokens": 28605358.0,
|
| 6936 |
-
"step": 6640
|
| 6937 |
-
},
|
| 6938 |
-
{
|
| 6939 |
-
"entropy": 0.2843796066939831,
|
| 6940 |
-
"epoch": 0.5037974204056895,
|
| 6941 |
-
"grad_norm": 0.3747865557670593,
|
| 6942 |
-
"learning_rate": 5.182752671352613e-05,
|
| 6943 |
-
"loss": 0.3391988754272461,
|
| 6944 |
-
"mean_token_accuracy": 0.9110558673739433,
|
| 6945 |
-
"num_tokens": 28709879.0,
|
| 6946 |
-
"step": 6650
|
| 6947 |
-
},
|
| 6948 |
-
{
|
| 6949 |
-
"entropy": 0.2728314265608788,
|
| 6950 |
-
"epoch": 0.5045550105115627,
|
| 6951 |
-
"grad_norm": 0.6186447739601135,
|
| 6952 |
-
"learning_rate": 5.170492318416712e-05,
|
| 6953 |
-
"loss": 0.3346660375595093,
|
| 6954 |
-
"mean_token_accuracy": 0.9131002396345138,
|
| 6955 |
-
"num_tokens": 28798336.0,
|
| 6956 |
-
"step": 6660
|
| 6957 |
-
},
|
| 6958 |
-
{
|
| 6959 |
-
"entropy": 0.299285851418972,
|
| 6960 |
-
"epoch": 0.5053126006174359,
|
| 6961 |
-
"grad_norm": 0.4465583264827728,
|
| 6962 |
-
"learning_rate": 5.158230939089392e-05,
|
| 6963 |
-
"loss": 0.3480525493621826,
|
| 6964 |
-
"mean_token_accuracy": 0.9050300002098084,
|
| 6965 |
-
"num_tokens": 28891167.0,
|
| 6966 |
-
"step": 6670
|
| 6967 |
-
},
|
| 6968 |
-
{
|
| 6969 |
-
"entropy": 0.34458486586809156,
|
| 6970 |
-
"epoch": 0.5060701907233092,
|
| 6971 |
-
"grad_norm": 0.36164185404777527,
|
| 6972 |
-
"learning_rate": 5.1459686071861435e-05,
|
| 6973 |
-
"loss": 0.3910751104354858,
|
| 6974 |
-
"mean_token_accuracy": 0.8874983996152878,
|
| 6975 |
-
"num_tokens": 28993611.0,
|
| 6976 |
-
"step": 6680
|
| 6977 |
-
},
|
| 6978 |
-
{
|
| 6979 |
-
"entropy": 0.3107647761702538,
|
| 6980 |
-
"epoch": 0.5068277808291823,
|
| 6981 |
-
"grad_norm": 0.39866575598716736,
|
| 6982 |
-
"learning_rate": 5.133705396528196e-05,
|
| 6983 |
-
"loss": 0.34510531425476076,
|
| 6984 |
-
"mean_token_accuracy": 0.9003084525465965,
|
| 6985 |
-
"num_tokens": 29089786.0,
|
| 6986 |
-
"step": 6690
|
| 6987 |
-
},
|
| 6988 |
-
{
|
| 6989 |
-
"entropy": 0.29254999160766604,
|
| 6990 |
-
"epoch": 0.5075853709350556,
|
| 6991 |
-
"grad_norm": 0.39197757840156555,
|
| 6992 |
-
"learning_rate": 5.121441380942066e-05,
|
| 6993 |
-
"loss": 0.32735280990600585,
|
| 6994 |
-
"mean_token_accuracy": 0.9101390585303306,
|
| 6995 |
-
"num_tokens": 29178857.0,
|
| 6996 |
-
"step": 6700
|
| 6997 |
-
},
|
| 6998 |
-
{
|
| 6999 |
-
"entropy": 0.3355607531964779,
|
| 7000 |
-
"epoch": 0.5083429610409288,
|
| 7001 |
-
"grad_norm": 0.3769339919090271,
|
| 7002 |
-
"learning_rate": 5.109176634259119e-05,
|
| 7003 |
-
"loss": 0.39189558029174804,
|
| 7004 |
-
"mean_token_accuracy": 0.8900659441947937,
|
| 7005 |
-
"num_tokens": 29281321.0,
|
| 7006 |
-
"step": 6710
|
| 7007 |
-
},
|
| 7008 |
-
{
|
| 7009 |
-
"entropy": 0.33913458716124295,
|
| 7010 |
-
"epoch": 0.509100551146802,
|
| 7011 |
-
"grad_norm": 0.38687586784362793,
|
| 7012 |
-
"learning_rate": 5.096911230315118e-05,
|
| 7013 |
-
"loss": 0.39292776584625244,
|
| 7014 |
-
"mean_token_accuracy": 0.8915379047393799,
|
| 7015 |
-
"num_tokens": 29385794.0,
|
| 7016 |
-
"step": 6720
|
| 7017 |
-
},
|
| 7018 |
-
{
|
| 7019 |
-
"entropy": 0.3371058743447065,
|
| 7020 |
-
"epoch": 0.5098581412526753,
|
| 7021 |
-
"grad_norm": 0.4083423316478729,
|
| 7022 |
-
"learning_rate": 5.084645242949787e-05,
|
| 7023 |
-
"loss": 0.38985190391540525,
|
| 7024 |
-
"mean_token_accuracy": 0.8895333856344223,
|
| 7025 |
-
"num_tokens": 29495430.0,
|
| 7026 |
-
"step": 6730
|
| 7027 |
-
},
|
| 7028 |
-
{
|
| 7029 |
-
"entropy": 0.2875635791569948,
|
| 7030 |
-
"epoch": 0.5106157313585484,
|
| 7031 |
-
"grad_norm": 0.4478307068347931,
|
| 7032 |
-
"learning_rate": 5.072378746006359e-05,
|
| 7033 |
-
"loss": 0.3266411066055298,
|
| 7034 |
-
"mean_token_accuracy": 0.9082275360822678,
|
| 7035 |
-
"num_tokens": 29588821.0,
|
| 7036 |
-
"step": 6740
|
| 7037 |
-
},
|
| 7038 |
-
{
|
| 7039 |
-
"entropy": 0.2940976222977042,
|
| 7040 |
-
"epoch": 0.5113733214644217,
|
| 7041 |
-
"grad_norm": 0.5036411881446838,
|
| 7042 |
-
"learning_rate": 5.0601118133311366e-05,
|
| 7043 |
-
"loss": 0.332806658744812,
|
| 7044 |
-
"mean_token_accuracy": 0.9071679145097733,
|
| 7045 |
-
"num_tokens": 29691240.0,
|
| 7046 |
-
"step": 6750
|
| 7047 |
-
},
|
| 7048 |
-
{
|
| 7049 |
-
"epoch": 0.5113733214644217,
|
| 7050 |
-
"eval_entropy": 0.2943603139133489,
|
| 7051 |
-
"eval_loss": 0.29613742232322693,
|
| 7052 |
-
"eval_mean_token_accuracy": 0.9047154851174086,
|
| 7053 |
-
"eval_num_tokens": 29691240.0,
|
| 7054 |
-
"eval_runtime": 358.159,
|
| 7055 |
-
"eval_samples_per_second": 5.958,
|
| 7056 |
-
"eval_steps_per_second": 1.491,
|
| 7057 |
-
"step": 6750
|
| 7058 |
}
|
| 7059 |
],
|
| 7060 |
"logging_steps": 10,
|
|
@@ -7074,7 +3942,7 @@
|
|
| 7074 |
"attributes": {}
|
| 7075 |
}
|
| 7076 |
},
|
| 7077 |
-
"total_flos":
|
| 7078 |
"train_batch_size": 2,
|
| 7079 |
"trial_name": null,
|
| 7080 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 3750,
|
| 3 |
+
"best_metric": 0.3008323907852173,
|
| 4 |
+
"best_model_checkpoint": "./lfm25-kodcode-lora/checkpoint-3750",
|
| 5 |
+
"epoch": 0.2840962897024565,
|
| 6 |
"eval_steps": 250,
|
| 7 |
+
"global_step": 3750,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3923 |
"eval_samples_per_second": 5.87,
|
| 3924 |
"eval_steps_per_second": 1.469,
|
| 3925 |
"step": 3750
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3926 |
}
|
| 3927 |
],
|
| 3928 |
"logging_steps": 10,
|
|
|
|
| 3942 |
"attributes": {}
|
| 3943 |
}
|
| 3944 |
},
|
| 3945 |
+
"total_flos": 3.23806700379308e+17,
|
| 3946 |
"train_batch_size": 2,
|
| 3947 |
"trial_name": null,
|
| 3948 |
"trial_params": null
|