tuantmdev commited on
Commit
45fb7f8
·
verified ·
1 Parent(s): 8e95064

Training in progress, step 200, checkpoint

Browse files
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6b767492292e2934ea3c30fcff6c6a19e7a5964603cd8ec03c783e96c6fe6e87
3
  size 74291604
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f8d860db2b51f3ffc9bce828907abd6793370f57ef3ee1b7ef77fdf9e03ea3c
3
  size 74291604
last-checkpoint/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:251d12d434d9fba29802bf5bdcbdc5943d943f4f3d170dcc34ffc801d60a4dc2
3
  size 14244
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98ff43aa37acd82806ac685199711e5d39670e1771e7ec46c38c053fb8c19ca2
3
  size 14244
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:500b1c78c9d3af1c55fbf9cc6e380394ac567579f265e54ad9607c3dc9f74108
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb148ad6cd6502048dc3eee274a6708e5b4f6b72c0d56ebc272935446c220acd
3
  size 1064
last-checkpoint/trainer_state.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.5602976581308821,
5
  "eval_steps": 40,
6
- "global_step": 160,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -75,6 +75,21 @@
75
  "eval_samples_per_second": 5.115,
76
  "eval_steps_per_second": 2.563,
77
  "step": 160
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78
  }
79
  ],
80
  "logging_steps": 40,
@@ -89,12 +104,12 @@
89
  "should_evaluate": false,
90
  "should_log": false,
91
  "should_save": true,
92
- "should_training_stop": false
93
  },
94
  "attributes": {}
95
  }
96
  },
97
- "total_flos": 4.011191962970358e+17,
98
  "train_batch_size": 2,
99
  "trial_name": null,
100
  "trial_params": null
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.7003720726636026,
5
  "eval_steps": 40,
6
+ "global_step": 200,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
75
  "eval_samples_per_second": 5.115,
76
  "eval_steps_per_second": 2.563,
77
  "step": 160
78
+ },
79
+ {
80
+ "epoch": 0.7003720726636026,
81
+ "grad_norm": NaN,
82
+ "learning_rate": 0.0,
83
+ "loss": 0.0,
84
+ "step": 200
85
+ },
86
+ {
87
+ "epoch": 0.7003720726636026,
88
+ "eval_loss": NaN,
89
+ "eval_runtime": 94.0957,
90
+ "eval_samples_per_second": 5.112,
91
+ "eval_steps_per_second": 2.561,
92
+ "step": 200
93
  }
94
  ],
95
  "logging_steps": 40,
 
104
  "should_evaluate": false,
105
  "should_log": false,
106
  "should_save": true,
107
+ "should_training_stop": true
108
  },
109
  "attributes": {}
110
  }
111
  },
112
+ "total_flos": 5.0224916726494e+17,
113
  "train_batch_size": 2,
114
  "trial_name": null,
115
  "trial_params": null