Training in progress, step 1050, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 54285928
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b5640fad40281fd5d639511356095a8552ab13034cf39c186ed59e06ab055a1
|
| 3 |
size 54285928
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 27753786
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1e01ad8da81a858e8e938a7bdca8254bfa5438d6b1c7251bdd13f12c56d5e1f
|
| 3 |
size 27753786
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:508337fd7bc9ca1cce78c0c53b3e5fba6c6a4bdf1bdeb4293058ee7e7b6238a1
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a5cc0d0ebf737ac8a43a5138b0ded560b01a769ad496c84f0d332e37eb84e28
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 1.
|
| 3 |
-
"best_model_checkpoint": "miner_id_24/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 150,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6363,6 +6363,1064 @@
|
|
| 6363 |
"eval_samples_per_second": 164.986,
|
| 6364 |
"eval_steps_per_second": 20.623,
|
| 6365 |
"step": 900
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6366 |
}
|
| 6367 |
],
|
| 6368 |
"logging_steps": 1,
|
|
@@ -6391,7 +7449,7 @@
|
|
| 6391 |
"attributes": {}
|
| 6392 |
}
|
| 6393 |
},
|
| 6394 |
-
"total_flos": 1.
|
| 6395 |
"train_batch_size": 12,
|
| 6396 |
"trial_name": null,
|
| 6397 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.3224910497665405,
|
| 3 |
+
"best_model_checkpoint": "miner_id_24/checkpoint-1050",
|
| 4 |
+
"epoch": 0.9657392504023914,
|
| 5 |
"eval_steps": 150,
|
| 6 |
+
"global_step": 1050,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6363 |
"eval_samples_per_second": 164.986,
|
| 6364 |
"eval_steps_per_second": 20.623,
|
| 6365 |
"step": 900
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.8286962520119567,
|
| 6369 |
+
"grad_norm": 0.8766337037086487,
|
| 6370 |
+
"learning_rate": 7.375678447088347e-06,
|
| 6371 |
+
"loss": 1.5154,
|
| 6372 |
+
"step": 901
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.8296160036790067,
|
| 6376 |
+
"grad_norm": 0.8737375140190125,
|
| 6377 |
+
"learning_rate": 7.298977877893687e-06,
|
| 6378 |
+
"loss": 1.4447,
|
| 6379 |
+
"step": 902
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.8305357553460566,
|
| 6383 |
+
"grad_norm": 0.9431170225143433,
|
| 6384 |
+
"learning_rate": 7.222646792325516e-06,
|
| 6385 |
+
"loss": 1.4588,
|
| 6386 |
+
"step": 903
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.8314555070131064,
|
| 6390 |
+
"grad_norm": 0.9367691874504089,
|
| 6391 |
+
"learning_rate": 7.146685850861851e-06,
|
| 6392 |
+
"loss": 1.4205,
|
| 6393 |
+
"step": 904
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.8323752586801564,
|
| 6397 |
+
"grad_norm": 0.812258780002594,
|
| 6398 |
+
"learning_rate": 7.071095710777925e-06,
|
| 6399 |
+
"loss": 1.4177,
|
| 6400 |
+
"step": 905
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.8332950103472062,
|
| 6404 |
+
"grad_norm": 0.7034198045730591,
|
| 6405 |
+
"learning_rate": 6.995877026140468e-06,
|
| 6406 |
+
"loss": 1.4146,
|
| 6407 |
+
"step": 906
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.8342147620142562,
|
| 6411 |
+
"grad_norm": 0.7884905934333801,
|
| 6412 |
+
"learning_rate": 6.921030447802146e-06,
|
| 6413 |
+
"loss": 1.4616,
|
| 6414 |
+
"step": 907
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.8351345136813061,
|
| 6418 |
+
"grad_norm": 0.8112537860870361,
|
| 6419 |
+
"learning_rate": 6.8465566233957945e-06,
|
| 6420 |
+
"loss": 1.3435,
|
| 6421 |
+
"step": 908
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.8360542653483559,
|
| 6425 |
+
"grad_norm": 0.7667593955993652,
|
| 6426 |
+
"learning_rate": 6.772456197328919e-06,
|
| 6427 |
+
"loss": 1.464,
|
| 6428 |
+
"step": 909
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.8369740170154059,
|
| 6432 |
+
"grad_norm": 0.762269914150238,
|
| 6433 |
+
"learning_rate": 6.698729810778065e-06,
|
| 6434 |
+
"loss": 1.4473,
|
| 6435 |
+
"step": 910
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.8378937686824557,
|
| 6439 |
+
"grad_norm": 0.852673351764679,
|
| 6440 |
+
"learning_rate": 6.625378101683316e-06,
|
| 6441 |
+
"loss": 1.4215,
|
| 6442 |
+
"step": 911
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.8388135203495056,
|
| 6446 |
+
"grad_norm": 0.7429057359695435,
|
| 6447 |
+
"learning_rate": 6.552401704742678e-06,
|
| 6448 |
+
"loss": 1.4426,
|
| 6449 |
+
"step": 912
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.8397332720165556,
|
| 6453 |
+
"grad_norm": 0.6884950995445251,
|
| 6454 |
+
"learning_rate": 6.4798012514067475e-06,
|
| 6455 |
+
"loss": 1.4016,
|
| 6456 |
+
"step": 913
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.8406530236836054,
|
| 6460 |
+
"grad_norm": 0.6550636291503906,
|
| 6461 |
+
"learning_rate": 6.407577369873069e-06,
|
| 6462 |
+
"loss": 1.4468,
|
| 6463 |
+
"step": 914
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.8415727753506553,
|
| 6467 |
+
"grad_norm": 0.5837852358818054,
|
| 6468 |
+
"learning_rate": 6.335730685080837e-06,
|
| 6469 |
+
"loss": 1.4036,
|
| 6470 |
+
"step": 915
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.8424925270177053,
|
| 6474 |
+
"grad_norm": 0.5570608377456665,
|
| 6475 |
+
"learning_rate": 6.264261818705419e-06,
|
| 6476 |
+
"loss": 1.3483,
|
| 6477 |
+
"step": 916
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.8434122786847551,
|
| 6481 |
+
"grad_norm": 0.7056939005851746,
|
| 6482 |
+
"learning_rate": 6.193171389152997e-06,
|
| 6483 |
+
"loss": 1.3397,
|
| 6484 |
+
"step": 917
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.844332030351805,
|
| 6488 |
+
"grad_norm": 0.623600423336029,
|
| 6489 |
+
"learning_rate": 6.122460011555187e-06,
|
| 6490 |
+
"loss": 1.4304,
|
| 6491 |
+
"step": 918
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.8452517820188549,
|
| 6495 |
+
"grad_norm": 0.6012278199195862,
|
| 6496 |
+
"learning_rate": 6.052128297763804e-06,
|
| 6497 |
+
"loss": 1.3684,
|
| 6498 |
+
"step": 919
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.8461715336859048,
|
| 6502 |
+
"grad_norm": 0.582744836807251,
|
| 6503 |
+
"learning_rate": 5.982176856345445e-06,
|
| 6504 |
+
"loss": 1.4205,
|
| 6505 |
+
"step": 920
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.8470912853529547,
|
| 6509 |
+
"grad_norm": 0.5616964101791382,
|
| 6510 |
+
"learning_rate": 5.912606292576283e-06,
|
| 6511 |
+
"loss": 1.3209,
|
| 6512 |
+
"step": 921
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.8480110370200046,
|
| 6516 |
+
"grad_norm": 0.5474282503128052,
|
| 6517 |
+
"learning_rate": 5.843417208436908e-06,
|
| 6518 |
+
"loss": 1.4125,
|
| 6519 |
+
"step": 922
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.8489307886870545,
|
| 6523 |
+
"grad_norm": 0.533388614654541,
|
| 6524 |
+
"learning_rate": 5.774610202606939e-06,
|
| 6525 |
+
"loss": 1.4116,
|
| 6526 |
+
"step": 923
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.8498505403541043,
|
| 6530 |
+
"grad_norm": 0.5694478154182434,
|
| 6531 |
+
"learning_rate": 5.706185870460018e-06,
|
| 6532 |
+
"loss": 1.509,
|
| 6533 |
+
"step": 924
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.8507702920211543,
|
| 6537 |
+
"grad_norm": 0.5748287439346313,
|
| 6538 |
+
"learning_rate": 5.638144804058559e-06,
|
| 6539 |
+
"loss": 1.3528,
|
| 6540 |
+
"step": 925
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.8516900436882042,
|
| 6544 |
+
"grad_norm": 0.6192615032196045,
|
| 6545 |
+
"learning_rate": 5.5704875921486655e-06,
|
| 6546 |
+
"loss": 1.3098,
|
| 6547 |
+
"step": 926
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.852609795355254,
|
| 6551 |
+
"grad_norm": 0.6460704207420349,
|
| 6552 |
+
"learning_rate": 5.503214820154978e-06,
|
| 6553 |
+
"loss": 1.3839,
|
| 6554 |
+
"step": 927
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.853529547022304,
|
| 6558 |
+
"grad_norm": 0.620794951915741,
|
| 6559 |
+
"learning_rate": 5.436327070175728e-06,
|
| 6560 |
+
"loss": 1.4197,
|
| 6561 |
+
"step": 928
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.8544492986893538,
|
| 6565 |
+
"grad_norm": 0.6275455355644226,
|
| 6566 |
+
"learning_rate": 5.369824920977568e-06,
|
| 6567 |
+
"loss": 1.2891,
|
| 6568 |
+
"step": 929
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.8553690503564038,
|
| 6572 |
+
"grad_norm": 0.5857694149017334,
|
| 6573 |
+
"learning_rate": 5.303708947990637e-06,
|
| 6574 |
+
"loss": 1.3334,
|
| 6575 |
+
"step": 930
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.8562888020234537,
|
| 6579 |
+
"grad_norm": 0.6003711819648743,
|
| 6580 |
+
"learning_rate": 5.2379797233035824e-06,
|
| 6581 |
+
"loss": 1.395,
|
| 6582 |
+
"step": 931
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.8572085536905035,
|
| 6586 |
+
"grad_norm": 0.6273806095123291,
|
| 6587 |
+
"learning_rate": 5.1726378156585816e-06,
|
| 6588 |
+
"loss": 1.2778,
|
| 6589 |
+
"step": 932
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.8581283053575535,
|
| 6593 |
+
"grad_norm": 0.6366182565689087,
|
| 6594 |
+
"learning_rate": 5.10768379044641e-06,
|
| 6595 |
+
"loss": 1.3508,
|
| 6596 |
+
"step": 933
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.8590480570246034,
|
| 6600 |
+
"grad_norm": 0.6845077872276306,
|
| 6601 |
+
"learning_rate": 5.043118209701631e-06,
|
| 6602 |
+
"loss": 1.2843,
|
| 6603 |
+
"step": 934
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.8599678086916532,
|
| 6607 |
+
"grad_norm": 0.6707909107208252,
|
| 6608 |
+
"learning_rate": 4.978941632097611e-06,
|
| 6609 |
+
"loss": 1.3239,
|
| 6610 |
+
"step": 935
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.8608875603587032,
|
| 6614 |
+
"grad_norm": 0.7041406631469727,
|
| 6615 |
+
"learning_rate": 4.9151546129417804e-06,
|
| 6616 |
+
"loss": 1.2556,
|
| 6617 |
+
"step": 936
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.861807312025753,
|
| 6621 |
+
"grad_norm": 0.6683023571968079,
|
| 6622 |
+
"learning_rate": 4.8517577041707955e-06,
|
| 6623 |
+
"loss": 1.289,
|
| 6624 |
+
"step": 937
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.8627270636928029,
|
| 6628 |
+
"grad_norm": 0.6463608741760254,
|
| 6629 |
+
"learning_rate": 4.788751454345763e-06,
|
| 6630 |
+
"loss": 1.225,
|
| 6631 |
+
"step": 938
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.8636468153598529,
|
| 6635 |
+
"grad_norm": 0.6901978254318237,
|
| 6636 |
+
"learning_rate": 4.726136408647464e-06,
|
| 6637 |
+
"loss": 1.2177,
|
| 6638 |
+
"step": 939
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.8645665670269027,
|
| 6642 |
+
"grad_norm": 0.6679742336273193,
|
| 6643 |
+
"learning_rate": 4.663913108871726e-06,
|
| 6644 |
+
"loss": 1.2586,
|
| 6645 |
+
"step": 940
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.8654863186939526,
|
| 6649 |
+
"grad_norm": 0.6778735518455505,
|
| 6650 |
+
"learning_rate": 4.60208209342462e-06,
|
| 6651 |
+
"loss": 1.183,
|
| 6652 |
+
"step": 941
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.8664060703610025,
|
| 6656 |
+
"grad_norm": 0.6251430511474609,
|
| 6657 |
+
"learning_rate": 4.540643897317887e-06,
|
| 6658 |
+
"loss": 1.2523,
|
| 6659 |
+
"step": 942
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.8673258220280524,
|
| 6663 |
+
"grad_norm": 0.6894196271896362,
|
| 6664 |
+
"learning_rate": 4.479599052164268e-06,
|
| 6665 |
+
"loss": 1.183,
|
| 6666 |
+
"step": 943
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.8682455736951024,
|
| 6670 |
+
"grad_norm": 0.6839209198951721,
|
| 6671 |
+
"learning_rate": 4.418948086172914e-06,
|
| 6672 |
+
"loss": 1.1992,
|
| 6673 |
+
"step": 944
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.8691653253621522,
|
| 6677 |
+
"grad_norm": 0.7572594285011292,
|
| 6678 |
+
"learning_rate": 4.35869152414482e-06,
|
| 6679 |
+
"loss": 1.1731,
|
| 6680 |
+
"step": 945
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.8700850770292021,
|
| 6684 |
+
"grad_norm": 0.7147699594497681,
|
| 6685 |
+
"learning_rate": 4.298829887468275e-06,
|
| 6686 |
+
"loss": 1.1665,
|
| 6687 |
+
"step": 946
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.8710048286962521,
|
| 6691 |
+
"grad_norm": 0.7666782736778259,
|
| 6692 |
+
"learning_rate": 4.2393636941143675e-06,
|
| 6693 |
+
"loss": 1.149,
|
| 6694 |
+
"step": 947
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.8719245803633019,
|
| 6698 |
+
"grad_norm": 0.7843433022499084,
|
| 6699 |
+
"learning_rate": 4.180293458632489e-06,
|
| 6700 |
+
"loss": 1.0903,
|
| 6701 |
+
"step": 948
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.8728443320303518,
|
| 6705 |
+
"grad_norm": 0.958113431930542,
|
| 6706 |
+
"learning_rate": 4.121619692145878e-06,
|
| 6707 |
+
"loss": 1.118,
|
| 6708 |
+
"step": 949
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.8737640836974017,
|
| 6712 |
+
"grad_norm": 1.1284202337265015,
|
| 6713 |
+
"learning_rate": 4.0633429023472e-06,
|
| 6714 |
+
"loss": 0.9711,
|
| 6715 |
+
"step": 950
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.8746838353644516,
|
| 6719 |
+
"grad_norm": 0.8368450403213501,
|
| 6720 |
+
"learning_rate": 4.005463593494163e-06,
|
| 6721 |
+
"loss": 1.4433,
|
| 6722 |
+
"step": 951
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.8756035870315015,
|
| 6726 |
+
"grad_norm": 0.6638758182525635,
|
| 6727 |
+
"learning_rate": 3.947982266405159e-06,
|
| 6728 |
+
"loss": 1.4285,
|
| 6729 |
+
"step": 952
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.8765233386985514,
|
| 6733 |
+
"grad_norm": 0.8789987564086914,
|
| 6734 |
+
"learning_rate": 3.890899418454913e-06,
|
| 6735 |
+
"loss": 1.4212,
|
| 6736 |
+
"step": 953
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.8774430903656013,
|
| 6740 |
+
"grad_norm": 0.847080409526825,
|
| 6741 |
+
"learning_rate": 3.834215543570191e-06,
|
| 6742 |
+
"loss": 1.4124,
|
| 6743 |
+
"step": 954
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.8783628420326511,
|
| 6747 |
+
"grad_norm": 0.9596214890480042,
|
| 6748 |
+
"learning_rate": 3.777931132225526e-06,
|
| 6749 |
+
"loss": 1.3723,
|
| 6750 |
+
"step": 955
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.8792825936997011,
|
| 6754 |
+
"grad_norm": 0.9075647592544556,
|
| 6755 |
+
"learning_rate": 3.72204667143895e-06,
|
| 6756 |
+
"loss": 1.493,
|
| 6757 |
+
"step": 956
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.880202345366751,
|
| 6761 |
+
"grad_norm": 0.780536413192749,
|
| 6762 |
+
"learning_rate": 3.6665626447678237e-06,
|
| 6763 |
+
"loss": 1.4126,
|
| 6764 |
+
"step": 957
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.8811220970338008,
|
| 6768 |
+
"grad_norm": 0.6997688412666321,
|
| 6769 |
+
"learning_rate": 3.611479532304618e-06,
|
| 6770 |
+
"loss": 1.389,
|
| 6771 |
+
"step": 958
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.8820418487008508,
|
| 6775 |
+
"grad_norm": 0.620875358581543,
|
| 6776 |
+
"learning_rate": 3.556797810672785e-06,
|
| 6777 |
+
"loss": 1.3514,
|
| 6778 |
+
"step": 959
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.8829616003679006,
|
| 6782 |
+
"grad_norm": 0.6854445338249207,
|
| 6783 |
+
"learning_rate": 3.5025179530225994e-06,
|
| 6784 |
+
"loss": 1.4661,
|
| 6785 |
+
"step": 960
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.8838813520349506,
|
| 6789 |
+
"grad_norm": 0.7020566463470459,
|
| 6790 |
+
"learning_rate": 3.4486404290271113e-06,
|
| 6791 |
+
"loss": 1.4115,
|
| 6792 |
+
"step": 961
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.8848011037020005,
|
| 6796 |
+
"grad_norm": 0.6943616271018982,
|
| 6797 |
+
"learning_rate": 3.3951657048780227e-06,
|
| 6798 |
+
"loss": 1.4774,
|
| 6799 |
+
"step": 962
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.8857208553690503,
|
| 6803 |
+
"grad_norm": 0.7479608654975891,
|
| 6804 |
+
"learning_rate": 3.3420942432817127e-06,
|
| 6805 |
+
"loss": 1.4625,
|
| 6806 |
+
"step": 963
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.8866406070361003,
|
| 6810 |
+
"grad_norm": 0.7025173902511597,
|
| 6811 |
+
"learning_rate": 3.289426503455201e-06,
|
| 6812 |
+
"loss": 1.4019,
|
| 6813 |
+
"step": 964
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.8875603587031502,
|
| 6817 |
+
"grad_norm": 0.673040509223938,
|
| 6818 |
+
"learning_rate": 3.2371629411221848e-06,
|
| 6819 |
+
"loss": 1.4343,
|
| 6820 |
+
"step": 965
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.8884801103702,
|
| 6824 |
+
"grad_norm": 0.728541910648346,
|
| 6825 |
+
"learning_rate": 3.185304008509077e-06,
|
| 6826 |
+
"loss": 1.5093,
|
| 6827 |
+
"step": 966
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.88939986203725,
|
| 6831 |
+
"grad_norm": 0.6773453950881958,
|
| 6832 |
+
"learning_rate": 3.133850154341139e-06,
|
| 6833 |
+
"loss": 1.4002,
|
| 6834 |
+
"step": 967
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.8903196137042998,
|
| 6838 |
+
"grad_norm": 0.6363242864608765,
|
| 6839 |
+
"learning_rate": 3.082801823838527e-06,
|
| 6840 |
+
"loss": 1.4272,
|
| 6841 |
+
"step": 968
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.8912393653713497,
|
| 6845 |
+
"grad_norm": 0.5722589492797852,
|
| 6846 |
+
"learning_rate": 3.032159458712508e-06,
|
| 6847 |
+
"loss": 1.3557,
|
| 6848 |
+
"step": 969
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.8921591170383997,
|
| 6852 |
+
"grad_norm": 0.5886601209640503,
|
| 6853 |
+
"learning_rate": 2.981923497161615e-06,
|
| 6854 |
+
"loss": 1.3874,
|
| 6855 |
+
"step": 970
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.8930788687054495,
|
| 6859 |
+
"grad_norm": 0.6230661273002625,
|
| 6860 |
+
"learning_rate": 2.9320943738678107e-06,
|
| 6861 |
+
"loss": 1.3784,
|
| 6862 |
+
"step": 971
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.8939986203724994,
|
| 6866 |
+
"grad_norm": 0.5844275951385498,
|
| 6867 |
+
"learning_rate": 2.882672519992824e-06,
|
| 6868 |
+
"loss": 1.4153,
|
| 6869 |
+
"step": 972
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.8949183720395493,
|
| 6873 |
+
"grad_norm": 0.6414538621902466,
|
| 6874 |
+
"learning_rate": 2.833658363174302e-06,
|
| 6875 |
+
"loss": 1.3611,
|
| 6876 |
+
"step": 973
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.8958381237065992,
|
| 6880 |
+
"grad_norm": 0.6074815392494202,
|
| 6881 |
+
"learning_rate": 2.785052327522214e-06,
|
| 6882 |
+
"loss": 1.3607,
|
| 6883 |
+
"step": 974
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.8967578753736491,
|
| 6887 |
+
"grad_norm": 0.5938957333564758,
|
| 6888 |
+
"learning_rate": 2.73685483361511e-06,
|
| 6889 |
+
"loss": 1.3765,
|
| 6890 |
+
"step": 975
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.897677627040699,
|
| 6894 |
+
"grad_norm": 0.5869003534317017,
|
| 6895 |
+
"learning_rate": 2.6890662984965232e-06,
|
| 6896 |
+
"loss": 1.392,
|
| 6897 |
+
"step": 976
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.8985973787077489,
|
| 6901 |
+
"grad_norm": 0.5588386654853821,
|
| 6902 |
+
"learning_rate": 2.6416871356713224e-06,
|
| 6903 |
+
"loss": 1.3047,
|
| 6904 |
+
"step": 977
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.8995171303747989,
|
| 6908 |
+
"grad_norm": 0.5922186970710754,
|
| 6909 |
+
"learning_rate": 2.594717755102205e-06,
|
| 6910 |
+
"loss": 1.3928,
|
| 6911 |
+
"step": 978
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.9004368820418487,
|
| 6915 |
+
"grad_norm": 0.5693724155426025,
|
| 6916 |
+
"learning_rate": 2.548158563206038e-06,
|
| 6917 |
+
"loss": 1.347,
|
| 6918 |
+
"step": 979
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.9013566337088986,
|
| 6922 |
+
"grad_norm": 0.6117263436317444,
|
| 6923 |
+
"learning_rate": 2.50200996285046e-06,
|
| 6924 |
+
"loss": 1.3568,
|
| 6925 |
+
"step": 980
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.9022763853759485,
|
| 6929 |
+
"grad_norm": 0.5885259509086609,
|
| 6930 |
+
"learning_rate": 2.4562723533503083e-06,
|
| 6931 |
+
"loss": 1.4184,
|
| 6932 |
+
"step": 981
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.9031961370429984,
|
| 6936 |
+
"grad_norm": 0.6112256646156311,
|
| 6937 |
+
"learning_rate": 2.4109461304642256e-06,
|
| 6938 |
+
"loss": 1.3344,
|
| 6939 |
+
"step": 982
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.9041158887100483,
|
| 6943 |
+
"grad_norm": 0.6500238180160522,
|
| 6944 |
+
"learning_rate": 2.366031686391168e-06,
|
| 6945 |
+
"loss": 1.3372,
|
| 6946 |
+
"step": 983
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.9050356403770982,
|
| 6950 |
+
"grad_norm": 0.6185190677642822,
|
| 6951 |
+
"learning_rate": 2.3215294097670925e-06,
|
| 6952 |
+
"loss": 1.2273,
|
| 6953 |
+
"step": 984
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.9059553920441481,
|
| 6957 |
+
"grad_norm": 0.6523995995521545,
|
| 6958 |
+
"learning_rate": 2.277439685661509e-06,
|
| 6959 |
+
"loss": 1.2538,
|
| 6960 |
+
"step": 985
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.9068751437111979,
|
| 6964 |
+
"grad_norm": 0.7136437296867371,
|
| 6965 |
+
"learning_rate": 2.2337628955742264e-06,
|
| 6966 |
+
"loss": 1.3739,
|
| 6967 |
+
"step": 986
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.9077948953782479,
|
| 6971 |
+
"grad_norm": 0.6043840050697327,
|
| 6972 |
+
"learning_rate": 2.1904994174319905e-06,
|
| 6973 |
+
"loss": 1.2184,
|
| 6974 |
+
"step": 987
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.9087146470452978,
|
| 6978 |
+
"grad_norm": 0.6362565159797668,
|
| 6979 |
+
"learning_rate": 2.1476496255852683e-06,
|
| 6980 |
+
"loss": 1.1398,
|
| 6981 |
+
"step": 988
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.9096343987123476,
|
| 6985 |
+
"grad_norm": 0.6597528457641602,
|
| 6986 |
+
"learning_rate": 2.1052138908049303e-06,
|
| 6987 |
+
"loss": 1.1972,
|
| 6988 |
+
"step": 989
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.9105541503793976,
|
| 6992 |
+
"grad_norm": 0.679057240486145,
|
| 6993 |
+
"learning_rate": 2.0631925802791606e-06,
|
| 6994 |
+
"loss": 1.2572,
|
| 6995 |
+
"step": 990
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.9114739020464474,
|
| 6999 |
+
"grad_norm": 0.6650072336196899,
|
| 7000 |
+
"learning_rate": 2.021586057610153e-06,
|
| 7001 |
+
"loss": 1.1868,
|
| 7002 |
+
"step": 991
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.9123936537134973,
|
| 7006 |
+
"grad_norm": 0.6258329749107361,
|
| 7007 |
+
"learning_rate": 1.9803946828110375e-06,
|
| 7008 |
+
"loss": 1.209,
|
| 7009 |
+
"step": 992
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 0.9133134053805473,
|
| 7013 |
+
"grad_norm": 0.6818736791610718,
|
| 7014 |
+
"learning_rate": 1.9396188123027737e-06,
|
| 7015 |
+
"loss": 1.2432,
|
| 7016 |
+
"step": 993
|
| 7017 |
+
},
|
| 7018 |
+
{
|
| 7019 |
+
"epoch": 0.9142331570475971,
|
| 7020 |
+
"grad_norm": 0.7300404906272888,
|
| 7021 |
+
"learning_rate": 1.8992587989110134e-06,
|
| 7022 |
+
"loss": 1.2549,
|
| 7023 |
+
"step": 994
|
| 7024 |
+
},
|
| 7025 |
+
{
|
| 7026 |
+
"epoch": 0.915152908714647,
|
| 7027 |
+
"grad_norm": 0.7216602563858032,
|
| 7028 |
+
"learning_rate": 1.8593149918630925e-06,
|
| 7029 |
+
"loss": 1.1911,
|
| 7030 |
+
"step": 995
|
| 7031 |
+
},
|
| 7032 |
+
{
|
| 7033 |
+
"epoch": 0.916072660381697,
|
| 7034 |
+
"grad_norm": 0.7485631704330444,
|
| 7035 |
+
"learning_rate": 1.8197877367849947e-06,
|
| 7036 |
+
"loss": 1.1326,
|
| 7037 |
+
"step": 996
|
| 7038 |
+
},
|
| 7039 |
+
{
|
| 7040 |
+
"epoch": 0.9169924120487468,
|
| 7041 |
+
"grad_norm": 0.8240882158279419,
|
| 7042 |
+
"learning_rate": 1.7806773756983642e-06,
|
| 7043 |
+
"loss": 1.1299,
|
| 7044 |
+
"step": 997
|
| 7045 |
+
},
|
| 7046 |
+
{
|
| 7047 |
+
"epoch": 0.9179121637157968,
|
| 7048 |
+
"grad_norm": 0.9147471189498901,
|
| 7049 |
+
"learning_rate": 1.7419842470175195e-06,
|
| 7050 |
+
"loss": 1.1179,
|
| 7051 |
+
"step": 998
|
| 7052 |
+
},
|
| 7053 |
+
{
|
| 7054 |
+
"epoch": 0.9188319153828466,
|
| 7055 |
+
"grad_norm": 0.9360700249671936,
|
| 7056 |
+
"learning_rate": 1.70370868554659e-06,
|
| 7057 |
+
"loss": 1.0562,
|
| 7058 |
+
"step": 999
|
| 7059 |
+
},
|
| 7060 |
+
{
|
| 7061 |
+
"epoch": 0.9197516670498965,
|
| 7062 |
+
"grad_norm": 1.174989104270935,
|
| 7063 |
+
"learning_rate": 1.6658510224765333e-06,
|
| 7064 |
+
"loss": 0.9121,
|
| 7065 |
+
"step": 1000
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 0.9206714187169465,
|
| 7069 |
+
"grad_norm": 0.8917292952537537,
|
| 7070 |
+
"learning_rate": 1.6284115853823445e-06,
|
| 7071 |
+
"loss": 1.4961,
|
| 7072 |
+
"step": 1001
|
| 7073 |
+
},
|
| 7074 |
+
{
|
| 7075 |
+
"epoch": 0.9215911703839963,
|
| 7076 |
+
"grad_norm": 0.6432257890701294,
|
| 7077 |
+
"learning_rate": 1.5913906982201742e-06,
|
| 7078 |
+
"loss": 1.488,
|
| 7079 |
+
"step": 1002
|
| 7080 |
+
},
|
| 7081 |
+
{
|
| 7082 |
+
"epoch": 0.9225109220510462,
|
| 7083 |
+
"grad_norm": 0.7689481973648071,
|
| 7084 |
+
"learning_rate": 1.5547886813245539e-06,
|
| 7085 |
+
"loss": 1.4265,
|
| 7086 |
+
"step": 1003
|
| 7087 |
+
},
|
| 7088 |
+
{
|
| 7089 |
+
"epoch": 0.9234306737180961,
|
| 7090 |
+
"grad_norm": 0.7164052128791809,
|
| 7091 |
+
"learning_rate": 1.5186058514055912e-06,
|
| 7092 |
+
"loss": 1.4054,
|
| 7093 |
+
"step": 1004
|
| 7094 |
+
},
|
| 7095 |
+
{
|
| 7096 |
+
"epoch": 0.924350425385146,
|
| 7097 |
+
"grad_norm": 0.8932134509086609,
|
| 7098 |
+
"learning_rate": 1.4828425215462848e-06,
|
| 7099 |
+
"loss": 1.403,
|
| 7100 |
+
"step": 1005
|
| 7101 |
+
},
|
| 7102 |
+
{
|
| 7103 |
+
"epoch": 0.9252701770521959,
|
| 7104 |
+
"grad_norm": 0.8750680685043335,
|
| 7105 |
+
"learning_rate": 1.447499001199748e-06,
|
| 7106 |
+
"loss": 1.3956,
|
| 7107 |
+
"step": 1006
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 0.9261899287192458,
|
| 7111 |
+
"grad_norm": 0.7176107168197632,
|
| 7112 |
+
"learning_rate": 1.4125755961865827e-06,
|
| 7113 |
+
"loss": 1.4235,
|
| 7114 |
+
"step": 1007
|
| 7115 |
+
},
|
| 7116 |
+
{
|
| 7117 |
+
"epoch": 0.9271096803862957,
|
| 7118 |
+
"grad_norm": 0.7204969525337219,
|
| 7119 |
+
"learning_rate": 1.3780726086922103e-06,
|
| 7120 |
+
"loss": 1.3773,
|
| 7121 |
+
"step": 1008
|
| 7122 |
+
},
|
| 7123 |
+
{
|
| 7124 |
+
"epoch": 0.9280294320533456,
|
| 7125 |
+
"grad_norm": 0.6472546458244324,
|
| 7126 |
+
"learning_rate": 1.3439903372642615e-06,
|
| 7127 |
+
"loss": 1.4734,
|
| 7128 |
+
"step": 1009
|
| 7129 |
+
},
|
| 7130 |
+
{
|
| 7131 |
+
"epoch": 0.9289491837203955,
|
| 7132 |
+
"grad_norm": 0.679750919342041,
|
| 7133 |
+
"learning_rate": 1.3103290768099797e-06,
|
| 7134 |
+
"loss": 1.5028,
|
| 7135 |
+
"step": 1010
|
| 7136 |
+
},
|
| 7137 |
+
{
|
| 7138 |
+
"epoch": 0.9298689353874454,
|
| 7139 |
+
"grad_norm": 0.6491613984107971,
|
| 7140 |
+
"learning_rate": 1.2770891185937105e-06,
|
| 7141 |
+
"loss": 1.403,
|
| 7142 |
+
"step": 1011
|
| 7143 |
+
},
|
| 7144 |
+
{
|
| 7145 |
+
"epoch": 0.9307886870544952,
|
| 7146 |
+
"grad_norm": 0.6442059278488159,
|
| 7147 |
+
"learning_rate": 1.2442707502343332e-06,
|
| 7148 |
+
"loss": 1.4124,
|
| 7149 |
+
"step": 1012
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 0.9317084387215452,
|
| 7153 |
+
"grad_norm": 0.5981637835502625,
|
| 7154 |
+
"learning_rate": 1.2118742557027884e-06,
|
| 7155 |
+
"loss": 1.459,
|
| 7156 |
+
"step": 1013
|
| 7157 |
+
},
|
| 7158 |
+
{
|
| 7159 |
+
"epoch": 0.9326281903885951,
|
| 7160 |
+
"grad_norm": 0.5459677577018738,
|
| 7161 |
+
"learning_rate": 1.1798999153196433e-06,
|
| 7162 |
+
"loss": 1.4171,
|
| 7163 |
+
"step": 1014
|
| 7164 |
+
},
|
| 7165 |
+
{
|
| 7166 |
+
"epoch": 0.933547942055645,
|
| 7167 |
+
"grad_norm": 0.5810702443122864,
|
| 7168 |
+
"learning_rate": 1.1483480057526363e-06,
|
| 7169 |
+
"loss": 1.3995,
|
| 7170 |
+
"step": 1015
|
| 7171 |
+
},
|
| 7172 |
+
{
|
| 7173 |
+
"epoch": 0.9344676937226949,
|
| 7174 |
+
"grad_norm": 0.5334146022796631,
|
| 7175 |
+
"learning_rate": 1.1172188000142802e-06,
|
| 7176 |
+
"loss": 1.4004,
|
| 7177 |
+
"step": 1016
|
| 7178 |
+
},
|
| 7179 |
+
{
|
| 7180 |
+
"epoch": 0.9353874453897447,
|
| 7181 |
+
"grad_norm": 0.5717347860336304,
|
| 7182 |
+
"learning_rate": 1.0865125674595466e-06,
|
| 7183 |
+
"loss": 1.3843,
|
| 7184 |
+
"step": 1017
|
| 7185 |
+
},
|
| 7186 |
+
{
|
| 7187 |
+
"epoch": 0.9363071970567947,
|
| 7188 |
+
"grad_norm": 0.5235407948493958,
|
| 7189 |
+
"learning_rate": 1.0562295737834737e-06,
|
| 7190 |
+
"loss": 1.3558,
|
| 7191 |
+
"step": 1018
|
| 7192 |
+
},
|
| 7193 |
+
{
|
| 7194 |
+
"epoch": 0.9372269487238446,
|
| 7195 |
+
"grad_norm": 0.5573782324790955,
|
| 7196 |
+
"learning_rate": 1.026370081018907e-06,
|
| 7197 |
+
"loss": 1.4016,
|
| 7198 |
+
"step": 1019
|
| 7199 |
+
},
|
| 7200 |
+
{
|
| 7201 |
+
"epoch": 0.9381467003908944,
|
| 7202 |
+
"grad_norm": 0.5528433322906494,
|
| 7203 |
+
"learning_rate": 9.969343475342285e-07,
|
| 7204 |
+
"loss": 1.3298,
|
| 7205 |
+
"step": 1020
|
| 7206 |
+
},
|
| 7207 |
+
{
|
| 7208 |
+
"epoch": 0.9390664520579444,
|
| 7209 |
+
"grad_norm": 0.573993980884552,
|
| 7210 |
+
"learning_rate": 9.679226280310982e-07,
|
| 7211 |
+
"loss": 1.3674,
|
| 7212 |
+
"step": 1021
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 0.9399862037249943,
|
| 7216 |
+
"grad_norm": 0.5446662902832031,
|
| 7217 |
+
"learning_rate": 9.393351735422773e-07,
|
| 7218 |
+
"loss": 1.3571,
|
| 7219 |
+
"step": 1022
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 0.9409059553920441,
|
| 7223 |
+
"grad_norm": 0.5892913937568665,
|
| 7224 |
+
"learning_rate": 9.111722314294358e-07,
|
| 7225 |
+
"loss": 1.3471,
|
| 7226 |
+
"step": 1023
|
| 7227 |
+
},
|
| 7228 |
+
{
|
| 7229 |
+
"epoch": 0.9418257070590941,
|
| 7230 |
+
"grad_norm": 0.6275593638420105,
|
| 7231 |
+
"learning_rate": 8.834340453810375e-07,
|
| 7232 |
+
"loss": 1.3269,
|
| 7233 |
+
"step": 1024
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 0.9427454587261439,
|
| 7237 |
+
"grad_norm": 0.6341751217842102,
|
| 7238 |
+
"learning_rate": 8.561208554101863e-07,
|
| 7239 |
+
"loss": 1.3899,
|
| 7240 |
+
"step": 1025
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 0.9436652103931938,
|
| 7244 |
+
"grad_norm": 0.6272470951080322,
|
| 7245 |
+
"learning_rate": 8.292328978526109e-07,
|
| 7246 |
+
"loss": 1.3545,
|
| 7247 |
+
"step": 1026
|
| 7248 |
+
},
|
| 7249 |
+
{
|
| 7250 |
+
"epoch": 0.9445849620602438,
|
| 7251 |
+
"grad_norm": 0.6651190519332886,
|
| 7252 |
+
"learning_rate": 8.027704053645613e-07,
|
| 7253 |
+
"loss": 1.3397,
|
| 7254 |
+
"step": 1027
|
| 7255 |
+
},
|
| 7256 |
+
{
|
| 7257 |
+
"epoch": 0.9455047137272936,
|
| 7258 |
+
"grad_norm": 0.6504070162773132,
|
| 7259 |
+
"learning_rate": 7.76733606920832e-07,
|
| 7260 |
+
"loss": 1.3889,
|
| 7261 |
+
"step": 1028
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 0.9464244653943436,
|
| 7265 |
+
"grad_norm": 0.639077365398407,
|
| 7266 |
+
"learning_rate": 7.511227278127697e-07,
|
| 7267 |
+
"loss": 1.3159,
|
| 7268 |
+
"step": 1029
|
| 7269 |
+
},
|
| 7270 |
+
{
|
| 7271 |
+
"epoch": 0.9473442170613934,
|
| 7272 |
+
"grad_norm": 0.685070812702179,
|
| 7273 |
+
"learning_rate": 7.259379896463247e-07,
|
| 7274 |
+
"loss": 1.312,
|
| 7275 |
+
"step": 1030
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 0.9482639687284433,
|
| 7279 |
+
"grad_norm": 0.705894947052002,
|
| 7280 |
+
"learning_rate": 7.011796103401191e-07,
|
| 7281 |
+
"loss": 1.325,
|
| 7282 |
+
"step": 1031
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 0.9491837203954933,
|
| 7286 |
+
"grad_norm": 0.6670310497283936,
|
| 7287 |
+
"learning_rate": 6.768478041236037e-07,
|
| 7288 |
+
"loss": 1.3582,
|
| 7289 |
+
"step": 1032
|
| 7290 |
+
},
|
| 7291 |
+
{
|
| 7292 |
+
"epoch": 0.9501034720625431,
|
| 7293 |
+
"grad_norm": 0.7927426695823669,
|
| 7294 |
+
"learning_rate": 6.529427815351374e-07,
|
| 7295 |
+
"loss": 1.3767,
|
| 7296 |
+
"step": 1033
|
| 7297 |
+
},
|
| 7298 |
+
{
|
| 7299 |
+
"epoch": 0.951023223729593,
|
| 7300 |
+
"grad_norm": 0.6605473160743713,
|
| 7301 |
+
"learning_rate": 6.294647494202444e-07,
|
| 7302 |
+
"loss": 1.2937,
|
| 7303 |
+
"step": 1034
|
| 7304 |
+
},
|
| 7305 |
+
{
|
| 7306 |
+
"epoch": 0.9519429753966429,
|
| 7307 |
+
"grad_norm": 0.599684476852417,
|
| 7308 |
+
"learning_rate": 6.064139109297485e-07,
|
| 7309 |
+
"loss": 1.2802,
|
| 7310 |
+
"step": 1035
|
| 7311 |
+
},
|
| 7312 |
+
{
|
| 7313 |
+
"epoch": 0.9528627270636928,
|
| 7314 |
+
"grad_norm": 0.6753445267677307,
|
| 7315 |
+
"learning_rate": 5.837904655180748e-07,
|
| 7316 |
+
"loss": 1.297,
|
| 7317 |
+
"step": 1036
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 0.9537824787307427,
|
| 7321 |
+
"grad_norm": 0.6682940125465393,
|
| 7322 |
+
"learning_rate": 5.615946089414736e-07,
|
| 7323 |
+
"loss": 1.3073,
|
| 7324 |
+
"step": 1037
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 0.9547022303977926,
|
| 7328 |
+
"grad_norm": 0.6744109392166138,
|
| 7329 |
+
"learning_rate": 5.398265332563934e-07,
|
| 7330 |
+
"loss": 1.1858,
|
| 7331 |
+
"step": 1038
|
| 7332 |
+
},
|
| 7333 |
+
{
|
| 7334 |
+
"epoch": 0.9556219820648425,
|
| 7335 |
+
"grad_norm": 0.6154145002365112,
|
| 7336 |
+
"learning_rate": 5.184864268177325e-07,
|
| 7337 |
+
"loss": 1.1648,
|
| 7338 |
+
"step": 1039
|
| 7339 |
+
},
|
| 7340 |
+
{
|
| 7341 |
+
"epoch": 0.9565417337318924,
|
| 7342 |
+
"grad_norm": 0.6836906671524048,
|
| 7343 |
+
"learning_rate": 4.975744742772848e-07,
|
| 7344 |
+
"loss": 1.2518,
|
| 7345 |
+
"step": 1040
|
| 7346 |
+
},
|
| 7347 |
+
{
|
| 7348 |
+
"epoch": 0.9574614853989423,
|
| 7349 |
+
"grad_norm": 0.6386029720306396,
|
| 7350 |
+
"learning_rate": 4.770908565820964e-07,
|
| 7351 |
+
"loss": 1.2142,
|
| 7352 |
+
"step": 1041
|
| 7353 |
+
},
|
| 7354 |
+
{
|
| 7355 |
+
"epoch": 0.9583812370659922,
|
| 7356 |
+
"grad_norm": 0.6528066992759705,
|
| 7357 |
+
"learning_rate": 4.5703575097292286e-07,
|
| 7358 |
+
"loss": 1.1931,
|
| 7359 |
+
"step": 1042
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 0.959300988733042,
|
| 7363 |
+
"grad_norm": 0.665433406829834,
|
| 7364 |
+
"learning_rate": 4.37409330982691e-07,
|
| 7365 |
+
"loss": 1.202,
|
| 7366 |
+
"step": 1043
|
| 7367 |
+
},
|
| 7368 |
+
{
|
| 7369 |
+
"epoch": 0.960220740400092,
|
| 7370 |
+
"grad_norm": 0.7009211182594299,
|
| 7371 |
+
"learning_rate": 4.182117664349783e-07,
|
| 7372 |
+
"loss": 1.2317,
|
| 7373 |
+
"step": 1044
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 0.9611404920671419,
|
| 7377 |
+
"grad_norm": 0.7533866167068481,
|
| 7378 |
+
"learning_rate": 3.99443223442586e-07,
|
| 7379 |
+
"loss": 1.2128,
|
| 7380 |
+
"step": 1045
|
| 7381 |
+
},
|
| 7382 |
+
{
|
| 7383 |
+
"epoch": 0.9620602437341917,
|
| 7384 |
+
"grad_norm": 0.7658700942993164,
|
| 7385 |
+
"learning_rate": 3.8110386440605164e-07,
|
| 7386 |
+
"loss": 1.1474,
|
| 7387 |
+
"step": 1046
|
| 7388 |
+
},
|
| 7389 |
+
{
|
| 7390 |
+
"epoch": 0.9629799954012417,
|
| 7391 |
+
"grad_norm": 0.7905300259590149,
|
| 7392 |
+
"learning_rate": 3.6319384801227763e-07,
|
| 7393 |
+
"loss": 1.1075,
|
| 7394 |
+
"step": 1047
|
| 7395 |
+
},
|
| 7396 |
+
{
|
| 7397 |
+
"epoch": 0.9638997470682915,
|
| 7398 |
+
"grad_norm": 0.9083186388015747,
|
| 7399 |
+
"learning_rate": 3.4571332923314936e-07,
|
| 7400 |
+
"loss": 1.1094,
|
| 7401 |
+
"step": 1048
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 0.9648194987353415,
|
| 7405 |
+
"grad_norm": 0.9923297762870789,
|
| 7406 |
+
"learning_rate": 3.2866245932418604e-07,
|
| 7407 |
+
"loss": 1.0341,
|
| 7408 |
+
"step": 1049
|
| 7409 |
+
},
|
| 7410 |
+
{
|
| 7411 |
+
"epoch": 0.9657392504023914,
|
| 7412 |
+
"grad_norm": 1.4956581592559814,
|
| 7413 |
+
"learning_rate": 3.120413858232474e-07,
|
| 7414 |
+
"loss": 0.9236,
|
| 7415 |
+
"step": 1050
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 0.9657392504023914,
|
| 7419 |
+
"eval_loss": 1.3224910497665405,
|
| 7420 |
+
"eval_runtime": 49.9198,
|
| 7421 |
+
"eval_samples_per_second": 165.065,
|
| 7422 |
+
"eval_steps_per_second": 20.633,
|
| 7423 |
+
"step": 1050
|
| 7424 |
}
|
| 7425 |
],
|
| 7426 |
"logging_steps": 1,
|
|
|
|
| 7449 |
"attributes": {}
|
| 7450 |
}
|
| 7451 |
},
|
| 7452 |
+
"total_flos": 1.9092013631668224e+17,
|
| 7453 |
"train_batch_size": 12,
|
| 7454 |
"trial_name": null,
|
| 7455 |
"trial_params": null
|