Training in progress, step 800, 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 335604696
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f1e1a151a5741a3af49bc404c3435bb2c87de0f0975a6c1caef52d07bc4df9bc
|
3 |
size 335604696
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 170920532
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e5ad5b608c3afe09851aad5be2f15a0fabc7e6f8e8b50eeb1b0b66f63b50b862
|
3 |
size 170920532
|
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:f984006bb3837e6a89eec0a2e65a93c4eb3466bdfa3832a69232239f8b636824
|
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 sha256:6cf13def61641dc944e9edbf416e7eb9f3aaa5a8dc09cbff5b8590b9b4880ce1
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
-
"best_model_checkpoint": "miner_id_24_1/checkpoint-
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 200,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4231,6 +4231,1414 @@
|
|
4231 |
"eval_samples_per_second": 8.764,
|
4232 |
"eval_steps_per_second": 2.191,
|
4233 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4234 |
}
|
4235 |
],
|
4236 |
"logging_steps": 1,
|
@@ -4259,7 +5667,7 @@
|
|
4259 |
"attributes": {}
|
4260 |
}
|
4261 |
},
|
4262 |
-
"total_flos": 1.
|
4263 |
"train_batch_size": 8,
|
4264 |
"trial_name": null,
|
4265 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.23512445390224457,
|
3 |
+
"best_model_checkpoint": "miner_id_24_1/checkpoint-800",
|
4 |
+
"epoch": 0.40738383195416933,
|
5 |
"eval_steps": 200,
|
6 |
+
"global_step": 800,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4231 |
"eval_samples_per_second": 8.764,
|
4232 |
"eval_steps_per_second": 2.191,
|
4233 |
"step": 600
|
4234 |
+
},
|
4235 |
+
{
|
4236 |
+
"epoch": 0.3060471037555697,
|
4237 |
+
"grad_norm": 5.079921722412109,
|
4238 |
+
"learning_rate": 3.5002545436149474e-05,
|
4239 |
+
"loss": 0.8877,
|
4240 |
+
"step": 601
|
4241 |
+
},
|
4242 |
+
{
|
4243 |
+
"epoch": 0.3065563335455124,
|
4244 |
+
"grad_norm": 5.308548927307129,
|
4245 |
+
"learning_rate": 3.485126066291364e-05,
|
4246 |
+
"loss": 0.8411,
|
4247 |
+
"step": 602
|
4248 |
+
},
|
4249 |
+
{
|
4250 |
+
"epoch": 0.30706556333545515,
|
4251 |
+
"grad_norm": 7.06653356552124,
|
4252 |
+
"learning_rate": 3.470012843731476e-05,
|
4253 |
+
"loss": 0.8479,
|
4254 |
+
"step": 603
|
4255 |
+
},
|
4256 |
+
{
|
4257 |
+
"epoch": 0.3075747931253978,
|
4258 |
+
"grad_norm": 2.4465208053588867,
|
4259 |
+
"learning_rate": 3.4549150281252636e-05,
|
4260 |
+
"loss": 0.0402,
|
4261 |
+
"step": 604
|
4262 |
+
},
|
4263 |
+
{
|
4264 |
+
"epoch": 0.30808402291534054,
|
4265 |
+
"grad_norm": 7.19598913192749,
|
4266 |
+
"learning_rate": 3.439832771507565e-05,
|
4267 |
+
"loss": 0.5183,
|
4268 |
+
"step": 605
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 0.30859325270528326,
|
4272 |
+
"grad_norm": 10.159460067749023,
|
4273 |
+
"learning_rate": 3.424766225756537e-05,
|
4274 |
+
"loss": 0.6912,
|
4275 |
+
"step": 606
|
4276 |
+
},
|
4277 |
+
{
|
4278 |
+
"epoch": 0.309102482495226,
|
4279 |
+
"grad_norm": 6.36456298828125,
|
4280 |
+
"learning_rate": 3.4097155425921254e-05,
|
4281 |
+
"loss": 0.4891,
|
4282 |
+
"step": 607
|
4283 |
+
},
|
4284 |
+
{
|
4285 |
+
"epoch": 0.3096117122851687,
|
4286 |
+
"grad_norm": 28.50994873046875,
|
4287 |
+
"learning_rate": 3.394680873574546e-05,
|
4288 |
+
"loss": 1.2461,
|
4289 |
+
"step": 608
|
4290 |
+
},
|
4291 |
+
{
|
4292 |
+
"epoch": 0.31012094207511137,
|
4293 |
+
"grad_norm": 11.220614433288574,
|
4294 |
+
"learning_rate": 3.3796623701027476e-05,
|
4295 |
+
"loss": 0.8775,
|
4296 |
+
"step": 609
|
4297 |
+
},
|
4298 |
+
{
|
4299 |
+
"epoch": 0.3106301718650541,
|
4300 |
+
"grad_norm": 9.175284385681152,
|
4301 |
+
"learning_rate": 3.364660183412892e-05,
|
4302 |
+
"loss": 0.8295,
|
4303 |
+
"step": 610
|
4304 |
+
},
|
4305 |
+
{
|
4306 |
+
"epoch": 0.3111394016549968,
|
4307 |
+
"grad_norm": 13.259340286254883,
|
4308 |
+
"learning_rate": 3.349674464576834e-05,
|
4309 |
+
"loss": 0.8886,
|
4310 |
+
"step": 611
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 0.31164863144493954,
|
4314 |
+
"grad_norm": 10.061495780944824,
|
4315 |
+
"learning_rate": 3.334705364500596e-05,
|
4316 |
+
"loss": 0.7599,
|
4317 |
+
"step": 612
|
4318 |
+
},
|
4319 |
+
{
|
4320 |
+
"epoch": 0.31215786123488226,
|
4321 |
+
"grad_norm": 4.028329849243164,
|
4322 |
+
"learning_rate": 3.3197530339228487e-05,
|
4323 |
+
"loss": 0.1656,
|
4324 |
+
"step": 613
|
4325 |
+
},
|
4326 |
+
{
|
4327 |
+
"epoch": 0.31266709102482493,
|
4328 |
+
"grad_norm": 2.669538736343384,
|
4329 |
+
"learning_rate": 3.304817623413397e-05,
|
4330 |
+
"loss": 0.124,
|
4331 |
+
"step": 614
|
4332 |
+
},
|
4333 |
+
{
|
4334 |
+
"epoch": 0.31317632081476765,
|
4335 |
+
"grad_norm": 3.3735191822052,
|
4336 |
+
"learning_rate": 3.289899283371657e-05,
|
4337 |
+
"loss": 0.1331,
|
4338 |
+
"step": 615
|
4339 |
+
},
|
4340 |
+
{
|
4341 |
+
"epoch": 0.3136855506047104,
|
4342 |
+
"grad_norm": 1.7686545848846436,
|
4343 |
+
"learning_rate": 3.274998164025148e-05,
|
4344 |
+
"loss": 0.1367,
|
4345 |
+
"step": 616
|
4346 |
+
},
|
4347 |
+
{
|
4348 |
+
"epoch": 0.3141947803946531,
|
4349 |
+
"grad_norm": 2.966285467147827,
|
4350 |
+
"learning_rate": 3.260114415427975e-05,
|
4351 |
+
"loss": 0.1389,
|
4352 |
+
"step": 617
|
4353 |
+
},
|
4354 |
+
{
|
4355 |
+
"epoch": 0.3147040101845958,
|
4356 |
+
"grad_norm": 6.92048454284668,
|
4357 |
+
"learning_rate": 3.2452481874593234e-05,
|
4358 |
+
"loss": 0.1157,
|
4359 |
+
"step": 618
|
4360 |
+
},
|
4361 |
+
{
|
4362 |
+
"epoch": 0.3152132399745385,
|
4363 |
+
"grad_norm": 8.68355655670166,
|
4364 |
+
"learning_rate": 3.230399629821942e-05,
|
4365 |
+
"loss": 0.1763,
|
4366 |
+
"step": 619
|
4367 |
+
},
|
4368 |
+
{
|
4369 |
+
"epoch": 0.3157224697644812,
|
4370 |
+
"grad_norm": 3.1240475177764893,
|
4371 |
+
"learning_rate": 3.215568892040641e-05,
|
4372 |
+
"loss": 0.1004,
|
4373 |
+
"step": 620
|
4374 |
+
},
|
4375 |
+
{
|
4376 |
+
"epoch": 0.31623169955442393,
|
4377 |
+
"grad_norm": 1.7762776613235474,
|
4378 |
+
"learning_rate": 3.200756123460788e-05,
|
4379 |
+
"loss": 0.0993,
|
4380 |
+
"step": 621
|
4381 |
+
},
|
4382 |
+
{
|
4383 |
+
"epoch": 0.31674092934436665,
|
4384 |
+
"grad_norm": 2.550874710083008,
|
4385 |
+
"learning_rate": 3.1859614732467954e-05,
|
4386 |
+
"loss": 0.0514,
|
4387 |
+
"step": 622
|
4388 |
+
},
|
4389 |
+
{
|
4390 |
+
"epoch": 0.3172501591343094,
|
4391 |
+
"grad_norm": 3.7702434062957764,
|
4392 |
+
"learning_rate": 3.171185090380628e-05,
|
4393 |
+
"loss": 0.0691,
|
4394 |
+
"step": 623
|
4395 |
+
},
|
4396 |
+
{
|
4397 |
+
"epoch": 0.3177593889242521,
|
4398 |
+
"grad_norm": 4.985774517059326,
|
4399 |
+
"learning_rate": 3.156427123660297e-05,
|
4400 |
+
"loss": 0.0982,
|
4401 |
+
"step": 624
|
4402 |
+
},
|
4403 |
+
{
|
4404 |
+
"epoch": 0.31826861871419476,
|
4405 |
+
"grad_norm": 9.22007942199707,
|
4406 |
+
"learning_rate": 3.141687721698363e-05,
|
4407 |
+
"loss": 0.123,
|
4408 |
+
"step": 625
|
4409 |
+
},
|
4410 |
+
{
|
4411 |
+
"epoch": 0.3187778485041375,
|
4412 |
+
"grad_norm": 3.031754970550537,
|
4413 |
+
"learning_rate": 3.12696703292044e-05,
|
4414 |
+
"loss": 0.0507,
|
4415 |
+
"step": 626
|
4416 |
+
},
|
4417 |
+
{
|
4418 |
+
"epoch": 0.3192870782940802,
|
4419 |
+
"grad_norm": 6.516983509063721,
|
4420 |
+
"learning_rate": 3.1122652055637015e-05,
|
4421 |
+
"loss": 0.0944,
|
4422 |
+
"step": 627
|
4423 |
+
},
|
4424 |
+
{
|
4425 |
+
"epoch": 0.31979630808402293,
|
4426 |
+
"grad_norm": 4.626400470733643,
|
4427 |
+
"learning_rate": 3.097582387675385e-05,
|
4428 |
+
"loss": 0.1333,
|
4429 |
+
"step": 628
|
4430 |
+
},
|
4431 |
+
{
|
4432 |
+
"epoch": 0.32030553787396565,
|
4433 |
+
"grad_norm": 5.953018665313721,
|
4434 |
+
"learning_rate": 3.082918727111304e-05,
|
4435 |
+
"loss": 0.1317,
|
4436 |
+
"step": 629
|
4437 |
+
},
|
4438 |
+
{
|
4439 |
+
"epoch": 0.3208147676639083,
|
4440 |
+
"grad_norm": 4.307436466217041,
|
4441 |
+
"learning_rate": 3.0682743715343564e-05,
|
4442 |
+
"loss": 0.0859,
|
4443 |
+
"step": 630
|
4444 |
+
},
|
4445 |
+
{
|
4446 |
+
"epoch": 0.32132399745385104,
|
4447 |
+
"grad_norm": 5.189202308654785,
|
4448 |
+
"learning_rate": 3.053649468413043e-05,
|
4449 |
+
"loss": 0.1205,
|
4450 |
+
"step": 631
|
4451 |
+
},
|
4452 |
+
{
|
4453 |
+
"epoch": 0.32183322724379376,
|
4454 |
+
"grad_norm": 7.541786193847656,
|
4455 |
+
"learning_rate": 3.0390441650199724e-05,
|
4456 |
+
"loss": 0.1567,
|
4457 |
+
"step": 632
|
4458 |
+
},
|
4459 |
+
{
|
4460 |
+
"epoch": 0.3223424570337365,
|
4461 |
+
"grad_norm": 5.107945442199707,
|
4462 |
+
"learning_rate": 3.0244586084303905e-05,
|
4463 |
+
"loss": 0.0836,
|
4464 |
+
"step": 633
|
4465 |
+
},
|
4466 |
+
{
|
4467 |
+
"epoch": 0.3228516868236792,
|
4468 |
+
"grad_norm": 5.2220563888549805,
|
4469 |
+
"learning_rate": 3.0098929455206904e-05,
|
4470 |
+
"loss": 0.0487,
|
4471 |
+
"step": 634
|
4472 |
+
},
|
4473 |
+
{
|
4474 |
+
"epoch": 0.3233609166136219,
|
4475 |
+
"grad_norm": 3.8281970024108887,
|
4476 |
+
"learning_rate": 2.9953473229669328e-05,
|
4477 |
+
"loss": 0.0783,
|
4478 |
+
"step": 635
|
4479 |
+
},
|
4480 |
+
{
|
4481 |
+
"epoch": 0.3238701464035646,
|
4482 |
+
"grad_norm": 4.100976943969727,
|
4483 |
+
"learning_rate": 2.9808218872433767e-05,
|
4484 |
+
"loss": 0.1336,
|
4485 |
+
"step": 636
|
4486 |
+
},
|
4487 |
+
{
|
4488 |
+
"epoch": 0.3243793761935073,
|
4489 |
+
"grad_norm": 4.79497766494751,
|
4490 |
+
"learning_rate": 2.9663167846209998e-05,
|
4491 |
+
"loss": 0.0657,
|
4492 |
+
"step": 637
|
4493 |
+
},
|
4494 |
+
{
|
4495 |
+
"epoch": 0.32488860598345004,
|
4496 |
+
"grad_norm": 5.689431190490723,
|
4497 |
+
"learning_rate": 2.9518321611660237e-05,
|
4498 |
+
"loss": 0.033,
|
4499 |
+
"step": 638
|
4500 |
+
},
|
4501 |
+
{
|
4502 |
+
"epoch": 0.32539783577339276,
|
4503 |
+
"grad_norm": 7.910229206085205,
|
4504 |
+
"learning_rate": 2.9373681627384447e-05,
|
4505 |
+
"loss": 0.0893,
|
4506 |
+
"step": 639
|
4507 |
+
},
|
4508 |
+
{
|
4509 |
+
"epoch": 0.32590706556333543,
|
4510 |
+
"grad_norm": 3.019529104232788,
|
4511 |
+
"learning_rate": 2.9229249349905684e-05,
|
4512 |
+
"loss": 0.0205,
|
4513 |
+
"step": 640
|
4514 |
+
},
|
4515 |
+
{
|
4516 |
+
"epoch": 0.32641629535327815,
|
4517 |
+
"grad_norm": 2.8111181259155273,
|
4518 |
+
"learning_rate": 2.9085026233655365e-05,
|
4519 |
+
"loss": 0.0133,
|
4520 |
+
"step": 641
|
4521 |
+
},
|
4522 |
+
{
|
4523 |
+
"epoch": 0.3269255251432209,
|
4524 |
+
"grad_norm": 4.1778974533081055,
|
4525 |
+
"learning_rate": 2.894101373095867e-05,
|
4526 |
+
"loss": 0.0162,
|
4527 |
+
"step": 642
|
4528 |
+
},
|
4529 |
+
{
|
4530 |
+
"epoch": 0.3274347549331636,
|
4531 |
+
"grad_norm": 40.197940826416016,
|
4532 |
+
"learning_rate": 2.8797213292019926e-05,
|
4533 |
+
"loss": 0.0561,
|
4534 |
+
"step": 643
|
4535 |
+
},
|
4536 |
+
{
|
4537 |
+
"epoch": 0.3279439847231063,
|
4538 |
+
"grad_norm": 5.798105239868164,
|
4539 |
+
"learning_rate": 2.8653626364907917e-05,
|
4540 |
+
"loss": 0.0468,
|
4541 |
+
"step": 644
|
4542 |
+
},
|
4543 |
+
{
|
4544 |
+
"epoch": 0.32845321451304904,
|
4545 |
+
"grad_norm": 0.09037820249795914,
|
4546 |
+
"learning_rate": 2.851025439554142e-05,
|
4547 |
+
"loss": 0.0004,
|
4548 |
+
"step": 645
|
4549 |
+
},
|
4550 |
+
{
|
4551 |
+
"epoch": 0.3289624443029917,
|
4552 |
+
"grad_norm": 6.750631809234619,
|
4553 |
+
"learning_rate": 2.8367098827674578e-05,
|
4554 |
+
"loss": 0.12,
|
4555 |
+
"step": 646
|
4556 |
+
},
|
4557 |
+
{
|
4558 |
+
"epoch": 0.32947167409293443,
|
4559 |
+
"grad_norm": 8.365553855895996,
|
4560 |
+
"learning_rate": 2.8224161102882397e-05,
|
4561 |
+
"loss": 0.0613,
|
4562 |
+
"step": 647
|
4563 |
+
},
|
4564 |
+
{
|
4565 |
+
"epoch": 0.32998090388287715,
|
4566 |
+
"grad_norm": 8.15538501739502,
|
4567 |
+
"learning_rate": 2.8081442660546125e-05,
|
4568 |
+
"loss": 0.1204,
|
4569 |
+
"step": 648
|
4570 |
+
},
|
4571 |
+
{
|
4572 |
+
"epoch": 0.3304901336728199,
|
4573 |
+
"grad_norm": 8.706940650939941,
|
4574 |
+
"learning_rate": 2.7938944937838923e-05,
|
4575 |
+
"loss": 0.0566,
|
4576 |
+
"step": 649
|
4577 |
+
},
|
4578 |
+
{
|
4579 |
+
"epoch": 0.3309993634627626,
|
4580 |
+
"grad_norm": 4.927478790283203,
|
4581 |
+
"learning_rate": 2.7796669369711294e-05,
|
4582 |
+
"loss": 0.1079,
|
4583 |
+
"step": 650
|
4584 |
+
},
|
4585 |
+
{
|
4586 |
+
"epoch": 0.33150859325270526,
|
4587 |
+
"grad_norm": 4.741428852081299,
|
4588 |
+
"learning_rate": 2.7654617388876615e-05,
|
4589 |
+
"loss": 0.7197,
|
4590 |
+
"step": 651
|
4591 |
+
},
|
4592 |
+
{
|
4593 |
+
"epoch": 0.332017823042648,
|
4594 |
+
"grad_norm": 4.9175944328308105,
|
4595 |
+
"learning_rate": 2.7512790425796718e-05,
|
4596 |
+
"loss": 0.5017,
|
4597 |
+
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.3325270528325907,
|
4601 |
+
"grad_norm": 3.583587884902954,
|
4602 |
+
"learning_rate": 2.7371189908667604e-05,
|
4603 |
+
"loss": 0.2062,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.33303628262253343,
|
4608 |
+
"grad_norm": 9.022587776184082,
|
4609 |
+
"learning_rate": 2.7229817263404866e-05,
|
4610 |
+
"loss": 0.7837,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.33354551241247615,
|
4615 |
+
"grad_norm": 9.906291007995605,
|
4616 |
+
"learning_rate": 2.708867391362948e-05,
|
4617 |
+
"loss": 0.6524,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.3340547422024188,
|
4622 |
+
"grad_norm": 5.258704662322998,
|
4623 |
+
"learning_rate": 2.694776128065345e-05,
|
4624 |
+
"loss": 0.3687,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.33456397199236154,
|
4629 |
+
"grad_norm": 7.302427768707275,
|
4630 |
+
"learning_rate": 2.6807080783465376e-05,
|
4631 |
+
"loss": 0.4876,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.33507320178230426,
|
4636 |
+
"grad_norm": 13.841546058654785,
|
4637 |
+
"learning_rate": 2.6666633838716314e-05,
|
4638 |
+
"loss": 1.4033,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.335582431572247,
|
4643 |
+
"grad_norm": 11.859368324279785,
|
4644 |
+
"learning_rate": 2.6526421860705473e-05,
|
4645 |
+
"loss": 0.9077,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.3360916613621897,
|
4650 |
+
"grad_norm": 13.534501075744629,
|
4651 |
+
"learning_rate": 2.638644626136587e-05,
|
4652 |
+
"loss": 0.7607,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.3366008911521324,
|
4657 |
+
"grad_norm": 7.9164958000183105,
|
4658 |
+
"learning_rate": 2.6246708450250256e-05,
|
4659 |
+
"loss": 0.599,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.3371101209420751,
|
4664 |
+
"grad_norm": 8.299764633178711,
|
4665 |
+
"learning_rate": 2.6107209834516854e-05,
|
4666 |
+
"loss": 1.0367,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.3376193507320178,
|
4671 |
+
"grad_norm": 4.36458683013916,
|
4672 |
+
"learning_rate": 2.596795181891514e-05,
|
4673 |
+
"loss": 0.4012,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.33812858052196054,
|
4678 |
+
"grad_norm": 2.5791258811950684,
|
4679 |
+
"learning_rate": 2.5828935805771802e-05,
|
4680 |
+
"loss": 0.1412,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.33863781031190326,
|
4685 |
+
"grad_norm": 4.219427585601807,
|
4686 |
+
"learning_rate": 2.5690163194976575e-05,
|
4687 |
+
"loss": 0.1824,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.33914704010184593,
|
4692 |
+
"grad_norm": 2.9232358932495117,
|
4693 |
+
"learning_rate": 2.5551635383968065e-05,
|
4694 |
+
"loss": 0.17,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.33965626989178865,
|
4699 |
+
"grad_norm": 2.1133458614349365,
|
4700 |
+
"learning_rate": 2.5413353767719805e-05,
|
4701 |
+
"loss": 0.1109,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.3401654996817314,
|
4706 |
+
"grad_norm": 0.8237698078155518,
|
4707 |
+
"learning_rate": 2.5275319738726165e-05,
|
4708 |
+
"loss": 0.0692,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.3406747294716741,
|
4713 |
+
"grad_norm": 2.2106404304504395,
|
4714 |
+
"learning_rate": 2.513753468698826e-05,
|
4715 |
+
"loss": 0.1241,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.3411839592616168,
|
4720 |
+
"grad_norm": 2.5900228023529053,
|
4721 |
+
"learning_rate": 2.500000000000001e-05,
|
4722 |
+
"loss": 0.1617,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.34169318905155954,
|
4727 |
+
"grad_norm": 0.9092320799827576,
|
4728 |
+
"learning_rate": 2.486271706273421e-05,
|
4729 |
+
"loss": 0.1002,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.3422024188415022,
|
4734 |
+
"grad_norm": 3.5675482749938965,
|
4735 |
+
"learning_rate": 2.4725687257628534e-05,
|
4736 |
+
"loss": 0.0889,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.34271164863144493,
|
4741 |
+
"grad_norm": 4.1621479988098145,
|
4742 |
+
"learning_rate": 2.4588911964571553e-05,
|
4743 |
+
"loss": 0.119,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.34322087842138765,
|
4748 |
+
"grad_norm": 1.428825855255127,
|
4749 |
+
"learning_rate": 2.4452392560888976e-05,
|
4750 |
+
"loss": 0.0871,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.3437301082113304,
|
4755 |
+
"grad_norm": 4.961071968078613,
|
4756 |
+
"learning_rate": 2.4316130421329697e-05,
|
4757 |
+
"loss": 0.0718,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.3442393380012731,
|
4762 |
+
"grad_norm": 5.432969570159912,
|
4763 |
+
"learning_rate": 2.418012691805191e-05,
|
4764 |
+
"loss": 0.0782,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.34474856779121577,
|
4769 |
+
"grad_norm": 4.57743501663208,
|
4770 |
+
"learning_rate": 2.4044383420609406e-05,
|
4771 |
+
"loss": 0.1596,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.3452577975811585,
|
4776 |
+
"grad_norm": 7.462673187255859,
|
4777 |
+
"learning_rate": 2.3908901295937713e-05,
|
4778 |
+
"loss": 0.1173,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.3457670273711012,
|
4783 |
+
"grad_norm": 4.732002258300781,
|
4784 |
+
"learning_rate": 2.3773681908340284e-05,
|
4785 |
+
"loss": 0.111,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.34627625716104393,
|
4790 |
+
"grad_norm": 5.835433483123779,
|
4791 |
+
"learning_rate": 2.363872661947488e-05,
|
4792 |
+
"loss": 0.1395,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.34678548695098665,
|
4797 |
+
"grad_norm": 4.240106105804443,
|
4798 |
+
"learning_rate": 2.350403678833976e-05,
|
4799 |
+
"loss": 0.0979,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.3472947167409293,
|
4804 |
+
"grad_norm": 6.862554550170898,
|
4805 |
+
"learning_rate": 2.336961377126001e-05,
|
4806 |
+
"loss": 0.1356,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.34780394653087204,
|
4811 |
+
"grad_norm": 5.05124568939209,
|
4812 |
+
"learning_rate": 2.3235458921873925e-05,
|
4813 |
+
"loss": 0.1112,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.34831317632081477,
|
4818 |
+
"grad_norm": 5.341157913208008,
|
4819 |
+
"learning_rate": 2.310157359111938e-05,
|
4820 |
+
"loss": 0.0908,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.3488224061107575,
|
4825 |
+
"grad_norm": 3.359665870666504,
|
4826 |
+
"learning_rate": 2.296795912722014e-05,
|
4827 |
+
"loss": 0.081,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.3493316359007002,
|
4832 |
+
"grad_norm": 4.865165710449219,
|
4833 |
+
"learning_rate": 2.283461687567236e-05,
|
4834 |
+
"loss": 0.1172,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.3498408656906429,
|
4839 |
+
"grad_norm": 8.436277389526367,
|
4840 |
+
"learning_rate": 2.2701548179231048e-05,
|
4841 |
+
"loss": 0.1962,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.3503500954805856,
|
4846 |
+
"grad_norm": 6.217014312744141,
|
4847 |
+
"learning_rate": 2.2568754377896516e-05,
|
4848 |
+
"loss": 0.0362,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.3508593252705283,
|
4853 |
+
"grad_norm": 19.41240119934082,
|
4854 |
+
"learning_rate": 2.2436236808900844e-05,
|
4855 |
+
"loss": 0.275,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.35136855506047104,
|
4860 |
+
"grad_norm": 6.9711456298828125,
|
4861 |
+
"learning_rate": 2.2303996806694488e-05,
|
4862 |
+
"loss": 0.0658,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.35187778485041377,
|
4867 |
+
"grad_norm": 6.803055763244629,
|
4868 |
+
"learning_rate": 2.2172035702932825e-05,
|
4869 |
+
"loss": 0.0384,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.3523870146403565,
|
4874 |
+
"grad_norm": 3.9196035861968994,
|
4875 |
+
"learning_rate": 2.2040354826462668e-05,
|
4876 |
+
"loss": 0.0957,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.35289624443029916,
|
4881 |
+
"grad_norm": 3.145606756210327,
|
4882 |
+
"learning_rate": 2.1908955503308993e-05,
|
4883 |
+
"loss": 0.0623,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.3534054742202419,
|
4888 |
+
"grad_norm": 16.734477996826172,
|
4889 |
+
"learning_rate": 2.1777839056661554e-05,
|
4890 |
+
"loss": 0.0769,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.3539147040101846,
|
4895 |
+
"grad_norm": 9.658271789550781,
|
4896 |
+
"learning_rate": 2.164700680686147e-05,
|
4897 |
+
"loss": 0.0265,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.3544239338001273,
|
4902 |
+
"grad_norm": 1.6071585416793823,
|
4903 |
+
"learning_rate": 2.1516460071388062e-05,
|
4904 |
+
"loss": 0.0115,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.35493316359007004,
|
4909 |
+
"grad_norm": 4.15781307220459,
|
4910 |
+
"learning_rate": 2.1386200164845526e-05,
|
4911 |
+
"loss": 0.0719,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.3554423933800127,
|
4916 |
+
"grad_norm": 1.0766063928604126,
|
4917 |
+
"learning_rate": 2.125622839894964e-05,
|
4918 |
+
"loss": 0.0062,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.35595162316995543,
|
4923 |
+
"grad_norm": 3.8494861125946045,
|
4924 |
+
"learning_rate": 2.1126546082514664e-05,
|
4925 |
+
"loss": 0.0104,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.35646085295989816,
|
4930 |
+
"grad_norm": 4.333930015563965,
|
4931 |
+
"learning_rate": 2.09971545214401e-05,
|
4932 |
+
"loss": 0.069,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.3569700827498409,
|
4937 |
+
"grad_norm": 5.104018211364746,
|
4938 |
+
"learning_rate": 2.086805501869749e-05,
|
4939 |
+
"loss": 0.735,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.3574793125397836,
|
4944 |
+
"grad_norm": 4.622631072998047,
|
4945 |
+
"learning_rate": 2.073924887431744e-05,
|
4946 |
+
"loss": 0.6267,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.35798854232972627,
|
4951 |
+
"grad_norm": 2.708951234817505,
|
4952 |
+
"learning_rate": 2.061073738537635e-05,
|
4953 |
+
"loss": 0.1031,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.358497772119669,
|
4958 |
+
"grad_norm": 6.911785125732422,
|
4959 |
+
"learning_rate": 2.048252184598352e-05,
|
4960 |
+
"loss": 0.5889,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.3590070019096117,
|
4965 |
+
"grad_norm": 8.187259674072266,
|
4966 |
+
"learning_rate": 2.0354603547267985e-05,
|
4967 |
+
"loss": 0.5466,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.35951623169955443,
|
4972 |
+
"grad_norm": 6.681227207183838,
|
4973 |
+
"learning_rate": 2.0226983777365604e-05,
|
4974 |
+
"loss": 0.5335,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.36002546148949716,
|
4979 |
+
"grad_norm": 13.172379493713379,
|
4980 |
+
"learning_rate": 2.0099663821406056e-05,
|
4981 |
+
"loss": 0.6291,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.3605346912794398,
|
4986 |
+
"grad_norm": 10.422917366027832,
|
4987 |
+
"learning_rate": 1.9972644961499854e-05,
|
4988 |
+
"loss": 0.8471,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.36104392106938255,
|
4993 |
+
"grad_norm": 13.669676780700684,
|
4994 |
+
"learning_rate": 1.9845928476725524e-05,
|
4995 |
+
"loss": 1.0402,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.36155315085932527,
|
5000 |
+
"grad_norm": 6.9141716957092285,
|
5001 |
+
"learning_rate": 1.9719515643116674e-05,
|
5002 |
+
"loss": 0.6035,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.362062380649268,
|
5007 |
+
"grad_norm": 8.188703536987305,
|
5008 |
+
"learning_rate": 1.959340773364911e-05,
|
5009 |
+
"loss": 0.6699,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.3625716104392107,
|
5014 |
+
"grad_norm": 7.684673309326172,
|
5015 |
+
"learning_rate": 1.946760601822809e-05,
|
5016 |
+
"loss": 0.7657,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.3630808402291534,
|
5021 |
+
"grad_norm": 2.5300745964050293,
|
5022 |
+
"learning_rate": 1.9342111763675512e-05,
|
5023 |
+
"loss": 0.1569,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.3635900700190961,
|
5028 |
+
"grad_norm": 1.3801597356796265,
|
5029 |
+
"learning_rate": 1.9216926233717085e-05,
|
5030 |
+
"loss": 0.081,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.3640992998090388,
|
5035 |
+
"grad_norm": 1.2587306499481201,
|
5036 |
+
"learning_rate": 1.9092050688969738e-05,
|
5037 |
+
"loss": 0.1421,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.36460852959898155,
|
5042 |
+
"grad_norm": 1.1372148990631104,
|
5043 |
+
"learning_rate": 1.8967486386928817e-05,
|
5044 |
+
"loss": 0.0961,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.36511775938892427,
|
5049 |
+
"grad_norm": 0.7794922590255737,
|
5050 |
+
"learning_rate": 1.8843234581955442e-05,
|
5051 |
+
"loss": 0.1008,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.365626989178867,
|
5056 |
+
"grad_norm": 1.1420189142227173,
|
5057 |
+
"learning_rate": 1.8719296525263922e-05,
|
5058 |
+
"loss": 0.1071,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.36613621896880966,
|
5063 |
+
"grad_norm": 1.6180529594421387,
|
5064 |
+
"learning_rate": 1.859567346490913e-05,
|
5065 |
+
"loss": 0.0875,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.3666454487587524,
|
5070 |
+
"grad_norm": 1.0748051404953003,
|
5071 |
+
"learning_rate": 1.847236664577389e-05,
|
5072 |
+
"loss": 0.1288,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.3671546785486951,
|
5077 |
+
"grad_norm": 1.0483026504516602,
|
5078 |
+
"learning_rate": 1.8349377309556486e-05,
|
5079 |
+
"loss": 0.0774,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.3676639083386378,
|
5084 |
+
"grad_norm": 1.506415843963623,
|
5085 |
+
"learning_rate": 1.8226706694758195e-05,
|
5086 |
+
"loss": 0.1362,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.36817313812858055,
|
5091 |
+
"grad_norm": 1.0291399955749512,
|
5092 |
+
"learning_rate": 1.810435603667075e-05,
|
5093 |
+
"loss": 0.1071,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.3686823679185232,
|
5098 |
+
"grad_norm": 5.942626953125,
|
5099 |
+
"learning_rate": 1.7982326567363888e-05,
|
5100 |
+
"loss": 0.1614,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.36919159770846594,
|
5105 |
+
"grad_norm": 12.163902282714844,
|
5106 |
+
"learning_rate": 1.7860619515673033e-05,
|
5107 |
+
"loss": 0.1458,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.36970082749840866,
|
5112 |
+
"grad_norm": 3.687087059020996,
|
5113 |
+
"learning_rate": 1.773923610718686e-05,
|
5114 |
+
"loss": 0.0919,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.3702100572883514,
|
5119 |
+
"grad_norm": 4.903958320617676,
|
5120 |
+
"learning_rate": 1.7618177564234905e-05,
|
5121 |
+
"loss": 0.1254,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.3707192870782941,
|
5126 |
+
"grad_norm": 4.808300495147705,
|
5127 |
+
"learning_rate": 1.7497445105875377e-05,
|
5128 |
+
"loss": 0.1473,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.37122851686823677,
|
5133 |
+
"grad_norm": 5.668433666229248,
|
5134 |
+
"learning_rate": 1.73770399478828e-05,
|
5135 |
+
"loss": 0.1524,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.3717377466581795,
|
5140 |
+
"grad_norm": 8.04811954498291,
|
5141 |
+
"learning_rate": 1.725696330273575e-05,
|
5142 |
+
"loss": 0.0919,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.3722469764481222,
|
5147 |
+
"grad_norm": 4.406056880950928,
|
5148 |
+
"learning_rate": 1.7137216379604727e-05,
|
5149 |
+
"loss": 0.1046,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.37275620623806494,
|
5154 |
+
"grad_norm": 7.855915546417236,
|
5155 |
+
"learning_rate": 1.7017800384339928e-05,
|
5156 |
+
"loss": 0.0709,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.37326543602800766,
|
5161 |
+
"grad_norm": 3.947338342666626,
|
5162 |
+
"learning_rate": 1.6898716519459074e-05,
|
5163 |
+
"loss": 0.072,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.3737746658179503,
|
5168 |
+
"grad_norm": 7.404916763305664,
|
5169 |
+
"learning_rate": 1.6779965984135377e-05,
|
5170 |
+
"loss": 0.1313,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.37428389560789305,
|
5175 |
+
"grad_norm": 4.375075817108154,
|
5176 |
+
"learning_rate": 1.6661549974185424e-05,
|
5177 |
+
"loss": 0.0829,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.37479312539783577,
|
5182 |
+
"grad_norm": 4.642003536224365,
|
5183 |
+
"learning_rate": 1.6543469682057106e-05,
|
5184 |
+
"loss": 0.0602,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.3753023551877785,
|
5189 |
+
"grad_norm": 5.987782001495361,
|
5190 |
+
"learning_rate": 1.6425726296817633e-05,
|
5191 |
+
"loss": 0.0868,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.3758115849777212,
|
5196 |
+
"grad_norm": 10.39226245880127,
|
5197 |
+
"learning_rate": 1.6308321004141607e-05,
|
5198 |
+
"loss": 0.1273,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.37632081476766394,
|
5203 |
+
"grad_norm": 6.080224514007568,
|
5204 |
+
"learning_rate": 1.619125498629904e-05,
|
5205 |
+
"loss": 0.0465,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.3768300445576066,
|
5210 |
+
"grad_norm": 4.397584915161133,
|
5211 |
+
"learning_rate": 1.60745294221434e-05,
|
5212 |
+
"loss": 0.0625,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.3773392743475493,
|
5217 |
+
"grad_norm": 8.622246742248535,
|
5218 |
+
"learning_rate": 1.595814548709983e-05,
|
5219 |
+
"loss": 0.0664,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.37784850413749205,
|
5224 |
+
"grad_norm": 8.313824653625488,
|
5225 |
+
"learning_rate": 1.5842104353153287e-05,
|
5226 |
+
"loss": 0.0271,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.37835773392743477,
|
5231 |
+
"grad_norm": 2.949397325515747,
|
5232 |
+
"learning_rate": 1.5726407188836673e-05,
|
5233 |
+
"loss": 0.0345,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.3788669637173775,
|
5238 |
+
"grad_norm": 3.27532696723938,
|
5239 |
+
"learning_rate": 1.5611055159219152e-05,
|
5240 |
+
"loss": 0.0561,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.37937619350732016,
|
5245 |
+
"grad_norm": 0.19333244860172272,
|
5246 |
+
"learning_rate": 1.549604942589441e-05,
|
5247 |
+
"loss": 0.0013,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.3798854232972629,
|
5252 |
+
"grad_norm": 4.911413192749023,
|
5253 |
+
"learning_rate": 1.5381391146968866e-05,
|
5254 |
+
"loss": 0.0187,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.3803946530872056,
|
5259 |
+
"grad_norm": 6.5160627365112305,
|
5260 |
+
"learning_rate": 1.526708147705013e-05,
|
5261 |
+
"loss": 0.0928,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.3809038828771483,
|
5266 |
+
"grad_norm": 5.048818111419678,
|
5267 |
+
"learning_rate": 1.5153121567235335e-05,
|
5268 |
+
"loss": 0.013,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.38141311266709105,
|
5273 |
+
"grad_norm": 4.852001667022705,
|
5274 |
+
"learning_rate": 1.5039512565099467e-05,
|
5275 |
+
"loss": 0.079,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.3819223424570337,
|
5280 |
+
"grad_norm": 3.08077335357666,
|
5281 |
+
"learning_rate": 1.4926255614683932e-05,
|
5282 |
+
"loss": 0.0403,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.38243157224697644,
|
5287 |
+
"grad_norm": 4.824317932128906,
|
5288 |
+
"learning_rate": 1.481335185648498e-05,
|
5289 |
+
"loss": 0.7268,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.38294080203691916,
|
5294 |
+
"grad_norm": 4.761933326721191,
|
5295 |
+
"learning_rate": 1.4700802427442179e-05,
|
5296 |
+
"loss": 0.5991,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.3834500318268619,
|
5301 |
+
"grad_norm": 4.185091495513916,
|
5302 |
+
"learning_rate": 1.458860846092705e-05,
|
5303 |
+
"loss": 0.2459,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.3839592616168046,
|
5308 |
+
"grad_norm": 10.620789527893066,
|
5309 |
+
"learning_rate": 1.4476771086731567e-05,
|
5310 |
+
"loss": 0.79,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.38446849140674727,
|
5315 |
+
"grad_norm": 11.214977264404297,
|
5316 |
+
"learning_rate": 1.4365291431056871e-05,
|
5317 |
+
"loss": 0.7926,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.38497772119669,
|
5322 |
+
"grad_norm": 5.3124895095825195,
|
5323 |
+
"learning_rate": 1.4254170616501827e-05,
|
5324 |
+
"loss": 0.4243,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.3854869509866327,
|
5329 |
+
"grad_norm": 11.870231628417969,
|
5330 |
+
"learning_rate": 1.414340976205183e-05,
|
5331 |
+
"loss": 0.6992,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.38599618077657544,
|
5336 |
+
"grad_norm": 9.170869827270508,
|
5337 |
+
"learning_rate": 1.4033009983067452e-05,
|
5338 |
+
"loss": 0.8608,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.38650541056651816,
|
5343 |
+
"grad_norm": 10.806239128112793,
|
5344 |
+
"learning_rate": 1.3922972391273226e-05,
|
5345 |
+
"loss": 0.8629,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.3870146403564608,
|
5350 |
+
"grad_norm": 9.26331901550293,
|
5351 |
+
"learning_rate": 1.3813298094746491e-05,
|
5352 |
+
"loss": 0.5738,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.38752387014640355,
|
5357 |
+
"grad_norm": 10.236473083496094,
|
5358 |
+
"learning_rate": 1.3703988197906209e-05,
|
5359 |
+
"loss": 0.6671,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.38803309993634627,
|
5364 |
+
"grad_norm": 6.966268539428711,
|
5365 |
+
"learning_rate": 1.3595043801501794e-05,
|
5366 |
+
"loss": 0.7432,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.388542329726289,
|
5371 |
+
"grad_norm": 5.45960807800293,
|
5372 |
+
"learning_rate": 1.3486466002602133e-05,
|
5373 |
+
"loss": 0.5434,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.3890515595162317,
|
5378 |
+
"grad_norm": 1.5403192043304443,
|
5379 |
+
"learning_rate": 1.3378255894584463e-05,
|
5380 |
+
"loss": 0.1034,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.38956078930617444,
|
5385 |
+
"grad_norm": 1.8921170234680176,
|
5386 |
+
"learning_rate": 1.327041456712334e-05,
|
5387 |
+
"loss": 0.084,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.3900700190961171,
|
5392 |
+
"grad_norm": 4.200987815856934,
|
5393 |
+
"learning_rate": 1.3162943106179749e-05,
|
5394 |
+
"loss": 0.1189,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.3905792488860598,
|
5399 |
+
"grad_norm": 2.609253406524658,
|
5400 |
+
"learning_rate": 1.3055842593990131e-05,
|
5401 |
+
"loss": 0.2064,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.39108847867600255,
|
5406 |
+
"grad_norm": 2.868013858795166,
|
5407 |
+
"learning_rate": 1.2949114109055415e-05,
|
5408 |
+
"loss": 0.1238,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.39159770846594527,
|
5413 |
+
"grad_norm": 2.5417683124542236,
|
5414 |
+
"learning_rate": 1.2842758726130283e-05,
|
5415 |
+
"loss": 0.0929,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.392106938255888,
|
5420 |
+
"grad_norm": 1.0600427389144897,
|
5421 |
+
"learning_rate": 1.2736777516212266e-05,
|
5422 |
+
"loss": 0.1323,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.39261616804583066,
|
5427 |
+
"grad_norm": 1.9886938333511353,
|
5428 |
+
"learning_rate": 1.2631171546530968e-05,
|
5429 |
+
"loss": 0.0904,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.3931253978357734,
|
5434 |
+
"grad_norm": 0.7915336489677429,
|
5435 |
+
"learning_rate": 1.2525941880537307e-05,
|
5436 |
+
"loss": 0.0964,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.3936346276257161,
|
5441 |
+
"grad_norm": 2.3213069438934326,
|
5442 |
+
"learning_rate": 1.2421089577892869e-05,
|
5443 |
+
"loss": 0.1248,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.3941438574156588,
|
5448 |
+
"grad_norm": 3.6811985969543457,
|
5449 |
+
"learning_rate": 1.2316615694459189e-05,
|
5450 |
+
"loss": 0.0861,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.39465308720560155,
|
5455 |
+
"grad_norm": 4.516244411468506,
|
5456 |
+
"learning_rate": 1.2212521282287092e-05,
|
5457 |
+
"loss": 0.1168,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.3951623169955442,
|
5462 |
+
"grad_norm": 4.109873294830322,
|
5463 |
+
"learning_rate": 1.2108807389606158e-05,
|
5464 |
+
"loss": 0.1002,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.39567154678548694,
|
5469 |
+
"grad_norm": 4.258279323577881,
|
5470 |
+
"learning_rate": 1.2005475060814159e-05,
|
5471 |
+
"loss": 0.1388,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.39618077657542966,
|
5476 |
+
"grad_norm": 3.9544007778167725,
|
5477 |
+
"learning_rate": 1.1902525336466464e-05,
|
5478 |
+
"loss": 0.1318,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.3966900063653724,
|
5483 |
+
"grad_norm": 5.926074981689453,
|
5484 |
+
"learning_rate": 1.1799959253265668e-05,
|
5485 |
+
"loss": 0.1125,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.3971992361553151,
|
5490 |
+
"grad_norm": 8.683218955993652,
|
5491 |
+
"learning_rate": 1.1697777844051105e-05,
|
5492 |
+
"loss": 0.0995,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.39770846594525777,
|
5497 |
+
"grad_norm": 3.9073028564453125,
|
5498 |
+
"learning_rate": 1.1595982137788403e-05,
|
5499 |
+
"loss": 0.0903,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.3982176957352005,
|
5504 |
+
"grad_norm": 5.088838577270508,
|
5505 |
+
"learning_rate": 1.1494573159559213e-05,
|
5506 |
+
"loss": 0.1193,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.3987269255251432,
|
5511 |
+
"grad_norm": 5.782503128051758,
|
5512 |
+
"learning_rate": 1.1393551930550828e-05,
|
5513 |
+
"loss": 0.0823,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.39923615531508594,
|
5518 |
+
"grad_norm": 5.548112869262695,
|
5519 |
+
"learning_rate": 1.1292919468045877e-05,
|
5520 |
+
"loss": 0.0797,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.39974538510502866,
|
5525 |
+
"grad_norm": 6.514894962310791,
|
5526 |
+
"learning_rate": 1.1192676785412154e-05,
|
5527 |
+
"loss": 0.1448,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.4002546148949714,
|
5532 |
+
"grad_norm": 4.215356826782227,
|
5533 |
+
"learning_rate": 1.1092824892092373e-05,
|
5534 |
+
"loss": 0.0659,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.40076384468491405,
|
5539 |
+
"grad_norm": 7.318501949310303,
|
5540 |
+
"learning_rate": 1.099336479359398e-05,
|
5541 |
+
"loss": 0.1557,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.4012730744748568,
|
5546 |
+
"grad_norm": 4.508605003356934,
|
5547 |
+
"learning_rate": 1.0894297491479045e-05,
|
5548 |
+
"loss": 0.0383,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.4017823042647995,
|
5553 |
+
"grad_norm": 2.6559200286865234,
|
5554 |
+
"learning_rate": 1.0795623983354215e-05,
|
5555 |
+
"loss": 0.0212,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.4022915340547422,
|
5560 |
+
"grad_norm": 1.4912053346633911,
|
5561 |
+
"learning_rate": 1.0697345262860636e-05,
|
5562 |
+
"loss": 0.0104,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.40280076384468494,
|
5567 |
+
"grad_norm": 7.236735820770264,
|
5568 |
+
"learning_rate": 1.0599462319663905e-05,
|
5569 |
+
"loss": 0.0706,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.4033099936346276,
|
5574 |
+
"grad_norm": 4.715954303741455,
|
5575 |
+
"learning_rate": 1.0501976139444191e-05,
|
5576 |
+
"loss": 0.048,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.40381922342457033,
|
5581 |
+
"grad_norm": 2.1985514163970947,
|
5582 |
+
"learning_rate": 1.0404887703886251e-05,
|
5583 |
+
"loss": 0.0125,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.40432845321451305,
|
5588 |
+
"grad_norm": 7.244698524475098,
|
5589 |
+
"learning_rate": 1.0308197990669538e-05,
|
5590 |
+
"loss": 0.0468,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.4048376830044558,
|
5595 |
+
"grad_norm": 1.7899662256240845,
|
5596 |
+
"learning_rate": 1.021190797345839e-05,
|
5597 |
+
"loss": 0.0033,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.4053469127943985,
|
5602 |
+
"grad_norm": 7.417436122894287,
|
5603 |
+
"learning_rate": 1.0116018621892237e-05,
|
5604 |
+
"loss": 0.1219,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.40585614258434116,
|
5609 |
+
"grad_norm": 2.332493782043457,
|
5610 |
+
"learning_rate": 1.0020530901575754e-05,
|
5611 |
+
"loss": 0.0177,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.4063653723742839,
|
5616 |
+
"grad_norm": 6.933743000030518,
|
5617 |
+
"learning_rate": 9.92544577406923e-06,
|
5618 |
+
"loss": 0.0993,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.4068746021642266,
|
5623 |
+
"grad_norm": 6.604279518127441,
|
5624 |
+
"learning_rate": 9.830764196878872e-06,
|
5625 |
+
"loss": 0.0523,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.40738383195416933,
|
5630 |
+
"grad_norm": 1.6020522117614746,
|
5631 |
+
"learning_rate": 9.73648712344707e-06,
|
5632 |
+
"loss": 0.0222,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.40738383195416933,
|
5637 |
+
"eval_loss": 0.23512445390224457,
|
5638 |
+
"eval_runtime": 376.2064,
|
5639 |
+
"eval_samples_per_second": 8.793,
|
5640 |
+
"eval_steps_per_second": 2.198,
|
5641 |
+
"step": 800
|
5642 |
}
|
5643 |
],
|
5644 |
"logging_steps": 1,
|
|
|
5667 |
"attributes": {}
|
5668 |
}
|
5669 |
},
|
5670 |
+
"total_flos": 1.6645191524564337e+18,
|
5671 |
"train_batch_size": 8,
|
5672 |
"trial_name": null,
|
5673 |
"trial_params": null
|