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 190757752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6852ee84275e7748b2a829103df9a363a97e6752106c6fae85e757ffd629b267
|
3 |
size 190757752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 97268276
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:61a9234a44f8fb1e04d6156ab6acd9d93b79330fbf529f367ee295021e280c31
|
3 |
size 97268276
|
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 sha256:198a91eb5a9c060c92cf14d26b19d6053f56ec2284335b64a07562732574666d
|
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/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": 35.062,
|
4232 |
"eval_steps_per_second": 8.769,
|
4233 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4234 |
}
|
4235 |
],
|
4236 |
"logging_steps": 1,
|
@@ -4259,7 +5667,7 @@
|
|
4259 |
"attributes": {}
|
4260 |
}
|
4261 |
},
|
4262 |
-
"total_flos":
|
4263 |
"train_batch_size": 8,
|
4264 |
"trial_name": null,
|
4265 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.040625762194395065,
|
3 |
+
"best_model_checkpoint": "miner_id_24/checkpoint-800",
|
4 |
+
"epoch": 0.5544005544005544,
|
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": 35.062,
|
4232 |
"eval_steps_per_second": 8.769,
|
4233 |
"step": 600
|
4234 |
+
},
|
4235 |
+
{
|
4236 |
+
"epoch": 0.4164934164934165,
|
4237 |
+
"grad_norm": 2.0430245399475098,
|
4238 |
+
"learning_rate": 3.5002545436149474e-05,
|
4239 |
+
"loss": 0.1009,
|
4240 |
+
"step": 601
|
4241 |
+
},
|
4242 |
+
{
|
4243 |
+
"epoch": 0.4171864171864172,
|
4244 |
+
"grad_norm": 2.058833599090576,
|
4245 |
+
"learning_rate": 3.485126066291364e-05,
|
4246 |
+
"loss": 0.0907,
|
4247 |
+
"step": 602
|
4248 |
+
},
|
4249 |
+
{
|
4250 |
+
"epoch": 0.4178794178794179,
|
4251 |
+
"grad_norm": 1.4888850450515747,
|
4252 |
+
"learning_rate": 3.470012843731476e-05,
|
4253 |
+
"loss": 0.0599,
|
4254 |
+
"step": 603
|
4255 |
+
},
|
4256 |
+
{
|
4257 |
+
"epoch": 0.41857241857241856,
|
4258 |
+
"grad_norm": 2.087881565093994,
|
4259 |
+
"learning_rate": 3.4549150281252636e-05,
|
4260 |
+
"loss": 0.0755,
|
4261 |
+
"step": 604
|
4262 |
+
},
|
4263 |
+
{
|
4264 |
+
"epoch": 0.41926541926541927,
|
4265 |
+
"grad_norm": 4.713494300842285,
|
4266 |
+
"learning_rate": 3.439832771507565e-05,
|
4267 |
+
"loss": 0.2284,
|
4268 |
+
"step": 605
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 0.41995841995842,
|
4272 |
+
"grad_norm": 1.2715479135513306,
|
4273 |
+
"learning_rate": 3.424766225756537e-05,
|
4274 |
+
"loss": 0.0305,
|
4275 |
+
"step": 606
|
4276 |
+
},
|
4277 |
+
{
|
4278 |
+
"epoch": 0.42065142065142064,
|
4279 |
+
"grad_norm": 1.7316052913665771,
|
4280 |
+
"learning_rate": 3.4097155425921254e-05,
|
4281 |
+
"loss": 0.0476,
|
4282 |
+
"step": 607
|
4283 |
+
},
|
4284 |
+
{
|
4285 |
+
"epoch": 0.42134442134442135,
|
4286 |
+
"grad_norm": 2.5955541133880615,
|
4287 |
+
"learning_rate": 3.394680873574546e-05,
|
4288 |
+
"loss": 0.1438,
|
4289 |
+
"step": 608
|
4290 |
+
},
|
4291 |
+
{
|
4292 |
+
"epoch": 0.42203742203742206,
|
4293 |
+
"grad_norm": 1.78447687625885,
|
4294 |
+
"learning_rate": 3.3796623701027476e-05,
|
4295 |
+
"loss": 0.0672,
|
4296 |
+
"step": 609
|
4297 |
+
},
|
4298 |
+
{
|
4299 |
+
"epoch": 0.4227304227304227,
|
4300 |
+
"grad_norm": 1.5245907306671143,
|
4301 |
+
"learning_rate": 3.364660183412892e-05,
|
4302 |
+
"loss": 0.0834,
|
4303 |
+
"step": 610
|
4304 |
+
},
|
4305 |
+
{
|
4306 |
+
"epoch": 0.42342342342342343,
|
4307 |
+
"grad_norm": 2.709289312362671,
|
4308 |
+
"learning_rate": 3.349674464576834e-05,
|
4309 |
+
"loss": 0.0825,
|
4310 |
+
"step": 611
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 0.42411642411642414,
|
4314 |
+
"grad_norm": 2.3717944622039795,
|
4315 |
+
"learning_rate": 3.334705364500596e-05,
|
4316 |
+
"loss": 0.0771,
|
4317 |
+
"step": 612
|
4318 |
+
},
|
4319 |
+
{
|
4320 |
+
"epoch": 0.4248094248094248,
|
4321 |
+
"grad_norm": 2.123501777648926,
|
4322 |
+
"learning_rate": 3.3197530339228487e-05,
|
4323 |
+
"loss": 0.0691,
|
4324 |
+
"step": 613
|
4325 |
+
},
|
4326 |
+
{
|
4327 |
+
"epoch": 0.4255024255024255,
|
4328 |
+
"grad_norm": 2.309788942337036,
|
4329 |
+
"learning_rate": 3.304817623413397e-05,
|
4330 |
+
"loss": 0.1484,
|
4331 |
+
"step": 614
|
4332 |
+
},
|
4333 |
+
{
|
4334 |
+
"epoch": 0.4261954261954262,
|
4335 |
+
"grad_norm": 1.5497885942459106,
|
4336 |
+
"learning_rate": 3.289899283371657e-05,
|
4337 |
+
"loss": 0.0443,
|
4338 |
+
"step": 615
|
4339 |
+
},
|
4340 |
+
{
|
4341 |
+
"epoch": 0.4268884268884269,
|
4342 |
+
"grad_norm": 1.267046570777893,
|
4343 |
+
"learning_rate": 3.274998164025148e-05,
|
4344 |
+
"loss": 0.0596,
|
4345 |
+
"step": 616
|
4346 |
+
},
|
4347 |
+
{
|
4348 |
+
"epoch": 0.4275814275814276,
|
4349 |
+
"grad_norm": 3.855762004852295,
|
4350 |
+
"learning_rate": 3.260114415427975e-05,
|
4351 |
+
"loss": 0.2714,
|
4352 |
+
"step": 617
|
4353 |
+
},
|
4354 |
+
{
|
4355 |
+
"epoch": 0.4282744282744283,
|
4356 |
+
"grad_norm": 1.7666049003601074,
|
4357 |
+
"learning_rate": 3.2452481874593234e-05,
|
4358 |
+
"loss": 0.0751,
|
4359 |
+
"step": 618
|
4360 |
+
},
|
4361 |
+
{
|
4362 |
+
"epoch": 0.42896742896742895,
|
4363 |
+
"grad_norm": 1.7175681591033936,
|
4364 |
+
"learning_rate": 3.230399629821942e-05,
|
4365 |
+
"loss": 0.0902,
|
4366 |
+
"step": 619
|
4367 |
+
},
|
4368 |
+
{
|
4369 |
+
"epoch": 0.42966042966042967,
|
4370 |
+
"grad_norm": 2.664297103881836,
|
4371 |
+
"learning_rate": 3.215568892040641e-05,
|
4372 |
+
"loss": 0.1185,
|
4373 |
+
"step": 620
|
4374 |
+
},
|
4375 |
+
{
|
4376 |
+
"epoch": 0.4303534303534304,
|
4377 |
+
"grad_norm": 1.8793284893035889,
|
4378 |
+
"learning_rate": 3.200756123460788e-05,
|
4379 |
+
"loss": 0.0478,
|
4380 |
+
"step": 621
|
4381 |
+
},
|
4382 |
+
{
|
4383 |
+
"epoch": 0.43104643104643103,
|
4384 |
+
"grad_norm": 2.2081196308135986,
|
4385 |
+
"learning_rate": 3.1859614732467954e-05,
|
4386 |
+
"loss": 0.0874,
|
4387 |
+
"step": 622
|
4388 |
+
},
|
4389 |
+
{
|
4390 |
+
"epoch": 0.43173943173943174,
|
4391 |
+
"grad_norm": 1.9158347845077515,
|
4392 |
+
"learning_rate": 3.171185090380628e-05,
|
4393 |
+
"loss": 0.0462,
|
4394 |
+
"step": 623
|
4395 |
+
},
|
4396 |
+
{
|
4397 |
+
"epoch": 0.43243243243243246,
|
4398 |
+
"grad_norm": 1.7025054693222046,
|
4399 |
+
"learning_rate": 3.156427123660297e-05,
|
4400 |
+
"loss": 0.0625,
|
4401 |
+
"step": 624
|
4402 |
+
},
|
4403 |
+
{
|
4404 |
+
"epoch": 0.4331254331254331,
|
4405 |
+
"grad_norm": 3.4043662548065186,
|
4406 |
+
"learning_rate": 3.141687721698363e-05,
|
4407 |
+
"loss": 0.081,
|
4408 |
+
"step": 625
|
4409 |
+
},
|
4410 |
+
{
|
4411 |
+
"epoch": 0.4338184338184338,
|
4412 |
+
"grad_norm": 1.7663445472717285,
|
4413 |
+
"learning_rate": 3.12696703292044e-05,
|
4414 |
+
"loss": 0.06,
|
4415 |
+
"step": 626
|
4416 |
+
},
|
4417 |
+
{
|
4418 |
+
"epoch": 0.43451143451143454,
|
4419 |
+
"grad_norm": 1.218408226966858,
|
4420 |
+
"learning_rate": 3.1122652055637015e-05,
|
4421 |
+
"loss": 0.0423,
|
4422 |
+
"step": 627
|
4423 |
+
},
|
4424 |
+
{
|
4425 |
+
"epoch": 0.4352044352044352,
|
4426 |
+
"grad_norm": 1.9166085720062256,
|
4427 |
+
"learning_rate": 3.097582387675385e-05,
|
4428 |
+
"loss": 0.0452,
|
4429 |
+
"step": 628
|
4430 |
+
},
|
4431 |
+
{
|
4432 |
+
"epoch": 0.4358974358974359,
|
4433 |
+
"grad_norm": 1.9915510416030884,
|
4434 |
+
"learning_rate": 3.082918727111304e-05,
|
4435 |
+
"loss": 0.0329,
|
4436 |
+
"step": 629
|
4437 |
+
},
|
4438 |
+
{
|
4439 |
+
"epoch": 0.4365904365904366,
|
4440 |
+
"grad_norm": 1.8710565567016602,
|
4441 |
+
"learning_rate": 3.0682743715343564e-05,
|
4442 |
+
"loss": 0.0452,
|
4443 |
+
"step": 630
|
4444 |
+
},
|
4445 |
+
{
|
4446 |
+
"epoch": 0.43728343728343727,
|
4447 |
+
"grad_norm": 1.6264441013336182,
|
4448 |
+
"learning_rate": 3.053649468413043e-05,
|
4449 |
+
"loss": 0.0617,
|
4450 |
+
"step": 631
|
4451 |
+
},
|
4452 |
+
{
|
4453 |
+
"epoch": 0.437976437976438,
|
4454 |
+
"grad_norm": 3.281435489654541,
|
4455 |
+
"learning_rate": 3.0390441650199724e-05,
|
4456 |
+
"loss": 0.0905,
|
4457 |
+
"step": 632
|
4458 |
+
},
|
4459 |
+
{
|
4460 |
+
"epoch": 0.4386694386694387,
|
4461 |
+
"grad_norm": 3.2243845462799072,
|
4462 |
+
"learning_rate": 3.0244586084303905e-05,
|
4463 |
+
"loss": 0.0823,
|
4464 |
+
"step": 633
|
4465 |
+
},
|
4466 |
+
{
|
4467 |
+
"epoch": 0.43936243936243935,
|
4468 |
+
"grad_norm": 2.279958724975586,
|
4469 |
+
"learning_rate": 3.0098929455206904e-05,
|
4470 |
+
"loss": 0.0779,
|
4471 |
+
"step": 634
|
4472 |
+
},
|
4473 |
+
{
|
4474 |
+
"epoch": 0.44005544005544006,
|
4475 |
+
"grad_norm": 1.3307725191116333,
|
4476 |
+
"learning_rate": 2.9953473229669328e-05,
|
4477 |
+
"loss": 0.0512,
|
4478 |
+
"step": 635
|
4479 |
+
},
|
4480 |
+
{
|
4481 |
+
"epoch": 0.4407484407484408,
|
4482 |
+
"grad_norm": 1.505265474319458,
|
4483 |
+
"learning_rate": 2.9808218872433767e-05,
|
4484 |
+
"loss": 0.0411,
|
4485 |
+
"step": 636
|
4486 |
+
},
|
4487 |
+
{
|
4488 |
+
"epoch": 0.44144144144144143,
|
4489 |
+
"grad_norm": 1.387364149093628,
|
4490 |
+
"learning_rate": 2.9663167846209998e-05,
|
4491 |
+
"loss": 0.0627,
|
4492 |
+
"step": 637
|
4493 |
+
},
|
4494 |
+
{
|
4495 |
+
"epoch": 0.44213444213444214,
|
4496 |
+
"grad_norm": 2.2626707553863525,
|
4497 |
+
"learning_rate": 2.9518321611660237e-05,
|
4498 |
+
"loss": 0.0887,
|
4499 |
+
"step": 638
|
4500 |
+
},
|
4501 |
+
{
|
4502 |
+
"epoch": 0.44282744282744285,
|
4503 |
+
"grad_norm": 3.1403772830963135,
|
4504 |
+
"learning_rate": 2.9373681627384447e-05,
|
4505 |
+
"loss": 0.0695,
|
4506 |
+
"step": 639
|
4507 |
+
},
|
4508 |
+
{
|
4509 |
+
"epoch": 0.4435204435204435,
|
4510 |
+
"grad_norm": 3.117483615875244,
|
4511 |
+
"learning_rate": 2.9229249349905684e-05,
|
4512 |
+
"loss": 0.0525,
|
4513 |
+
"step": 640
|
4514 |
+
},
|
4515 |
+
{
|
4516 |
+
"epoch": 0.4442134442134442,
|
4517 |
+
"grad_norm": 1.9548671245574951,
|
4518 |
+
"learning_rate": 2.9085026233655365e-05,
|
4519 |
+
"loss": 0.0654,
|
4520 |
+
"step": 641
|
4521 |
+
},
|
4522 |
+
{
|
4523 |
+
"epoch": 0.44490644490644493,
|
4524 |
+
"grad_norm": 2.2557075023651123,
|
4525 |
+
"learning_rate": 2.894101373095867e-05,
|
4526 |
+
"loss": 0.0611,
|
4527 |
+
"step": 642
|
4528 |
+
},
|
4529 |
+
{
|
4530 |
+
"epoch": 0.4455994455994456,
|
4531 |
+
"grad_norm": 1.508870244026184,
|
4532 |
+
"learning_rate": 2.8797213292019926e-05,
|
4533 |
+
"loss": 0.0566,
|
4534 |
+
"step": 643
|
4535 |
+
},
|
4536 |
+
{
|
4537 |
+
"epoch": 0.4462924462924463,
|
4538 |
+
"grad_norm": 1.16555655002594,
|
4539 |
+
"learning_rate": 2.8653626364907917e-05,
|
4540 |
+
"loss": 0.0206,
|
4541 |
+
"step": 644
|
4542 |
+
},
|
4543 |
+
{
|
4544 |
+
"epoch": 0.446985446985447,
|
4545 |
+
"grad_norm": 0.8943382501602173,
|
4546 |
+
"learning_rate": 2.851025439554142e-05,
|
4547 |
+
"loss": 0.0146,
|
4548 |
+
"step": 645
|
4549 |
+
},
|
4550 |
+
{
|
4551 |
+
"epoch": 0.44767844767844767,
|
4552 |
+
"grad_norm": 1.6789727210998535,
|
4553 |
+
"learning_rate": 2.8367098827674578e-05,
|
4554 |
+
"loss": 0.023,
|
4555 |
+
"step": 646
|
4556 |
+
},
|
4557 |
+
{
|
4558 |
+
"epoch": 0.4483714483714484,
|
4559 |
+
"grad_norm": 2.898833751678467,
|
4560 |
+
"learning_rate": 2.8224161102882397e-05,
|
4561 |
+
"loss": 0.0378,
|
4562 |
+
"step": 647
|
4563 |
+
},
|
4564 |
+
{
|
4565 |
+
"epoch": 0.4490644490644491,
|
4566 |
+
"grad_norm": 1.326059341430664,
|
4567 |
+
"learning_rate": 2.8081442660546125e-05,
|
4568 |
+
"loss": 0.0283,
|
4569 |
+
"step": 648
|
4570 |
+
},
|
4571 |
+
{
|
4572 |
+
"epoch": 0.44975744975744975,
|
4573 |
+
"grad_norm": 0.012674244120717049,
|
4574 |
+
"learning_rate": 2.7938944937838923e-05,
|
4575 |
+
"loss": 0.0001,
|
4576 |
+
"step": 649
|
4577 |
+
},
|
4578 |
+
{
|
4579 |
+
"epoch": 0.45045045045045046,
|
4580 |
+
"grad_norm": 0.04887093976140022,
|
4581 |
+
"learning_rate": 2.7796669369711294e-05,
|
4582 |
+
"loss": 0.0002,
|
4583 |
+
"step": 650
|
4584 |
+
},
|
4585 |
+
{
|
4586 |
+
"epoch": 0.45114345114345117,
|
4587 |
+
"grad_norm": 1.1527273654937744,
|
4588 |
+
"learning_rate": 2.7654617388876615e-05,
|
4589 |
+
"loss": 0.045,
|
4590 |
+
"step": 651
|
4591 |
+
},
|
4592 |
+
{
|
4593 |
+
"epoch": 0.4518364518364518,
|
4594 |
+
"grad_norm": 2.142963409423828,
|
4595 |
+
"learning_rate": 2.7512790425796718e-05,
|
4596 |
+
"loss": 0.1299,
|
4597 |
+
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.45252945252945254,
|
4601 |
+
"grad_norm": 3.2838921546936035,
|
4602 |
+
"learning_rate": 2.7371189908667604e-05,
|
4603 |
+
"loss": 0.1249,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.45322245322245325,
|
4608 |
+
"grad_norm": 1.5524722337722778,
|
4609 |
+
"learning_rate": 2.7229817263404866e-05,
|
4610 |
+
"loss": 0.0692,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.4539154539154539,
|
4615 |
+
"grad_norm": 2.0327537059783936,
|
4616 |
+
"learning_rate": 2.708867391362948e-05,
|
4617 |
+
"loss": 0.0729,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.4546084546084546,
|
4622 |
+
"grad_norm": 1.5686278343200684,
|
4623 |
+
"learning_rate": 2.694776128065345e-05,
|
4624 |
+
"loss": 0.0535,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.4553014553014553,
|
4629 |
+
"grad_norm": 1.42728590965271,
|
4630 |
+
"learning_rate": 2.6807080783465376e-05,
|
4631 |
+
"loss": 0.0835,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.455994455994456,
|
4636 |
+
"grad_norm": 1.6944345235824585,
|
4637 |
+
"learning_rate": 2.6666633838716314e-05,
|
4638 |
+
"loss": 0.0471,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.4566874566874567,
|
4643 |
+
"grad_norm": 1.3432730436325073,
|
4644 |
+
"learning_rate": 2.6526421860705473e-05,
|
4645 |
+
"loss": 0.07,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.4573804573804574,
|
4650 |
+
"grad_norm": 2.6595051288604736,
|
4651 |
+
"learning_rate": 2.638644626136587e-05,
|
4652 |
+
"loss": 0.1548,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.45807345807345806,
|
4657 |
+
"grad_norm": 1.3710395097732544,
|
4658 |
+
"learning_rate": 2.6246708450250256e-05,
|
4659 |
+
"loss": 0.0638,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.4587664587664588,
|
4664 |
+
"grad_norm": 1.750665307044983,
|
4665 |
+
"learning_rate": 2.6107209834516854e-05,
|
4666 |
+
"loss": 0.0547,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.4594594594594595,
|
4671 |
+
"grad_norm": 1.6597323417663574,
|
4672 |
+
"learning_rate": 2.596795181891514e-05,
|
4673 |
+
"loss": 0.0719,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.46015246015246014,
|
4678 |
+
"grad_norm": 3.2295982837677,
|
4679 |
+
"learning_rate": 2.5828935805771802e-05,
|
4680 |
+
"loss": 0.0542,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.46084546084546085,
|
4685 |
+
"grad_norm": 2.0197107791900635,
|
4686 |
+
"learning_rate": 2.5690163194976575e-05,
|
4687 |
+
"loss": 0.0688,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.46153846153846156,
|
4692 |
+
"grad_norm": 1.264258861541748,
|
4693 |
+
"learning_rate": 2.5551635383968065e-05,
|
4694 |
+
"loss": 0.0459,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.4622314622314622,
|
4699 |
+
"grad_norm": 1.3104031085968018,
|
4700 |
+
"learning_rate": 2.5413353767719805e-05,
|
4701 |
+
"loss": 0.0596,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.46292446292446293,
|
4706 |
+
"grad_norm": 2.645916223526001,
|
4707 |
+
"learning_rate": 2.5275319738726165e-05,
|
4708 |
+
"loss": 0.1307,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.46361746361746364,
|
4713 |
+
"grad_norm": 1.6166561841964722,
|
4714 |
+
"learning_rate": 2.513753468698826e-05,
|
4715 |
+
"loss": 0.0582,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.4643104643104643,
|
4720 |
+
"grad_norm": 1.8327372074127197,
|
4721 |
+
"learning_rate": 2.500000000000001e-05,
|
4722 |
+
"loss": 0.0721,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.465003465003465,
|
4727 |
+
"grad_norm": 1.8945083618164062,
|
4728 |
+
"learning_rate": 2.486271706273421e-05,
|
4729 |
+
"loss": 0.1175,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.4656964656964657,
|
4734 |
+
"grad_norm": 1.4678902626037598,
|
4735 |
+
"learning_rate": 2.4725687257628534e-05,
|
4736 |
+
"loss": 0.0621,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.4663894663894664,
|
4741 |
+
"grad_norm": 1.4501336812973022,
|
4742 |
+
"learning_rate": 2.4588911964571553e-05,
|
4743 |
+
"loss": 0.0723,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.4670824670824671,
|
4748 |
+
"grad_norm": 1.9211817979812622,
|
4749 |
+
"learning_rate": 2.4452392560888976e-05,
|
4750 |
+
"loss": 0.0762,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.4677754677754678,
|
4755 |
+
"grad_norm": 1.600954532623291,
|
4756 |
+
"learning_rate": 2.4316130421329697e-05,
|
4757 |
+
"loss": 0.0642,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.46846846846846846,
|
4762 |
+
"grad_norm": 2.006173610687256,
|
4763 |
+
"learning_rate": 2.418012691805191e-05,
|
4764 |
+
"loss": 0.1227,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.46916146916146917,
|
4769 |
+
"grad_norm": 1.272980809211731,
|
4770 |
+
"learning_rate": 2.4044383420609406e-05,
|
4771 |
+
"loss": 0.0575,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.4698544698544699,
|
4776 |
+
"grad_norm": 1.5940930843353271,
|
4777 |
+
"learning_rate": 2.3908901295937713e-05,
|
4778 |
+
"loss": 0.065,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.47054747054747054,
|
4783 |
+
"grad_norm": 3.0212154388427734,
|
4784 |
+
"learning_rate": 2.3773681908340284e-05,
|
4785 |
+
"loss": 0.1318,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.47124047124047125,
|
4790 |
+
"grad_norm": 1.466066837310791,
|
4791 |
+
"learning_rate": 2.363872661947488e-05,
|
4792 |
+
"loss": 0.0401,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.47193347193347196,
|
4797 |
+
"grad_norm": 1.1249910593032837,
|
4798 |
+
"learning_rate": 2.350403678833976e-05,
|
4799 |
+
"loss": 0.0356,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.4726264726264726,
|
4804 |
+
"grad_norm": 1.8696436882019043,
|
4805 |
+
"learning_rate": 2.336961377126001e-05,
|
4806 |
+
"loss": 0.0614,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.4733194733194733,
|
4811 |
+
"grad_norm": 1.967720627784729,
|
4812 |
+
"learning_rate": 2.3235458921873925e-05,
|
4813 |
+
"loss": 0.0413,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.47401247401247404,
|
4818 |
+
"grad_norm": 1.4729372262954712,
|
4819 |
+
"learning_rate": 2.310157359111938e-05,
|
4820 |
+
"loss": 0.061,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.4747054747054747,
|
4825 |
+
"grad_norm": 1.9776396751403809,
|
4826 |
+
"learning_rate": 2.296795912722014e-05,
|
4827 |
+
"loss": 0.1042,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.4753984753984754,
|
4832 |
+
"grad_norm": 1.8598027229309082,
|
4833 |
+
"learning_rate": 2.283461687567236e-05,
|
4834 |
+
"loss": 0.0559,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.4760914760914761,
|
4839 |
+
"grad_norm": 2.0203568935394287,
|
4840 |
+
"learning_rate": 2.2701548179231048e-05,
|
4841 |
+
"loss": 0.0916,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.4767844767844768,
|
4846 |
+
"grad_norm": 1.7815768718719482,
|
4847 |
+
"learning_rate": 2.2568754377896516e-05,
|
4848 |
+
"loss": 0.0452,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.4774774774774775,
|
4853 |
+
"grad_norm": 1.6619998216629028,
|
4854 |
+
"learning_rate": 2.2436236808900844e-05,
|
4855 |
+
"loss": 0.0441,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.4781704781704782,
|
4860 |
+
"grad_norm": 2.220858097076416,
|
4861 |
+
"learning_rate": 2.2303996806694488e-05,
|
4862 |
+
"loss": 0.1036,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.47886347886347885,
|
4867 |
+
"grad_norm": 0.9709250330924988,
|
4868 |
+
"learning_rate": 2.2172035702932825e-05,
|
4869 |
+
"loss": 0.0258,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.47955647955647956,
|
4874 |
+
"grad_norm": 1.4434762001037598,
|
4875 |
+
"learning_rate": 2.2040354826462668e-05,
|
4876 |
+
"loss": 0.0316,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.4802494802494803,
|
4881 |
+
"grad_norm": 1.7352850437164307,
|
4882 |
+
"learning_rate": 2.1908955503308993e-05,
|
4883 |
+
"loss": 0.0494,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.48094248094248093,
|
4888 |
+
"grad_norm": 2.2967495918273926,
|
4889 |
+
"learning_rate": 2.1777839056661554e-05,
|
4890 |
+
"loss": 0.0652,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.48163548163548164,
|
4895 |
+
"grad_norm": 2.0756208896636963,
|
4896 |
+
"learning_rate": 2.164700680686147e-05,
|
4897 |
+
"loss": 0.0404,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.48232848232848236,
|
4902 |
+
"grad_norm": 5.310906887054443,
|
4903 |
+
"learning_rate": 2.1516460071388062e-05,
|
4904 |
+
"loss": 0.0521,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.483021483021483,
|
4909 |
+
"grad_norm": 3.2012720108032227,
|
4910 |
+
"learning_rate": 2.1386200164845526e-05,
|
4911 |
+
"loss": 0.1225,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.4837144837144837,
|
4916 |
+
"grad_norm": 8.336894989013672,
|
4917 |
+
"learning_rate": 2.125622839894964e-05,
|
4918 |
+
"loss": 0.0514,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.48440748440748443,
|
4923 |
+
"grad_norm": 0.009585398249328136,
|
4924 |
+
"learning_rate": 2.1126546082514664e-05,
|
4925 |
+
"loss": 0.0001,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.4851004851004851,
|
4930 |
+
"grad_norm": 6.890432834625244,
|
4931 |
+
"learning_rate": 2.09971545214401e-05,
|
4932 |
+
"loss": 0.0131,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.4857934857934858,
|
4937 |
+
"grad_norm": 2.0048975944519043,
|
4938 |
+
"learning_rate": 2.086805501869749e-05,
|
4939 |
+
"loss": 0.1861,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.4864864864864865,
|
4944 |
+
"grad_norm": 1.4275426864624023,
|
4945 |
+
"learning_rate": 2.073924887431744e-05,
|
4946 |
+
"loss": 0.0573,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.48717948717948717,
|
4951 |
+
"grad_norm": 2.745793104171753,
|
4952 |
+
"learning_rate": 2.061073738537635e-05,
|
4953 |
+
"loss": 0.0999,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.4878724878724879,
|
4958 |
+
"grad_norm": 2.188532590866089,
|
4959 |
+
"learning_rate": 2.048252184598352e-05,
|
4960 |
+
"loss": 0.0662,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.4885654885654886,
|
4965 |
+
"grad_norm": 0.9374767541885376,
|
4966 |
+
"learning_rate": 2.0354603547267985e-05,
|
4967 |
+
"loss": 0.022,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.48925848925848925,
|
4972 |
+
"grad_norm": 1.1383275985717773,
|
4973 |
+
"learning_rate": 2.0226983777365604e-05,
|
4974 |
+
"loss": 0.0426,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.48995148995148996,
|
4979 |
+
"grad_norm": 1.7358490228652954,
|
4980 |
+
"learning_rate": 2.0099663821406056e-05,
|
4981 |
+
"loss": 0.0774,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.49064449064449067,
|
4986 |
+
"grad_norm": 3.148839235305786,
|
4987 |
+
"learning_rate": 1.9972644961499854e-05,
|
4988 |
+
"loss": 0.0728,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.49133749133749133,
|
4993 |
+
"grad_norm": 1.354243278503418,
|
4994 |
+
"learning_rate": 1.9845928476725524e-05,
|
4995 |
+
"loss": 0.058,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.49203049203049204,
|
5000 |
+
"grad_norm": 1.5222508907318115,
|
5001 |
+
"learning_rate": 1.9719515643116674e-05,
|
5002 |
+
"loss": 0.052,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.49272349272349275,
|
5007 |
+
"grad_norm": 1.3979942798614502,
|
5008 |
+
"learning_rate": 1.959340773364911e-05,
|
5009 |
+
"loss": 0.0301,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.4934164934164934,
|
5014 |
+
"grad_norm": 2.327789068222046,
|
5015 |
+
"learning_rate": 1.946760601822809e-05,
|
5016 |
+
"loss": 0.0385,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.4941094941094941,
|
5021 |
+
"grad_norm": 2.022264242172241,
|
5022 |
+
"learning_rate": 1.9342111763675512e-05,
|
5023 |
+
"loss": 0.046,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.49480249480249483,
|
5028 |
+
"grad_norm": 1.3128710985183716,
|
5029 |
+
"learning_rate": 1.9216926233717085e-05,
|
5030 |
+
"loss": 0.046,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.4954954954954955,
|
5035 |
+
"grad_norm": 2.2289204597473145,
|
5036 |
+
"learning_rate": 1.9092050688969738e-05,
|
5037 |
+
"loss": 0.0395,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.4961884961884962,
|
5042 |
+
"grad_norm": 1.4785228967666626,
|
5043 |
+
"learning_rate": 1.8967486386928817e-05,
|
5044 |
+
"loss": 0.0482,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.4968814968814969,
|
5049 |
+
"grad_norm": 1.7745870351791382,
|
5050 |
+
"learning_rate": 1.8843234581955442e-05,
|
5051 |
+
"loss": 0.0501,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.49757449757449757,
|
5056 |
+
"grad_norm": 4.233173370361328,
|
5057 |
+
"learning_rate": 1.8719296525263922e-05,
|
5058 |
+
"loss": 0.2729,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.4982674982674983,
|
5063 |
+
"grad_norm": 1.4577674865722656,
|
5064 |
+
"learning_rate": 1.859567346490913e-05,
|
5065 |
+
"loss": 0.0879,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.498960498960499,
|
5070 |
+
"grad_norm": 2.581014394760132,
|
5071 |
+
"learning_rate": 1.847236664577389e-05,
|
5072 |
+
"loss": 0.1143,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.49965349965349964,
|
5077 |
+
"grad_norm": 1.708632469177246,
|
5078 |
+
"learning_rate": 1.8349377309556486e-05,
|
5079 |
+
"loss": 0.0614,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.5003465003465003,
|
5084 |
+
"grad_norm": 4.064712047576904,
|
5085 |
+
"learning_rate": 1.8226706694758195e-05,
|
5086 |
+
"loss": 0.0652,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.501039501039501,
|
5091 |
+
"grad_norm": 2.1657605171203613,
|
5092 |
+
"learning_rate": 1.810435603667075e-05,
|
5093 |
+
"loss": 0.062,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.5017325017325017,
|
5098 |
+
"grad_norm": 2.218892812728882,
|
5099 |
+
"learning_rate": 1.7982326567363888e-05,
|
5100 |
+
"loss": 0.0313,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.5024255024255024,
|
5105 |
+
"grad_norm": 2.087905168533325,
|
5106 |
+
"learning_rate": 1.7860619515673033e-05,
|
5107 |
+
"loss": 0.0699,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.5031185031185031,
|
5112 |
+
"grad_norm": 2.7413132190704346,
|
5113 |
+
"learning_rate": 1.773923610718686e-05,
|
5114 |
+
"loss": 0.0602,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.5038115038115039,
|
5119 |
+
"grad_norm": 2.1985816955566406,
|
5120 |
+
"learning_rate": 1.7618177564234905e-05,
|
5121 |
+
"loss": 0.1066,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.5045045045045045,
|
5126 |
+
"grad_norm": 1.0309902429580688,
|
5127 |
+
"learning_rate": 1.7497445105875377e-05,
|
5128 |
+
"loss": 0.0388,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.5051975051975052,
|
5133 |
+
"grad_norm": 1.1991498470306396,
|
5134 |
+
"learning_rate": 1.73770399478828e-05,
|
5135 |
+
"loss": 0.025,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.5058905058905059,
|
5140 |
+
"grad_norm": 2.0643422603607178,
|
5141 |
+
"learning_rate": 1.725696330273575e-05,
|
5142 |
+
"loss": 0.0574,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.5065835065835066,
|
5147 |
+
"grad_norm": 1.4626661539077759,
|
5148 |
+
"learning_rate": 1.7137216379604727e-05,
|
5149 |
+
"loss": 0.1035,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.5072765072765073,
|
5154 |
+
"grad_norm": 1.5174840688705444,
|
5155 |
+
"learning_rate": 1.7017800384339928e-05,
|
5156 |
+
"loss": 0.0722,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.507969507969508,
|
5161 |
+
"grad_norm": 2.180429458618164,
|
5162 |
+
"learning_rate": 1.6898716519459074e-05,
|
5163 |
+
"loss": 0.1261,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.5086625086625086,
|
5168 |
+
"grad_norm": 1.6073616743087769,
|
5169 |
+
"learning_rate": 1.6779965984135377e-05,
|
5170 |
+
"loss": 0.0837,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.5093555093555093,
|
5175 |
+
"grad_norm": 3.2812795639038086,
|
5176 |
+
"learning_rate": 1.6661549974185424e-05,
|
5177 |
+
"loss": 0.146,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.51004851004851,
|
5182 |
+
"grad_norm": 2.0749173164367676,
|
5183 |
+
"learning_rate": 1.6543469682057106e-05,
|
5184 |
+
"loss": 0.082,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.5107415107415108,
|
5189 |
+
"grad_norm": 1.5799446105957031,
|
5190 |
+
"learning_rate": 1.6425726296817633e-05,
|
5191 |
+
"loss": 0.0335,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.5114345114345115,
|
5196 |
+
"grad_norm": 1.534477949142456,
|
5197 |
+
"learning_rate": 1.6308321004141607e-05,
|
5198 |
+
"loss": 0.0605,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.5121275121275122,
|
5203 |
+
"grad_norm": 1.5507577657699585,
|
5204 |
+
"learning_rate": 1.619125498629904e-05,
|
5205 |
+
"loss": 0.0465,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.5128205128205128,
|
5210 |
+
"grad_norm": 4.244449615478516,
|
5211 |
+
"learning_rate": 1.60745294221434e-05,
|
5212 |
+
"loss": 0.0472,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.5135135135135135,
|
5217 |
+
"grad_norm": 0.9149932265281677,
|
5218 |
+
"learning_rate": 1.595814548709983e-05,
|
5219 |
+
"loss": 0.027,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.5142065142065142,
|
5224 |
+
"grad_norm": 0.7855058908462524,
|
5225 |
+
"learning_rate": 1.5842104353153287e-05,
|
5226 |
+
"loss": 0.0213,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.5148995148995149,
|
5231 |
+
"grad_norm": 1.9645469188690186,
|
5232 |
+
"learning_rate": 1.5726407188836673e-05,
|
5233 |
+
"loss": 0.026,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.5155925155925156,
|
5238 |
+
"grad_norm": 1.3487372398376465,
|
5239 |
+
"learning_rate": 1.5611055159219152e-05,
|
5240 |
+
"loss": 0.0399,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.5162855162855163,
|
5245 |
+
"grad_norm": 2.7862071990966797,
|
5246 |
+
"learning_rate": 1.549604942589441e-05,
|
5247 |
+
"loss": 0.0253,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.5169785169785169,
|
5252 |
+
"grad_norm": 1.8288354873657227,
|
5253 |
+
"learning_rate": 1.5381391146968866e-05,
|
5254 |
+
"loss": 0.0505,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.5176715176715176,
|
5259 |
+
"grad_norm": 0.9688133001327515,
|
5260 |
+
"learning_rate": 1.526708147705013e-05,
|
5261 |
+
"loss": 0.023,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.5183645183645184,
|
5266 |
+
"grad_norm": 1.158333420753479,
|
5267 |
+
"learning_rate": 1.5153121567235335e-05,
|
5268 |
+
"loss": 0.0239,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.5190575190575191,
|
5273 |
+
"grad_norm": 0.25449123978614807,
|
5274 |
+
"learning_rate": 1.5039512565099467e-05,
|
5275 |
+
"loss": 0.0013,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.5197505197505198,
|
5280 |
+
"grad_norm": 0.23921255767345428,
|
5281 |
+
"learning_rate": 1.4926255614683932e-05,
|
5282 |
+
"loss": 0.0002,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.5204435204435205,
|
5287 |
+
"grad_norm": 1.5712229013442993,
|
5288 |
+
"learning_rate": 1.481335185648498e-05,
|
5289 |
+
"loss": 0.268,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.5211365211365211,
|
5294 |
+
"grad_norm": 1.4475122690200806,
|
5295 |
+
"learning_rate": 1.4700802427442179e-05,
|
5296 |
+
"loss": 0.0789,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.5218295218295218,
|
5301 |
+
"grad_norm": 2.140687942504883,
|
5302 |
+
"learning_rate": 1.458860846092705e-05,
|
5303 |
+
"loss": 0.1106,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.5225225225225225,
|
5308 |
+
"grad_norm": 1.6918996572494507,
|
5309 |
+
"learning_rate": 1.4476771086731567e-05,
|
5310 |
+
"loss": 0.0567,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.5232155232155232,
|
5315 |
+
"grad_norm": 1.683551549911499,
|
5316 |
+
"learning_rate": 1.4365291431056871e-05,
|
5317 |
+
"loss": 0.1074,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.5239085239085239,
|
5322 |
+
"grad_norm": 1.1498463153839111,
|
5323 |
+
"learning_rate": 1.4254170616501827e-05,
|
5324 |
+
"loss": 0.0149,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.5246015246015246,
|
5329 |
+
"grad_norm": 1.2035645246505737,
|
5330 |
+
"learning_rate": 1.414340976205183e-05,
|
5331 |
+
"loss": 0.0424,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.5252945252945252,
|
5336 |
+
"grad_norm": 1.7896965742111206,
|
5337 |
+
"learning_rate": 1.4033009983067452e-05,
|
5338 |
+
"loss": 0.1129,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.525987525987526,
|
5343 |
+
"grad_norm": 1.246095895767212,
|
5344 |
+
"learning_rate": 1.3922972391273226e-05,
|
5345 |
+
"loss": 0.0636,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.5266805266805267,
|
5350 |
+
"grad_norm": 2.1839723587036133,
|
5351 |
+
"learning_rate": 1.3813298094746491e-05,
|
5352 |
+
"loss": 0.1153,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.5273735273735274,
|
5357 |
+
"grad_norm": 1.7188761234283447,
|
5358 |
+
"learning_rate": 1.3703988197906209e-05,
|
5359 |
+
"loss": 0.0778,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.5280665280665281,
|
5364 |
+
"grad_norm": 1.6612868309020996,
|
5365 |
+
"learning_rate": 1.3595043801501794e-05,
|
5366 |
+
"loss": 0.0354,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.5287595287595288,
|
5371 |
+
"grad_norm": 1.7637569904327393,
|
5372 |
+
"learning_rate": 1.3486466002602133e-05,
|
5373 |
+
"loss": 0.0885,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.5294525294525294,
|
5378 |
+
"grad_norm": 2.42702317237854,
|
5379 |
+
"learning_rate": 1.3378255894584463e-05,
|
5380 |
+
"loss": 0.1145,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.5301455301455301,
|
5385 |
+
"grad_norm": 1.406791090965271,
|
5386 |
+
"learning_rate": 1.327041456712334e-05,
|
5387 |
+
"loss": 0.0686,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.5308385308385308,
|
5392 |
+
"grad_norm": 2.8606255054473877,
|
5393 |
+
"learning_rate": 1.3162943106179749e-05,
|
5394 |
+
"loss": 0.1681,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.5315315315315315,
|
5399 |
+
"grad_norm": 0.9502009153366089,
|
5400 |
+
"learning_rate": 1.3055842593990131e-05,
|
5401 |
+
"loss": 0.0418,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.5322245322245323,
|
5406 |
+
"grad_norm": 1.3333024978637695,
|
5407 |
+
"learning_rate": 1.2949114109055415e-05,
|
5408 |
+
"loss": 0.0683,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.532917532917533,
|
5413 |
+
"grad_norm": 2.184300661087036,
|
5414 |
+
"learning_rate": 1.2842758726130283e-05,
|
5415 |
+
"loss": 0.0293,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.5336105336105336,
|
5420 |
+
"grad_norm": 1.6312488317489624,
|
5421 |
+
"learning_rate": 1.2736777516212266e-05,
|
5422 |
+
"loss": 0.071,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.5343035343035343,
|
5427 |
+
"grad_norm": 3.242823600769043,
|
5428 |
+
"learning_rate": 1.2631171546530968e-05,
|
5429 |
+
"loss": 0.082,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.534996534996535,
|
5434 |
+
"grad_norm": 2.8160717487335205,
|
5435 |
+
"learning_rate": 1.2525941880537307e-05,
|
5436 |
+
"loss": 0.0488,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.5356895356895357,
|
5441 |
+
"grad_norm": 1.718396544456482,
|
5442 |
+
"learning_rate": 1.2421089577892869e-05,
|
5443 |
+
"loss": 0.0722,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.5363825363825364,
|
5448 |
+
"grad_norm": 2.165678024291992,
|
5449 |
+
"learning_rate": 1.2316615694459189e-05,
|
5450 |
+
"loss": 0.085,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.5370755370755371,
|
5455 |
+
"grad_norm": 4.593423843383789,
|
5456 |
+
"learning_rate": 1.2212521282287092e-05,
|
5457 |
+
"loss": 0.2502,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.5377685377685377,
|
5462 |
+
"grad_norm": 1.3617160320281982,
|
5463 |
+
"learning_rate": 1.2108807389606158e-05,
|
5464 |
+
"loss": 0.0314,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.5384615384615384,
|
5469 |
+
"grad_norm": 1.9156297445297241,
|
5470 |
+
"learning_rate": 1.2005475060814159e-05,
|
5471 |
+
"loss": 0.1195,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.5391545391545391,
|
5476 |
+
"grad_norm": 1.6965415477752686,
|
5477 |
+
"learning_rate": 1.1902525336466464e-05,
|
5478 |
+
"loss": 0.1157,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.5398475398475399,
|
5483 |
+
"grad_norm": 2.171086072921753,
|
5484 |
+
"learning_rate": 1.1799959253265668e-05,
|
5485 |
+
"loss": 0.0866,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.5405405405405406,
|
5490 |
+
"grad_norm": 3.2741446495056152,
|
5491 |
+
"learning_rate": 1.1697777844051105e-05,
|
5492 |
+
"loss": 0.1048,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.5412335412335413,
|
5497 |
+
"grad_norm": 1.4844908714294434,
|
5498 |
+
"learning_rate": 1.1595982137788403e-05,
|
5499 |
+
"loss": 0.0816,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.5419265419265419,
|
5504 |
+
"grad_norm": 1.7694414854049683,
|
5505 |
+
"learning_rate": 1.1494573159559213e-05,
|
5506 |
+
"loss": 0.0412,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.5426195426195426,
|
5511 |
+
"grad_norm": 1.2457849979400635,
|
5512 |
+
"learning_rate": 1.1393551930550828e-05,
|
5513 |
+
"loss": 0.0593,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.5433125433125433,
|
5518 |
+
"grad_norm": 0.8710437417030334,
|
5519 |
+
"learning_rate": 1.1292919468045877e-05,
|
5520 |
+
"loss": 0.036,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.544005544005544,
|
5525 |
+
"grad_norm": 2.328735828399658,
|
5526 |
+
"learning_rate": 1.1192676785412154e-05,
|
5527 |
+
"loss": 0.0848,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.5446985446985447,
|
5532 |
+
"grad_norm": 1.1400730609893799,
|
5533 |
+
"learning_rate": 1.1092824892092373e-05,
|
5534 |
+
"loss": 0.0359,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.5453915453915454,
|
5539 |
+
"grad_norm": 2.1870477199554443,
|
5540 |
+
"learning_rate": 1.099336479359398e-05,
|
5541 |
+
"loss": 0.0828,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.546084546084546,
|
5546 |
+
"grad_norm": 1.8485313653945923,
|
5547 |
+
"learning_rate": 1.0894297491479045e-05,
|
5548 |
+
"loss": 0.0594,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.5467775467775468,
|
5553 |
+
"grad_norm": 2.052849054336548,
|
5554 |
+
"learning_rate": 1.0795623983354215e-05,
|
5555 |
+
"loss": 0.0442,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.5474705474705475,
|
5560 |
+
"grad_norm": 4.594583034515381,
|
5561 |
+
"learning_rate": 1.0697345262860636e-05,
|
5562 |
+
"loss": 0.0964,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.5481635481635482,
|
5567 |
+
"grad_norm": 2.2317960262298584,
|
5568 |
+
"learning_rate": 1.0599462319663905e-05,
|
5569 |
+
"loss": 0.0453,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.5488565488565489,
|
5574 |
+
"grad_norm": 1.04276442527771,
|
5575 |
+
"learning_rate": 1.0501976139444191e-05,
|
5576 |
+
"loss": 0.0307,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.5495495495495496,
|
5581 |
+
"grad_norm": 1.3961700201034546,
|
5582 |
+
"learning_rate": 1.0404887703886251e-05,
|
5583 |
+
"loss": 0.0503,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.5502425502425502,
|
5588 |
+
"grad_norm": 1.7015811204910278,
|
5589 |
+
"learning_rate": 1.0308197990669538e-05,
|
5590 |
+
"loss": 0.0435,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.5509355509355509,
|
5595 |
+
"grad_norm": 1.8332734107971191,
|
5596 |
+
"learning_rate": 1.021190797345839e-05,
|
5597 |
+
"loss": 0.0263,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.5516285516285516,
|
5602 |
+
"grad_norm": 1.551506519317627,
|
5603 |
+
"learning_rate": 1.0116018621892237e-05,
|
5604 |
+
"loss": 0.0402,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.5523215523215523,
|
5609 |
+
"grad_norm": 2.0567147731781006,
|
5610 |
+
"learning_rate": 1.0020530901575754e-05,
|
5611 |
+
"loss": 0.0177,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.553014553014553,
|
5616 |
+
"grad_norm": 1.6934616565704346,
|
5617 |
+
"learning_rate": 9.92544577406923e-06,
|
5618 |
+
"loss": 0.0304,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.5537075537075538,
|
5623 |
+
"grad_norm": 0.8998746275901794,
|
5624 |
+
"learning_rate": 9.830764196878872e-06,
|
5625 |
+
"loss": 0.0012,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.5544005544005544,
|
5630 |
+
"grad_norm": 1.843292474746704,
|
5631 |
+
"learning_rate": 9.73648712344707e-06,
|
5632 |
+
"loss": 0.0003,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.5544005544005544,
|
5637 |
+
"eval_loss": 0.040625762194395065,
|
5638 |
+
"eval_runtime": 69.5659,
|
5639 |
+
"eval_samples_per_second": 34.945,
|
5640 |
+
"eval_steps_per_second": 8.74,
|
5641 |
+
"step": 800
|
5642 |
}
|
5643 |
],
|
5644 |
"logging_steps": 1,
|
|
|
5667 |
"attributes": {}
|
5668 |
}
|
5669 |
},
|
5670 |
+
"total_flos": 4.621122051778806e+17,
|
5671 |
"train_batch_size": 8,
|
5672 |
"trial_name": null,
|
5673 |
"trial_params": null
|