Training in progress, step 30000, checkpoint
Browse files- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state_0.pth +1 -1
- last-checkpoint/rng_state_1.pth +1 -1
- last-checkpoint/rng_state_2.pth +1 -1
- last-checkpoint/rng_state_3.pth +1 -1
- last-checkpoint/rng_state_4.pth +1 -1
- last-checkpoint/rng_state_5.pth +1 -1
- last-checkpoint/rng_state_6.pth +1 -1
- last-checkpoint/rng_state_7.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +3511 -3
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 306619286
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:877ef9b1ef203d1c02f31d1f5d81565b3878eb39121c8153f356847a2fd8eef8
|
| 3 |
size 306619286
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 919972410
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78438408549281a57107eeb546cb6b697cb5ef9532b1f55bd2b74f24bbf771c3
|
| 3 |
size 919972410
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb18ac8d6db3307b1c242f7cb069fc8b8dab957434ddfcafcac997cfd6a43abf
|
| 3 |
size 15984
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4bdab708057b5f34a402d9a2b4443f5f93a8e8ee2ddb66d955f0a15ad394ecc5
|
| 3 |
size 15984
|
last-checkpoint/rng_state_2.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:599882a30c163a5a2a000c4e74b320ecc4a55aa1b079882fd66aa3d2559d19e7
|
| 3 |
size 15984
|
last-checkpoint/rng_state_3.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:567c3b482c209c2778fc017e39a38642c488edda20673ef29f571ef7177ad81e
|
| 3 |
size 15984
|
last-checkpoint/rng_state_4.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f9ffe9a916e778423aaed4ec842923c9ccfdd3d7a4fbad10dc6a3bfc278fb8e
|
| 3 |
size 15984
|
last-checkpoint/rng_state_5.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7ede8a81aa3c780fb9c3cb57537752a782c4aed1dcecb7aafd6ca5a7ea90252
|
| 3 |
size 15984
|
last-checkpoint/rng_state_6.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b1c5c0c0afa907d332467e631e6cee80ba476689aa0caa77689ca273d83b3e4
|
| 3 |
size 15984
|
last-checkpoint/rng_state_7.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73025ac422abb13303ee974109cf39f6f848de7f7013e828d04aa4e2ec0e6757
|
| 3 |
size 15984
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1909da172bf01f799e2ef8934ccdab3f4895ac7509eafdfbd8adccaf29c2176
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 5000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -17547,6 +17547,3514 @@
|
|
| 17547 |
"eval_samples_per_second": 3202.139,
|
| 17548 |
"eval_steps_per_second": 50.035,
|
| 17549 |
"step": 25000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17550 |
}
|
| 17551 |
],
|
| 17552 |
"logging_steps": 10,
|
|
@@ -17566,7 +21074,7 @@
|
|
| 17566 |
"attributes": {}
|
| 17567 |
}
|
| 17568 |
},
|
| 17569 |
-
"total_flos":
|
| 17570 |
"train_batch_size": 8,
|
| 17571 |
"trial_name": null,
|
| 17572 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5567192534394811,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 30000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 17547 |
"eval_samples_per_second": 3202.139,
|
| 17548 |
"eval_steps_per_second": 50.035,
|
| 17549 |
"step": 25000
|
| 17550 |
+
},
|
| 17551 |
+
{
|
| 17552 |
+
"epoch": 0.46411828428404744,
|
| 17553 |
+
"grad_norm": 37.1875,
|
| 17554 |
+
"learning_rate": 9.927481539107923e-06,
|
| 17555 |
+
"loss": 21.032,
|
| 17556 |
+
"step": 25010
|
| 17557 |
+
},
|
| 17558 |
+
{
|
| 17559 |
+
"epoch": 0.46430385736852725,
|
| 17560 |
+
"grad_norm": 34.0,
|
| 17561 |
+
"learning_rate": 9.927452543321879e-06,
|
| 17562 |
+
"loss": 20.8725,
|
| 17563 |
+
"step": 25020
|
| 17564 |
+
},
|
| 17565 |
+
{
|
| 17566 |
+
"epoch": 0.4644894304530071,
|
| 17567 |
+
"grad_norm": 35.34375,
|
| 17568 |
+
"learning_rate": 9.927423547535836e-06,
|
| 17569 |
+
"loss": 20.9236,
|
| 17570 |
+
"step": 25030
|
| 17571 |
+
},
|
| 17572 |
+
{
|
| 17573 |
+
"epoch": 0.4646750035374869,
|
| 17574 |
+
"grad_norm": 35.90625,
|
| 17575 |
+
"learning_rate": 9.927394551749796e-06,
|
| 17576 |
+
"loss": 20.7448,
|
| 17577 |
+
"step": 25040
|
| 17578 |
+
},
|
| 17579 |
+
{
|
| 17580 |
+
"epoch": 0.4648605766219667,
|
| 17581 |
+
"grad_norm": 33.875,
|
| 17582 |
+
"learning_rate": 9.927365555963753e-06,
|
| 17583 |
+
"loss": 21.2387,
|
| 17584 |
+
"step": 25050
|
| 17585 |
+
},
|
| 17586 |
+
{
|
| 17587 |
+
"epoch": 0.4650461497064466,
|
| 17588 |
+
"grad_norm": 34.8125,
|
| 17589 |
+
"learning_rate": 9.92733656017771e-06,
|
| 17590 |
+
"loss": 20.801,
|
| 17591 |
+
"step": 25060
|
| 17592 |
+
},
|
| 17593 |
+
{
|
| 17594 |
+
"epoch": 0.4652317227909264,
|
| 17595 |
+
"grad_norm": 36.1875,
|
| 17596 |
+
"learning_rate": 9.927307564391668e-06,
|
| 17597 |
+
"loss": 20.7705,
|
| 17598 |
+
"step": 25070
|
| 17599 |
+
},
|
| 17600 |
+
{
|
| 17601 |
+
"epoch": 0.46541729587540626,
|
| 17602 |
+
"grad_norm": 34.53125,
|
| 17603 |
+
"learning_rate": 9.927278568605625e-06,
|
| 17604 |
+
"loss": 20.7148,
|
| 17605 |
+
"step": 25080
|
| 17606 |
+
},
|
| 17607 |
+
{
|
| 17608 |
+
"epoch": 0.46560286895988606,
|
| 17609 |
+
"grad_norm": 34.875,
|
| 17610 |
+
"learning_rate": 9.927249572819583e-06,
|
| 17611 |
+
"loss": 20.4595,
|
| 17612 |
+
"step": 25090
|
| 17613 |
+
},
|
| 17614 |
+
{
|
| 17615 |
+
"epoch": 0.46578844204436587,
|
| 17616 |
+
"grad_norm": 35.46875,
|
| 17617 |
+
"learning_rate": 9.92722057703354e-06,
|
| 17618 |
+
"loss": 20.9314,
|
| 17619 |
+
"step": 25100
|
| 17620 |
+
},
|
| 17621 |
+
{
|
| 17622 |
+
"epoch": 0.46597401512884573,
|
| 17623 |
+
"grad_norm": 33.90625,
|
| 17624 |
+
"learning_rate": 9.927191581247499e-06,
|
| 17625 |
+
"loss": 20.3269,
|
| 17626 |
+
"step": 25110
|
| 17627 |
+
},
|
| 17628 |
+
{
|
| 17629 |
+
"epoch": 0.46615958821332554,
|
| 17630 |
+
"grad_norm": 36.3125,
|
| 17631 |
+
"learning_rate": 9.927162585461456e-06,
|
| 17632 |
+
"loss": 21.2732,
|
| 17633 |
+
"step": 25120
|
| 17634 |
+
},
|
| 17635 |
+
{
|
| 17636 |
+
"epoch": 0.46634516129780534,
|
| 17637 |
+
"grad_norm": 33.46875,
|
| 17638 |
+
"learning_rate": 9.927133589675412e-06,
|
| 17639 |
+
"loss": 21.0565,
|
| 17640 |
+
"step": 25130
|
| 17641 |
+
},
|
| 17642 |
+
{
|
| 17643 |
+
"epoch": 0.4665307343822852,
|
| 17644 |
+
"grad_norm": 35.53125,
|
| 17645 |
+
"learning_rate": 9.927104593889371e-06,
|
| 17646 |
+
"loss": 21.2097,
|
| 17647 |
+
"step": 25140
|
| 17648 |
+
},
|
| 17649 |
+
{
|
| 17650 |
+
"epoch": 0.466716307466765,
|
| 17651 |
+
"grad_norm": 34.78125,
|
| 17652 |
+
"learning_rate": 9.927075598103329e-06,
|
| 17653 |
+
"loss": 20.6643,
|
| 17654 |
+
"step": 25150
|
| 17655 |
+
},
|
| 17656 |
+
{
|
| 17657 |
+
"epoch": 0.4669018805512449,
|
| 17658 |
+
"grad_norm": 33.71875,
|
| 17659 |
+
"learning_rate": 9.927046602317286e-06,
|
| 17660 |
+
"loss": 20.9921,
|
| 17661 |
+
"step": 25160
|
| 17662 |
+
},
|
| 17663 |
+
{
|
| 17664 |
+
"epoch": 0.4670874536357247,
|
| 17665 |
+
"grad_norm": 31.984375,
|
| 17666 |
+
"learning_rate": 9.927017606531244e-06,
|
| 17667 |
+
"loss": 20.5886,
|
| 17668 |
+
"step": 25170
|
| 17669 |
+
},
|
| 17670 |
+
{
|
| 17671 |
+
"epoch": 0.4672730267202045,
|
| 17672 |
+
"grad_norm": 33.90625,
|
| 17673 |
+
"learning_rate": 9.926988610745201e-06,
|
| 17674 |
+
"loss": 20.725,
|
| 17675 |
+
"step": 25180
|
| 17676 |
+
},
|
| 17677 |
+
{
|
| 17678 |
+
"epoch": 0.46745859980468435,
|
| 17679 |
+
"grad_norm": 35.1875,
|
| 17680 |
+
"learning_rate": 9.926959614959158e-06,
|
| 17681 |
+
"loss": 20.9366,
|
| 17682 |
+
"step": 25190
|
| 17683 |
+
},
|
| 17684 |
+
{
|
| 17685 |
+
"epoch": 0.46764417288916416,
|
| 17686 |
+
"grad_norm": 34.6875,
|
| 17687 |
+
"learning_rate": 9.926930619173116e-06,
|
| 17688 |
+
"loss": 21.0935,
|
| 17689 |
+
"step": 25200
|
| 17690 |
+
},
|
| 17691 |
+
{
|
| 17692 |
+
"epoch": 0.46782974597364396,
|
| 17693 |
+
"grad_norm": 34.6875,
|
| 17694 |
+
"learning_rate": 9.926901623387075e-06,
|
| 17695 |
+
"loss": 20.3594,
|
| 17696 |
+
"step": 25210
|
| 17697 |
+
},
|
| 17698 |
+
{
|
| 17699 |
+
"epoch": 0.4680153190581238,
|
| 17700 |
+
"grad_norm": 36.96875,
|
| 17701 |
+
"learning_rate": 9.926872627601032e-06,
|
| 17702 |
+
"loss": 20.7214,
|
| 17703 |
+
"step": 25220
|
| 17704 |
+
},
|
| 17705 |
+
{
|
| 17706 |
+
"epoch": 0.46820089214260363,
|
| 17707 |
+
"grad_norm": 37.375,
|
| 17708 |
+
"learning_rate": 9.92684363181499e-06,
|
| 17709 |
+
"loss": 20.9528,
|
| 17710 |
+
"step": 25230
|
| 17711 |
+
},
|
| 17712 |
+
{
|
| 17713 |
+
"epoch": 0.46838646522708344,
|
| 17714 |
+
"grad_norm": 34.1875,
|
| 17715 |
+
"learning_rate": 9.926814636028947e-06,
|
| 17716 |
+
"loss": 20.7266,
|
| 17717 |
+
"step": 25240
|
| 17718 |
+
},
|
| 17719 |
+
{
|
| 17720 |
+
"epoch": 0.4685720383115633,
|
| 17721 |
+
"grad_norm": 34.5625,
|
| 17722 |
+
"learning_rate": 9.926785640242904e-06,
|
| 17723 |
+
"loss": 20.522,
|
| 17724 |
+
"step": 25250
|
| 17725 |
+
},
|
| 17726 |
+
{
|
| 17727 |
+
"epoch": 0.4687576113960431,
|
| 17728 |
+
"grad_norm": 33.65625,
|
| 17729 |
+
"learning_rate": 9.926756644456862e-06,
|
| 17730 |
+
"loss": 20.7444,
|
| 17731 |
+
"step": 25260
|
| 17732 |
+
},
|
| 17733 |
+
{
|
| 17734 |
+
"epoch": 0.46894318448052297,
|
| 17735 |
+
"grad_norm": 32.21875,
|
| 17736 |
+
"learning_rate": 9.92672764867082e-06,
|
| 17737 |
+
"loss": 20.1346,
|
| 17738 |
+
"step": 25270
|
| 17739 |
+
},
|
| 17740 |
+
{
|
| 17741 |
+
"epoch": 0.4691287575650028,
|
| 17742 |
+
"grad_norm": 34.96875,
|
| 17743 |
+
"learning_rate": 9.926698652884777e-06,
|
| 17744 |
+
"loss": 20.6407,
|
| 17745 |
+
"step": 25280
|
| 17746 |
+
},
|
| 17747 |
+
{
|
| 17748 |
+
"epoch": 0.4693143306494826,
|
| 17749 |
+
"grad_norm": 33.875,
|
| 17750 |
+
"learning_rate": 9.926669657098734e-06,
|
| 17751 |
+
"loss": 20.6497,
|
| 17752 |
+
"step": 25290
|
| 17753 |
+
},
|
| 17754 |
+
{
|
| 17755 |
+
"epoch": 0.46949990373396244,
|
| 17756 |
+
"grad_norm": 34.75,
|
| 17757 |
+
"learning_rate": 9.926640661312692e-06,
|
| 17758 |
+
"loss": 20.4089,
|
| 17759 |
+
"step": 25300
|
| 17760 |
+
},
|
| 17761 |
+
{
|
| 17762 |
+
"epoch": 0.46968547681844225,
|
| 17763 |
+
"grad_norm": 34.8125,
|
| 17764 |
+
"learning_rate": 9.92661166552665e-06,
|
| 17765 |
+
"loss": 20.2013,
|
| 17766 |
+
"step": 25310
|
| 17767 |
+
},
|
| 17768 |
+
{
|
| 17769 |
+
"epoch": 0.46987104990292206,
|
| 17770 |
+
"grad_norm": 34.84375,
|
| 17771 |
+
"learning_rate": 9.926582669740608e-06,
|
| 17772 |
+
"loss": 20.8357,
|
| 17773 |
+
"step": 25320
|
| 17774 |
+
},
|
| 17775 |
+
{
|
| 17776 |
+
"epoch": 0.4700566229874019,
|
| 17777 |
+
"grad_norm": 34.375,
|
| 17778 |
+
"learning_rate": 9.926553673954565e-06,
|
| 17779 |
+
"loss": 20.3461,
|
| 17780 |
+
"step": 25330
|
| 17781 |
+
},
|
| 17782 |
+
{
|
| 17783 |
+
"epoch": 0.4702421960718817,
|
| 17784 |
+
"grad_norm": 33.78125,
|
| 17785 |
+
"learning_rate": 9.926524678168523e-06,
|
| 17786 |
+
"loss": 20.1499,
|
| 17787 |
+
"step": 25340
|
| 17788 |
+
},
|
| 17789 |
+
{
|
| 17790 |
+
"epoch": 0.4704277691563616,
|
| 17791 |
+
"grad_norm": 35.71875,
|
| 17792 |
+
"learning_rate": 9.92649568238248e-06,
|
| 17793 |
+
"loss": 20.4552,
|
| 17794 |
+
"step": 25350
|
| 17795 |
+
},
|
| 17796 |
+
{
|
| 17797 |
+
"epoch": 0.4706133422408414,
|
| 17798 |
+
"grad_norm": 34.28125,
|
| 17799 |
+
"learning_rate": 9.926466686596438e-06,
|
| 17800 |
+
"loss": 20.6206,
|
| 17801 |
+
"step": 25360
|
| 17802 |
+
},
|
| 17803 |
+
{
|
| 17804 |
+
"epoch": 0.4707989153253212,
|
| 17805 |
+
"grad_norm": 35.28125,
|
| 17806 |
+
"learning_rate": 9.926437690810395e-06,
|
| 17807 |
+
"loss": 20.7498,
|
| 17808 |
+
"step": 25370
|
| 17809 |
+
},
|
| 17810 |
+
{
|
| 17811 |
+
"epoch": 0.47098448840980106,
|
| 17812 |
+
"grad_norm": 34.1875,
|
| 17813 |
+
"learning_rate": 9.926408695024352e-06,
|
| 17814 |
+
"loss": 20.6467,
|
| 17815 |
+
"step": 25380
|
| 17816 |
+
},
|
| 17817 |
+
{
|
| 17818 |
+
"epoch": 0.47117006149428087,
|
| 17819 |
+
"grad_norm": 33.0625,
|
| 17820 |
+
"learning_rate": 9.926379699238312e-06,
|
| 17821 |
+
"loss": 20.624,
|
| 17822 |
+
"step": 25390
|
| 17823 |
+
},
|
| 17824 |
+
{
|
| 17825 |
+
"epoch": 0.4713556345787607,
|
| 17826 |
+
"grad_norm": 33.21875,
|
| 17827 |
+
"learning_rate": 9.926350703452267e-06,
|
| 17828 |
+
"loss": 20.5546,
|
| 17829 |
+
"step": 25400
|
| 17830 |
+
},
|
| 17831 |
+
{
|
| 17832 |
+
"epoch": 0.47154120766324054,
|
| 17833 |
+
"grad_norm": 35.875,
|
| 17834 |
+
"learning_rate": 9.926321707666225e-06,
|
| 17835 |
+
"loss": 20.8728,
|
| 17836 |
+
"step": 25410
|
| 17837 |
+
},
|
| 17838 |
+
{
|
| 17839 |
+
"epoch": 0.47172678074772034,
|
| 17840 |
+
"grad_norm": 35.03125,
|
| 17841 |
+
"learning_rate": 9.926292711880184e-06,
|
| 17842 |
+
"loss": 20.7638,
|
| 17843 |
+
"step": 25420
|
| 17844 |
+
},
|
| 17845 |
+
{
|
| 17846 |
+
"epoch": 0.4719123538322002,
|
| 17847 |
+
"grad_norm": 33.6875,
|
| 17848 |
+
"learning_rate": 9.926263716094141e-06,
|
| 17849 |
+
"loss": 20.8226,
|
| 17850 |
+
"step": 25430
|
| 17851 |
+
},
|
| 17852 |
+
{
|
| 17853 |
+
"epoch": 0.47209792691668,
|
| 17854 |
+
"grad_norm": 35.09375,
|
| 17855 |
+
"learning_rate": 9.926234720308099e-06,
|
| 17856 |
+
"loss": 20.8098,
|
| 17857 |
+
"step": 25440
|
| 17858 |
+
},
|
| 17859 |
+
{
|
| 17860 |
+
"epoch": 0.4722835000011598,
|
| 17861 |
+
"grad_norm": 35.90625,
|
| 17862 |
+
"learning_rate": 9.926205724522056e-06,
|
| 17863 |
+
"loss": 20.0728,
|
| 17864 |
+
"step": 25450
|
| 17865 |
+
},
|
| 17866 |
+
{
|
| 17867 |
+
"epoch": 0.4724690730856397,
|
| 17868 |
+
"grad_norm": 35.65625,
|
| 17869 |
+
"learning_rate": 9.926176728736013e-06,
|
| 17870 |
+
"loss": 20.8324,
|
| 17871 |
+
"step": 25460
|
| 17872 |
+
},
|
| 17873 |
+
{
|
| 17874 |
+
"epoch": 0.4726546461701195,
|
| 17875 |
+
"grad_norm": 36.15625,
|
| 17876 |
+
"learning_rate": 9.92614773294997e-06,
|
| 17877 |
+
"loss": 20.5694,
|
| 17878 |
+
"step": 25470
|
| 17879 |
+
},
|
| 17880 |
+
{
|
| 17881 |
+
"epoch": 0.4728402192545993,
|
| 17882 |
+
"grad_norm": 34.4375,
|
| 17883 |
+
"learning_rate": 9.926118737163928e-06,
|
| 17884 |
+
"loss": 21.1526,
|
| 17885 |
+
"step": 25480
|
| 17886 |
+
},
|
| 17887 |
+
{
|
| 17888 |
+
"epoch": 0.47302579233907915,
|
| 17889 |
+
"grad_norm": 35.875,
|
| 17890 |
+
"learning_rate": 9.926089741377887e-06,
|
| 17891 |
+
"loss": 21.0011,
|
| 17892 |
+
"step": 25490
|
| 17893 |
+
},
|
| 17894 |
+
{
|
| 17895 |
+
"epoch": 0.47321136542355896,
|
| 17896 |
+
"grad_norm": 36.5625,
|
| 17897 |
+
"learning_rate": 9.926060745591843e-06,
|
| 17898 |
+
"loss": 20.2465,
|
| 17899 |
+
"step": 25500
|
| 17900 |
+
},
|
| 17901 |
+
{
|
| 17902 |
+
"epoch": 0.47339693850803877,
|
| 17903 |
+
"grad_norm": 37.40625,
|
| 17904 |
+
"learning_rate": 9.9260317498058e-06,
|
| 17905 |
+
"loss": 20.8743,
|
| 17906 |
+
"step": 25510
|
| 17907 |
+
},
|
| 17908 |
+
{
|
| 17909 |
+
"epoch": 0.47358251159251863,
|
| 17910 |
+
"grad_norm": 35.4375,
|
| 17911 |
+
"learning_rate": 9.92600275401976e-06,
|
| 17912 |
+
"loss": 20.9921,
|
| 17913 |
+
"step": 25520
|
| 17914 |
+
},
|
| 17915 |
+
{
|
| 17916 |
+
"epoch": 0.47376808467699844,
|
| 17917 |
+
"grad_norm": 33.03125,
|
| 17918 |
+
"learning_rate": 9.925973758233717e-06,
|
| 17919 |
+
"loss": 20.6059,
|
| 17920 |
+
"step": 25530
|
| 17921 |
+
},
|
| 17922 |
+
{
|
| 17923 |
+
"epoch": 0.4739536577614783,
|
| 17924 |
+
"grad_norm": 35.09375,
|
| 17925 |
+
"learning_rate": 9.925944762447674e-06,
|
| 17926 |
+
"loss": 20.8631,
|
| 17927 |
+
"step": 25540
|
| 17928 |
+
},
|
| 17929 |
+
{
|
| 17930 |
+
"epoch": 0.4741392308459581,
|
| 17931 |
+
"grad_norm": 35.0,
|
| 17932 |
+
"learning_rate": 9.925915766661632e-06,
|
| 17933 |
+
"loss": 20.3443,
|
| 17934 |
+
"step": 25550
|
| 17935 |
+
},
|
| 17936 |
+
{
|
| 17937 |
+
"epoch": 0.4743248039304379,
|
| 17938 |
+
"grad_norm": 36.75,
|
| 17939 |
+
"learning_rate": 9.92588677087559e-06,
|
| 17940 |
+
"loss": 20.7723,
|
| 17941 |
+
"step": 25560
|
| 17942 |
+
},
|
| 17943 |
+
{
|
| 17944 |
+
"epoch": 0.4745103770149178,
|
| 17945 |
+
"grad_norm": 34.6875,
|
| 17946 |
+
"learning_rate": 9.925857775089547e-06,
|
| 17947 |
+
"loss": 20.4458,
|
| 17948 |
+
"step": 25570
|
| 17949 |
+
},
|
| 17950 |
+
{
|
| 17951 |
+
"epoch": 0.4746959500993976,
|
| 17952 |
+
"grad_norm": 35.96875,
|
| 17953 |
+
"learning_rate": 9.925828779303504e-06,
|
| 17954 |
+
"loss": 20.3749,
|
| 17955 |
+
"step": 25580
|
| 17956 |
+
},
|
| 17957 |
+
{
|
| 17958 |
+
"epoch": 0.4748815231838774,
|
| 17959 |
+
"grad_norm": 36.625,
|
| 17960 |
+
"learning_rate": 9.925799783517463e-06,
|
| 17961 |
+
"loss": 20.3563,
|
| 17962 |
+
"step": 25590
|
| 17963 |
+
},
|
| 17964 |
+
{
|
| 17965 |
+
"epoch": 0.47506709626835725,
|
| 17966 |
+
"grad_norm": 34.875,
|
| 17967 |
+
"learning_rate": 9.92577078773142e-06,
|
| 17968 |
+
"loss": 20.9151,
|
| 17969 |
+
"step": 25600
|
| 17970 |
+
},
|
| 17971 |
+
{
|
| 17972 |
+
"epoch": 0.47525266935283705,
|
| 17973 |
+
"grad_norm": 35.28125,
|
| 17974 |
+
"learning_rate": 9.925741791945376e-06,
|
| 17975 |
+
"loss": 20.473,
|
| 17976 |
+
"step": 25610
|
| 17977 |
+
},
|
| 17978 |
+
{
|
| 17979 |
+
"epoch": 0.4754382424373169,
|
| 17980 |
+
"grad_norm": 36.3125,
|
| 17981 |
+
"learning_rate": 9.925712796159335e-06,
|
| 17982 |
+
"loss": 20.9318,
|
| 17983 |
+
"step": 25620
|
| 17984 |
+
},
|
| 17985 |
+
{
|
| 17986 |
+
"epoch": 0.4756238155217967,
|
| 17987 |
+
"grad_norm": 36.28125,
|
| 17988 |
+
"learning_rate": 9.925683800373293e-06,
|
| 17989 |
+
"loss": 20.9008,
|
| 17990 |
+
"step": 25630
|
| 17991 |
+
},
|
| 17992 |
+
{
|
| 17993 |
+
"epoch": 0.47580938860627653,
|
| 17994 |
+
"grad_norm": 35.71875,
|
| 17995 |
+
"learning_rate": 9.92565480458725e-06,
|
| 17996 |
+
"loss": 20.6804,
|
| 17997 |
+
"step": 25640
|
| 17998 |
+
},
|
| 17999 |
+
{
|
| 18000 |
+
"epoch": 0.4759949616907564,
|
| 18001 |
+
"grad_norm": 36.625,
|
| 18002 |
+
"learning_rate": 9.925625808801208e-06,
|
| 18003 |
+
"loss": 20.6793,
|
| 18004 |
+
"step": 25650
|
| 18005 |
+
},
|
| 18006 |
+
{
|
| 18007 |
+
"epoch": 0.4761805347752362,
|
| 18008 |
+
"grad_norm": 36.40625,
|
| 18009 |
+
"learning_rate": 9.925596813015167e-06,
|
| 18010 |
+
"loss": 20.6986,
|
| 18011 |
+
"step": 25660
|
| 18012 |
+
},
|
| 18013 |
+
{
|
| 18014 |
+
"epoch": 0.476366107859716,
|
| 18015 |
+
"grad_norm": 36.0,
|
| 18016 |
+
"learning_rate": 9.925567817229122e-06,
|
| 18017 |
+
"loss": 20.4197,
|
| 18018 |
+
"step": 25670
|
| 18019 |
+
},
|
| 18020 |
+
{
|
| 18021 |
+
"epoch": 0.47655168094419587,
|
| 18022 |
+
"grad_norm": 34.21875,
|
| 18023 |
+
"learning_rate": 9.92553882144308e-06,
|
| 18024 |
+
"loss": 20.203,
|
| 18025 |
+
"step": 25680
|
| 18026 |
+
},
|
| 18027 |
+
{
|
| 18028 |
+
"epoch": 0.47673725402867567,
|
| 18029 |
+
"grad_norm": 37.375,
|
| 18030 |
+
"learning_rate": 9.925509825657039e-06,
|
| 18031 |
+
"loss": 21.2316,
|
| 18032 |
+
"step": 25690
|
| 18033 |
+
},
|
| 18034 |
+
{
|
| 18035 |
+
"epoch": 0.47692282711315553,
|
| 18036 |
+
"grad_norm": 36.03125,
|
| 18037 |
+
"learning_rate": 9.925480829870996e-06,
|
| 18038 |
+
"loss": 20.5119,
|
| 18039 |
+
"step": 25700
|
| 18040 |
+
},
|
| 18041 |
+
{
|
| 18042 |
+
"epoch": 0.47710840019763534,
|
| 18043 |
+
"grad_norm": 34.25,
|
| 18044 |
+
"learning_rate": 9.925451834084954e-06,
|
| 18045 |
+
"loss": 20.4633,
|
| 18046 |
+
"step": 25710
|
| 18047 |
+
},
|
| 18048 |
+
{
|
| 18049 |
+
"epoch": 0.47729397328211515,
|
| 18050 |
+
"grad_norm": 36.15625,
|
| 18051 |
+
"learning_rate": 9.925422838298911e-06,
|
| 18052 |
+
"loss": 20.5057,
|
| 18053 |
+
"step": 25720
|
| 18054 |
+
},
|
| 18055 |
+
{
|
| 18056 |
+
"epoch": 0.477479546366595,
|
| 18057 |
+
"grad_norm": 35.3125,
|
| 18058 |
+
"learning_rate": 9.925393842512868e-06,
|
| 18059 |
+
"loss": 20.8923,
|
| 18060 |
+
"step": 25730
|
| 18061 |
+
},
|
| 18062 |
+
{
|
| 18063 |
+
"epoch": 0.4776651194510748,
|
| 18064 |
+
"grad_norm": 33.59375,
|
| 18065 |
+
"learning_rate": 9.925364846726826e-06,
|
| 18066 |
+
"loss": 20.7508,
|
| 18067 |
+
"step": 25740
|
| 18068 |
+
},
|
| 18069 |
+
{
|
| 18070 |
+
"epoch": 0.4778506925355546,
|
| 18071 |
+
"grad_norm": 34.5625,
|
| 18072 |
+
"learning_rate": 9.925335850940783e-06,
|
| 18073 |
+
"loss": 20.3849,
|
| 18074 |
+
"step": 25750
|
| 18075 |
+
},
|
| 18076 |
+
{
|
| 18077 |
+
"epoch": 0.4780362656200345,
|
| 18078 |
+
"grad_norm": 34.875,
|
| 18079 |
+
"learning_rate": 9.925306855154742e-06,
|
| 18080 |
+
"loss": 21.1023,
|
| 18081 |
+
"step": 25760
|
| 18082 |
+
},
|
| 18083 |
+
{
|
| 18084 |
+
"epoch": 0.4782218387045143,
|
| 18085 |
+
"grad_norm": 35.25,
|
| 18086 |
+
"learning_rate": 9.925277859368698e-06,
|
| 18087 |
+
"loss": 20.5539,
|
| 18088 |
+
"step": 25770
|
| 18089 |
+
},
|
| 18090 |
+
{
|
| 18091 |
+
"epoch": 0.4784074117889941,
|
| 18092 |
+
"grad_norm": 35.75,
|
| 18093 |
+
"learning_rate": 9.925248863582656e-06,
|
| 18094 |
+
"loss": 20.4852,
|
| 18095 |
+
"step": 25780
|
| 18096 |
+
},
|
| 18097 |
+
{
|
| 18098 |
+
"epoch": 0.47859298487347396,
|
| 18099 |
+
"grad_norm": 33.40625,
|
| 18100 |
+
"learning_rate": 9.925219867796615e-06,
|
| 18101 |
+
"loss": 20.5779,
|
| 18102 |
+
"step": 25790
|
| 18103 |
+
},
|
| 18104 |
+
{
|
| 18105 |
+
"epoch": 0.47877855795795377,
|
| 18106 |
+
"grad_norm": 34.96875,
|
| 18107 |
+
"learning_rate": 9.925190872010572e-06,
|
| 18108 |
+
"loss": 20.5704,
|
| 18109 |
+
"step": 25800
|
| 18110 |
+
},
|
| 18111 |
+
{
|
| 18112 |
+
"epoch": 0.4789641310424336,
|
| 18113 |
+
"grad_norm": 34.78125,
|
| 18114 |
+
"learning_rate": 9.92516187622453e-06,
|
| 18115 |
+
"loss": 20.5373,
|
| 18116 |
+
"step": 25810
|
| 18117 |
+
},
|
| 18118 |
+
{
|
| 18119 |
+
"epoch": 0.47914970412691343,
|
| 18120 |
+
"grad_norm": 34.5,
|
| 18121 |
+
"learning_rate": 9.925132880438487e-06,
|
| 18122 |
+
"loss": 20.758,
|
| 18123 |
+
"step": 25820
|
| 18124 |
+
},
|
| 18125 |
+
{
|
| 18126 |
+
"epoch": 0.47933527721139324,
|
| 18127 |
+
"grad_norm": 35.9375,
|
| 18128 |
+
"learning_rate": 9.925103884652444e-06,
|
| 18129 |
+
"loss": 20.5638,
|
| 18130 |
+
"step": 25830
|
| 18131 |
+
},
|
| 18132 |
+
{
|
| 18133 |
+
"epoch": 0.4795208502958731,
|
| 18134 |
+
"grad_norm": 33.5625,
|
| 18135 |
+
"learning_rate": 9.925074888866402e-06,
|
| 18136 |
+
"loss": 20.8422,
|
| 18137 |
+
"step": 25840
|
| 18138 |
+
},
|
| 18139 |
+
{
|
| 18140 |
+
"epoch": 0.4797064233803529,
|
| 18141 |
+
"grad_norm": 33.5625,
|
| 18142 |
+
"learning_rate": 9.925045893080359e-06,
|
| 18143 |
+
"loss": 20.8382,
|
| 18144 |
+
"step": 25850
|
| 18145 |
+
},
|
| 18146 |
+
{
|
| 18147 |
+
"epoch": 0.4798919964648327,
|
| 18148 |
+
"grad_norm": 35.375,
|
| 18149 |
+
"learning_rate": 9.925016897294316e-06,
|
| 18150 |
+
"loss": 20.6228,
|
| 18151 |
+
"step": 25860
|
| 18152 |
+
},
|
| 18153 |
+
{
|
| 18154 |
+
"epoch": 0.4800775695493126,
|
| 18155 |
+
"grad_norm": 33.96875,
|
| 18156 |
+
"learning_rate": 9.924987901508276e-06,
|
| 18157 |
+
"loss": 20.3393,
|
| 18158 |
+
"step": 25870
|
| 18159 |
+
},
|
| 18160 |
+
{
|
| 18161 |
+
"epoch": 0.4802631426337924,
|
| 18162 |
+
"grad_norm": 35.09375,
|
| 18163 |
+
"learning_rate": 9.924958905722231e-06,
|
| 18164 |
+
"loss": 20.7853,
|
| 18165 |
+
"step": 25880
|
| 18166 |
+
},
|
| 18167 |
+
{
|
| 18168 |
+
"epoch": 0.48044871571827225,
|
| 18169 |
+
"grad_norm": 33.53125,
|
| 18170 |
+
"learning_rate": 9.92492990993619e-06,
|
| 18171 |
+
"loss": 21.0041,
|
| 18172 |
+
"step": 25890
|
| 18173 |
+
},
|
| 18174 |
+
{
|
| 18175 |
+
"epoch": 0.48063428880275205,
|
| 18176 |
+
"grad_norm": 38.21875,
|
| 18177 |
+
"learning_rate": 9.924900914150148e-06,
|
| 18178 |
+
"loss": 20.4117,
|
| 18179 |
+
"step": 25900
|
| 18180 |
+
},
|
| 18181 |
+
{
|
| 18182 |
+
"epoch": 0.48081986188723186,
|
| 18183 |
+
"grad_norm": 34.0625,
|
| 18184 |
+
"learning_rate": 9.924871918364105e-06,
|
| 18185 |
+
"loss": 20.6509,
|
| 18186 |
+
"step": 25910
|
| 18187 |
+
},
|
| 18188 |
+
{
|
| 18189 |
+
"epoch": 0.4810054349717117,
|
| 18190 |
+
"grad_norm": 36.0,
|
| 18191 |
+
"learning_rate": 9.924842922578063e-06,
|
| 18192 |
+
"loss": 20.895,
|
| 18193 |
+
"step": 25920
|
| 18194 |
+
},
|
| 18195 |
+
{
|
| 18196 |
+
"epoch": 0.4811910080561915,
|
| 18197 |
+
"grad_norm": 36.28125,
|
| 18198 |
+
"learning_rate": 9.92481392679202e-06,
|
| 18199 |
+
"loss": 20.8673,
|
| 18200 |
+
"step": 25930
|
| 18201 |
+
},
|
| 18202 |
+
{
|
| 18203 |
+
"epoch": 0.48137658114067133,
|
| 18204 |
+
"grad_norm": 34.25,
|
| 18205 |
+
"learning_rate": 9.924784931005977e-06,
|
| 18206 |
+
"loss": 20.5624,
|
| 18207 |
+
"step": 25940
|
| 18208 |
+
},
|
| 18209 |
+
{
|
| 18210 |
+
"epoch": 0.4815621542251512,
|
| 18211 |
+
"grad_norm": 35.125,
|
| 18212 |
+
"learning_rate": 9.924755935219935e-06,
|
| 18213 |
+
"loss": 21.1527,
|
| 18214 |
+
"step": 25950
|
| 18215 |
+
},
|
| 18216 |
+
{
|
| 18217 |
+
"epoch": 0.481747727309631,
|
| 18218 |
+
"grad_norm": 36.71875,
|
| 18219 |
+
"learning_rate": 9.924726939433892e-06,
|
| 18220 |
+
"loss": 20.7524,
|
| 18221 |
+
"step": 25960
|
| 18222 |
+
},
|
| 18223 |
+
{
|
| 18224 |
+
"epoch": 0.48193330039411086,
|
| 18225 |
+
"grad_norm": 33.65625,
|
| 18226 |
+
"learning_rate": 9.924697943647851e-06,
|
| 18227 |
+
"loss": 21.1041,
|
| 18228 |
+
"step": 25970
|
| 18229 |
+
},
|
| 18230 |
+
{
|
| 18231 |
+
"epoch": 0.48211887347859067,
|
| 18232 |
+
"grad_norm": 36.59375,
|
| 18233 |
+
"learning_rate": 9.924668947861809e-06,
|
| 18234 |
+
"loss": 20.6638,
|
| 18235 |
+
"step": 25980
|
| 18236 |
+
},
|
| 18237 |
+
{
|
| 18238 |
+
"epoch": 0.4823044465630705,
|
| 18239 |
+
"grad_norm": 33.59375,
|
| 18240 |
+
"learning_rate": 9.924639952075764e-06,
|
| 18241 |
+
"loss": 20.6987,
|
| 18242 |
+
"step": 25990
|
| 18243 |
+
},
|
| 18244 |
+
{
|
| 18245 |
+
"epoch": 0.48249001964755034,
|
| 18246 |
+
"grad_norm": 34.625,
|
| 18247 |
+
"learning_rate": 9.924610956289724e-06,
|
| 18248 |
+
"loss": 20.7549,
|
| 18249 |
+
"step": 26000
|
| 18250 |
+
},
|
| 18251 |
+
{
|
| 18252 |
+
"epoch": 0.48267559273203015,
|
| 18253 |
+
"grad_norm": 36.125,
|
| 18254 |
+
"learning_rate": 9.924581960503681e-06,
|
| 18255 |
+
"loss": 20.5746,
|
| 18256 |
+
"step": 26010
|
| 18257 |
+
},
|
| 18258 |
+
{
|
| 18259 |
+
"epoch": 0.48286116581650995,
|
| 18260 |
+
"grad_norm": 33.40625,
|
| 18261 |
+
"learning_rate": 9.924552964717638e-06,
|
| 18262 |
+
"loss": 20.6575,
|
| 18263 |
+
"step": 26020
|
| 18264 |
+
},
|
| 18265 |
+
{
|
| 18266 |
+
"epoch": 0.4830467389009898,
|
| 18267 |
+
"grad_norm": 34.0625,
|
| 18268 |
+
"learning_rate": 9.924523968931596e-06,
|
| 18269 |
+
"loss": 20.1084,
|
| 18270 |
+
"step": 26030
|
| 18271 |
+
},
|
| 18272 |
+
{
|
| 18273 |
+
"epoch": 0.4832323119854696,
|
| 18274 |
+
"grad_norm": 37.03125,
|
| 18275 |
+
"learning_rate": 9.924494973145553e-06,
|
| 18276 |
+
"loss": 20.3671,
|
| 18277 |
+
"step": 26040
|
| 18278 |
+
},
|
| 18279 |
+
{
|
| 18280 |
+
"epoch": 0.4834178850699495,
|
| 18281 |
+
"grad_norm": 34.125,
|
| 18282 |
+
"learning_rate": 9.92446597735951e-06,
|
| 18283 |
+
"loss": 20.4181,
|
| 18284 |
+
"step": 26050
|
| 18285 |
+
},
|
| 18286 |
+
{
|
| 18287 |
+
"epoch": 0.4836034581544293,
|
| 18288 |
+
"grad_norm": 35.875,
|
| 18289 |
+
"learning_rate": 9.924436981573468e-06,
|
| 18290 |
+
"loss": 20.5477,
|
| 18291 |
+
"step": 26060
|
| 18292 |
+
},
|
| 18293 |
+
{
|
| 18294 |
+
"epoch": 0.4837890312389091,
|
| 18295 |
+
"grad_norm": 34.90625,
|
| 18296 |
+
"learning_rate": 9.924407985787427e-06,
|
| 18297 |
+
"loss": 20.2599,
|
| 18298 |
+
"step": 26070
|
| 18299 |
+
},
|
| 18300 |
+
{
|
| 18301 |
+
"epoch": 0.48397460432338896,
|
| 18302 |
+
"grad_norm": 33.1875,
|
| 18303 |
+
"learning_rate": 9.924378990001385e-06,
|
| 18304 |
+
"loss": 20.6862,
|
| 18305 |
+
"step": 26080
|
| 18306 |
+
},
|
| 18307 |
+
{
|
| 18308 |
+
"epoch": 0.48416017740786876,
|
| 18309 |
+
"grad_norm": 35.1875,
|
| 18310 |
+
"learning_rate": 9.92434999421534e-06,
|
| 18311 |
+
"loss": 20.6373,
|
| 18312 |
+
"step": 26090
|
| 18313 |
+
},
|
| 18314 |
+
{
|
| 18315 |
+
"epoch": 0.48434575049234857,
|
| 18316 |
+
"grad_norm": 34.8125,
|
| 18317 |
+
"learning_rate": 9.9243209984293e-06,
|
| 18318 |
+
"loss": 20.8698,
|
| 18319 |
+
"step": 26100
|
| 18320 |
+
},
|
| 18321 |
+
{
|
| 18322 |
+
"epoch": 0.48453132357682843,
|
| 18323 |
+
"grad_norm": 35.5625,
|
| 18324 |
+
"learning_rate": 9.924292002643257e-06,
|
| 18325 |
+
"loss": 20.5171,
|
| 18326 |
+
"step": 26110
|
| 18327 |
+
},
|
| 18328 |
+
{
|
| 18329 |
+
"epoch": 0.48471689666130824,
|
| 18330 |
+
"grad_norm": 36.0625,
|
| 18331 |
+
"learning_rate": 9.924263006857214e-06,
|
| 18332 |
+
"loss": 20.0415,
|
| 18333 |
+
"step": 26120
|
| 18334 |
+
},
|
| 18335 |
+
{
|
| 18336 |
+
"epoch": 0.48490246974578805,
|
| 18337 |
+
"grad_norm": 34.46875,
|
| 18338 |
+
"learning_rate": 9.924234011071172e-06,
|
| 18339 |
+
"loss": 20.0809,
|
| 18340 |
+
"step": 26130
|
| 18341 |
+
},
|
| 18342 |
+
{
|
| 18343 |
+
"epoch": 0.4850880428302679,
|
| 18344 |
+
"grad_norm": 35.5625,
|
| 18345 |
+
"learning_rate": 9.92420501528513e-06,
|
| 18346 |
+
"loss": 20.8953,
|
| 18347 |
+
"step": 26140
|
| 18348 |
+
},
|
| 18349 |
+
{
|
| 18350 |
+
"epoch": 0.4852736159147477,
|
| 18351 |
+
"grad_norm": 33.59375,
|
| 18352 |
+
"learning_rate": 9.924176019499086e-06,
|
| 18353 |
+
"loss": 20.5997,
|
| 18354 |
+
"step": 26150
|
| 18355 |
+
},
|
| 18356 |
+
{
|
| 18357 |
+
"epoch": 0.4854591889992276,
|
| 18358 |
+
"grad_norm": 33.6875,
|
| 18359 |
+
"learning_rate": 9.924147023713044e-06,
|
| 18360 |
+
"loss": 20.538,
|
| 18361 |
+
"step": 26160
|
| 18362 |
+
},
|
| 18363 |
+
{
|
| 18364 |
+
"epoch": 0.4856447620837074,
|
| 18365 |
+
"grad_norm": 32.59375,
|
| 18366 |
+
"learning_rate": 9.924118027927003e-06,
|
| 18367 |
+
"loss": 20.5563,
|
| 18368 |
+
"step": 26170
|
| 18369 |
+
},
|
| 18370 |
+
{
|
| 18371 |
+
"epoch": 0.4858303351681872,
|
| 18372 |
+
"grad_norm": 35.28125,
|
| 18373 |
+
"learning_rate": 9.92408903214096e-06,
|
| 18374 |
+
"loss": 20.3517,
|
| 18375 |
+
"step": 26180
|
| 18376 |
+
},
|
| 18377 |
+
{
|
| 18378 |
+
"epoch": 0.48601590825266705,
|
| 18379 |
+
"grad_norm": 34.34375,
|
| 18380 |
+
"learning_rate": 9.924060036354918e-06,
|
| 18381 |
+
"loss": 20.4909,
|
| 18382 |
+
"step": 26190
|
| 18383 |
+
},
|
| 18384 |
+
{
|
| 18385 |
+
"epoch": 0.48620148133714686,
|
| 18386 |
+
"grad_norm": 36.0625,
|
| 18387 |
+
"learning_rate": 9.924031040568875e-06,
|
| 18388 |
+
"loss": 20.3212,
|
| 18389 |
+
"step": 26200
|
| 18390 |
+
},
|
| 18391 |
+
{
|
| 18392 |
+
"epoch": 0.48638705442162666,
|
| 18393 |
+
"grad_norm": 33.84375,
|
| 18394 |
+
"learning_rate": 9.924002044782832e-06,
|
| 18395 |
+
"loss": 20.7427,
|
| 18396 |
+
"step": 26210
|
| 18397 |
+
},
|
| 18398 |
+
{
|
| 18399 |
+
"epoch": 0.4865726275061065,
|
| 18400 |
+
"grad_norm": 35.1875,
|
| 18401 |
+
"learning_rate": 9.92397304899679e-06,
|
| 18402 |
+
"loss": 20.5698,
|
| 18403 |
+
"step": 26220
|
| 18404 |
+
},
|
| 18405 |
+
{
|
| 18406 |
+
"epoch": 0.48675820059058633,
|
| 18407 |
+
"grad_norm": 38.0625,
|
| 18408 |
+
"learning_rate": 9.923944053210747e-06,
|
| 18409 |
+
"loss": 20.6616,
|
| 18410 |
+
"step": 26230
|
| 18411 |
+
},
|
| 18412 |
+
{
|
| 18413 |
+
"epoch": 0.4869437736750662,
|
| 18414 |
+
"grad_norm": 33.6875,
|
| 18415 |
+
"learning_rate": 9.923915057424706e-06,
|
| 18416 |
+
"loss": 20.3389,
|
| 18417 |
+
"step": 26240
|
| 18418 |
+
},
|
| 18419 |
+
{
|
| 18420 |
+
"epoch": 0.487129346759546,
|
| 18421 |
+
"grad_norm": 35.59375,
|
| 18422 |
+
"learning_rate": 9.923886061638664e-06,
|
| 18423 |
+
"loss": 20.3957,
|
| 18424 |
+
"step": 26250
|
| 18425 |
+
},
|
| 18426 |
+
{
|
| 18427 |
+
"epoch": 0.4873149198440258,
|
| 18428 |
+
"grad_norm": 34.25,
|
| 18429 |
+
"learning_rate": 9.92385706585262e-06,
|
| 18430 |
+
"loss": 20.4382,
|
| 18431 |
+
"step": 26260
|
| 18432 |
+
},
|
| 18433 |
+
{
|
| 18434 |
+
"epoch": 0.48750049292850567,
|
| 18435 |
+
"grad_norm": 34.84375,
|
| 18436 |
+
"learning_rate": 9.923828070066579e-06,
|
| 18437 |
+
"loss": 20.4939,
|
| 18438 |
+
"step": 26270
|
| 18439 |
+
},
|
| 18440 |
+
{
|
| 18441 |
+
"epoch": 0.4876860660129855,
|
| 18442 |
+
"grad_norm": 34.875,
|
| 18443 |
+
"learning_rate": 9.923799074280536e-06,
|
| 18444 |
+
"loss": 20.4227,
|
| 18445 |
+
"step": 26280
|
| 18446 |
+
},
|
| 18447 |
+
{
|
| 18448 |
+
"epoch": 0.4878716390974653,
|
| 18449 |
+
"grad_norm": 34.875,
|
| 18450 |
+
"learning_rate": 9.923770078494493e-06,
|
| 18451 |
+
"loss": 20.8318,
|
| 18452 |
+
"step": 26290
|
| 18453 |
+
},
|
| 18454 |
+
{
|
| 18455 |
+
"epoch": 0.48805721218194514,
|
| 18456 |
+
"grad_norm": 33.09375,
|
| 18457 |
+
"learning_rate": 9.923741082708451e-06,
|
| 18458 |
+
"loss": 20.5788,
|
| 18459 |
+
"step": 26300
|
| 18460 |
+
},
|
| 18461 |
+
{
|
| 18462 |
+
"epoch": 0.48824278526642495,
|
| 18463 |
+
"grad_norm": 35.0625,
|
| 18464 |
+
"learning_rate": 9.923712086922408e-06,
|
| 18465 |
+
"loss": 20.7702,
|
| 18466 |
+
"step": 26310
|
| 18467 |
+
},
|
| 18468 |
+
{
|
| 18469 |
+
"epoch": 0.4884283583509048,
|
| 18470 |
+
"grad_norm": 35.40625,
|
| 18471 |
+
"learning_rate": 9.923683091136366e-06,
|
| 18472 |
+
"loss": 20.8507,
|
| 18473 |
+
"step": 26320
|
| 18474 |
+
},
|
| 18475 |
+
{
|
| 18476 |
+
"epoch": 0.4886139314353846,
|
| 18477 |
+
"grad_norm": 34.75,
|
| 18478 |
+
"learning_rate": 9.923654095350323e-06,
|
| 18479 |
+
"loss": 20.8756,
|
| 18480 |
+
"step": 26330
|
| 18481 |
+
},
|
| 18482 |
+
{
|
| 18483 |
+
"epoch": 0.4887995045198644,
|
| 18484 |
+
"grad_norm": 34.0625,
|
| 18485 |
+
"learning_rate": 9.923625099564282e-06,
|
| 18486 |
+
"loss": 20.7579,
|
| 18487 |
+
"step": 26340
|
| 18488 |
+
},
|
| 18489 |
+
{
|
| 18490 |
+
"epoch": 0.4889850776043443,
|
| 18491 |
+
"grad_norm": 34.25,
|
| 18492 |
+
"learning_rate": 9.92359610377824e-06,
|
| 18493 |
+
"loss": 20.6045,
|
| 18494 |
+
"step": 26350
|
| 18495 |
+
},
|
| 18496 |
+
{
|
| 18497 |
+
"epoch": 0.4891706506888241,
|
| 18498 |
+
"grad_norm": 35.3125,
|
| 18499 |
+
"learning_rate": 9.923567107992195e-06,
|
| 18500 |
+
"loss": 20.2022,
|
| 18501 |
+
"step": 26360
|
| 18502 |
+
},
|
| 18503 |
+
{
|
| 18504 |
+
"epoch": 0.4893562237733039,
|
| 18505 |
+
"grad_norm": 34.0625,
|
| 18506 |
+
"learning_rate": 9.923538112206154e-06,
|
| 18507 |
+
"loss": 20.2575,
|
| 18508 |
+
"step": 26370
|
| 18509 |
+
},
|
| 18510 |
+
{
|
| 18511 |
+
"epoch": 0.48954179685778376,
|
| 18512 |
+
"grad_norm": 35.0625,
|
| 18513 |
+
"learning_rate": 9.923509116420112e-06,
|
| 18514 |
+
"loss": 20.9015,
|
| 18515 |
+
"step": 26380
|
| 18516 |
+
},
|
| 18517 |
+
{
|
| 18518 |
+
"epoch": 0.48972736994226357,
|
| 18519 |
+
"grad_norm": 33.875,
|
| 18520 |
+
"learning_rate": 9.92348012063407e-06,
|
| 18521 |
+
"loss": 20.532,
|
| 18522 |
+
"step": 26390
|
| 18523 |
+
},
|
| 18524 |
+
{
|
| 18525 |
+
"epoch": 0.4899129430267434,
|
| 18526 |
+
"grad_norm": 33.65625,
|
| 18527 |
+
"learning_rate": 9.923451124848027e-06,
|
| 18528 |
+
"loss": 20.6322,
|
| 18529 |
+
"step": 26400
|
| 18530 |
+
},
|
| 18531 |
+
{
|
| 18532 |
+
"epoch": 0.49009851611122324,
|
| 18533 |
+
"grad_norm": 34.5625,
|
| 18534 |
+
"learning_rate": 9.923422129061984e-06,
|
| 18535 |
+
"loss": 20.2689,
|
| 18536 |
+
"step": 26410
|
| 18537 |
+
},
|
| 18538 |
+
{
|
| 18539 |
+
"epoch": 0.49028408919570304,
|
| 18540 |
+
"grad_norm": 33.9375,
|
| 18541 |
+
"learning_rate": 9.923393133275941e-06,
|
| 18542 |
+
"loss": 20.7293,
|
| 18543 |
+
"step": 26420
|
| 18544 |
+
},
|
| 18545 |
+
{
|
| 18546 |
+
"epoch": 0.4904696622801829,
|
| 18547 |
+
"grad_norm": 34.5,
|
| 18548 |
+
"learning_rate": 9.923364137489899e-06,
|
| 18549 |
+
"loss": 20.89,
|
| 18550 |
+
"step": 26430
|
| 18551 |
+
},
|
| 18552 |
+
{
|
| 18553 |
+
"epoch": 0.4906552353646627,
|
| 18554 |
+
"grad_norm": 33.5625,
|
| 18555 |
+
"learning_rate": 9.923335141703856e-06,
|
| 18556 |
+
"loss": 19.9809,
|
| 18557 |
+
"step": 26440
|
| 18558 |
+
},
|
| 18559 |
+
{
|
| 18560 |
+
"epoch": 0.4908408084491425,
|
| 18561 |
+
"grad_norm": 36.375,
|
| 18562 |
+
"learning_rate": 9.923306145917815e-06,
|
| 18563 |
+
"loss": 20.7523,
|
| 18564 |
+
"step": 26450
|
| 18565 |
+
},
|
| 18566 |
+
{
|
| 18567 |
+
"epoch": 0.4910263815336224,
|
| 18568 |
+
"grad_norm": 36.5,
|
| 18569 |
+
"learning_rate": 9.923277150131773e-06,
|
| 18570 |
+
"loss": 20.9323,
|
| 18571 |
+
"step": 26460
|
| 18572 |
+
},
|
| 18573 |
+
{
|
| 18574 |
+
"epoch": 0.4912119546181022,
|
| 18575 |
+
"grad_norm": 37.65625,
|
| 18576 |
+
"learning_rate": 9.923248154345728e-06,
|
| 18577 |
+
"loss": 20.9648,
|
| 18578 |
+
"step": 26470
|
| 18579 |
+
},
|
| 18580 |
+
{
|
| 18581 |
+
"epoch": 0.491397527702582,
|
| 18582 |
+
"grad_norm": 33.28125,
|
| 18583 |
+
"learning_rate": 9.923219158559688e-06,
|
| 18584 |
+
"loss": 20.694,
|
| 18585 |
+
"step": 26480
|
| 18586 |
+
},
|
| 18587 |
+
{
|
| 18588 |
+
"epoch": 0.49158310078706186,
|
| 18589 |
+
"grad_norm": 34.75,
|
| 18590 |
+
"learning_rate": 9.923190162773645e-06,
|
| 18591 |
+
"loss": 20.6665,
|
| 18592 |
+
"step": 26490
|
| 18593 |
+
},
|
| 18594 |
+
{
|
| 18595 |
+
"epoch": 0.49176867387154166,
|
| 18596 |
+
"grad_norm": 34.25,
|
| 18597 |
+
"learning_rate": 9.923161166987602e-06,
|
| 18598 |
+
"loss": 20.4606,
|
| 18599 |
+
"step": 26500
|
| 18600 |
+
},
|
| 18601 |
+
{
|
| 18602 |
+
"epoch": 0.4919542469560215,
|
| 18603 |
+
"grad_norm": 35.375,
|
| 18604 |
+
"learning_rate": 9.92313217120156e-06,
|
| 18605 |
+
"loss": 20.5022,
|
| 18606 |
+
"step": 26510
|
| 18607 |
+
},
|
| 18608 |
+
{
|
| 18609 |
+
"epoch": 0.49213982004050133,
|
| 18610 |
+
"grad_norm": 34.9375,
|
| 18611 |
+
"learning_rate": 9.923103175415517e-06,
|
| 18612 |
+
"loss": 20.5076,
|
| 18613 |
+
"step": 26520
|
| 18614 |
+
},
|
| 18615 |
+
{
|
| 18616 |
+
"epoch": 0.49232539312498114,
|
| 18617 |
+
"grad_norm": 33.3125,
|
| 18618 |
+
"learning_rate": 9.923074179629475e-06,
|
| 18619 |
+
"loss": 20.7874,
|
| 18620 |
+
"step": 26530
|
| 18621 |
+
},
|
| 18622 |
+
{
|
| 18623 |
+
"epoch": 0.492510966209461,
|
| 18624 |
+
"grad_norm": 35.4375,
|
| 18625 |
+
"learning_rate": 9.923045183843432e-06,
|
| 18626 |
+
"loss": 20.9914,
|
| 18627 |
+
"step": 26540
|
| 18628 |
+
},
|
| 18629 |
+
{
|
| 18630 |
+
"epoch": 0.4926965392939408,
|
| 18631 |
+
"grad_norm": 35.25,
|
| 18632 |
+
"learning_rate": 9.923016188057391e-06,
|
| 18633 |
+
"loss": 19.7607,
|
| 18634 |
+
"step": 26550
|
| 18635 |
+
},
|
| 18636 |
+
{
|
| 18637 |
+
"epoch": 0.4928821123784206,
|
| 18638 |
+
"grad_norm": 34.9375,
|
| 18639 |
+
"learning_rate": 9.922987192271349e-06,
|
| 18640 |
+
"loss": 21.0187,
|
| 18641 |
+
"step": 26560
|
| 18642 |
+
},
|
| 18643 |
+
{
|
| 18644 |
+
"epoch": 0.4930676854629005,
|
| 18645 |
+
"grad_norm": 35.96875,
|
| 18646 |
+
"learning_rate": 9.922958196485306e-06,
|
| 18647 |
+
"loss": 20.2126,
|
| 18648 |
+
"step": 26570
|
| 18649 |
+
},
|
| 18650 |
+
{
|
| 18651 |
+
"epoch": 0.4932532585473803,
|
| 18652 |
+
"grad_norm": 35.0,
|
| 18653 |
+
"learning_rate": 9.922929200699263e-06,
|
| 18654 |
+
"loss": 20.2824,
|
| 18655 |
+
"step": 26580
|
| 18656 |
+
},
|
| 18657 |
+
{
|
| 18658 |
+
"epoch": 0.49343883163186014,
|
| 18659 |
+
"grad_norm": 37.78125,
|
| 18660 |
+
"learning_rate": 9.92290020491322e-06,
|
| 18661 |
+
"loss": 21.1446,
|
| 18662 |
+
"step": 26590
|
| 18663 |
+
},
|
| 18664 |
+
{
|
| 18665 |
+
"epoch": 0.49362440471633995,
|
| 18666 |
+
"grad_norm": 35.9375,
|
| 18667 |
+
"learning_rate": 9.922871209127178e-06,
|
| 18668 |
+
"loss": 20.6024,
|
| 18669 |
+
"step": 26600
|
| 18670 |
+
},
|
| 18671 |
+
{
|
| 18672 |
+
"epoch": 0.49380997780081975,
|
| 18673 |
+
"grad_norm": 34.75,
|
| 18674 |
+
"learning_rate": 9.922842213341136e-06,
|
| 18675 |
+
"loss": 20.4418,
|
| 18676 |
+
"step": 26610
|
| 18677 |
+
},
|
| 18678 |
+
{
|
| 18679 |
+
"epoch": 0.4939955508852996,
|
| 18680 |
+
"grad_norm": 33.78125,
|
| 18681 |
+
"learning_rate": 9.922813217555095e-06,
|
| 18682 |
+
"loss": 20.5188,
|
| 18683 |
+
"step": 26620
|
| 18684 |
+
},
|
| 18685 |
+
{
|
| 18686 |
+
"epoch": 0.4941811239697794,
|
| 18687 |
+
"grad_norm": 34.21875,
|
| 18688 |
+
"learning_rate": 9.92278422176905e-06,
|
| 18689 |
+
"loss": 20.2447,
|
| 18690 |
+
"step": 26630
|
| 18691 |
+
},
|
| 18692 |
+
{
|
| 18693 |
+
"epoch": 0.49436669705425923,
|
| 18694 |
+
"grad_norm": 35.59375,
|
| 18695 |
+
"learning_rate": 9.922755225983008e-06,
|
| 18696 |
+
"loss": 20.735,
|
| 18697 |
+
"step": 26640
|
| 18698 |
+
},
|
| 18699 |
+
{
|
| 18700 |
+
"epoch": 0.4945522701387391,
|
| 18701 |
+
"grad_norm": 37.03125,
|
| 18702 |
+
"learning_rate": 9.922726230196967e-06,
|
| 18703 |
+
"loss": 21.0627,
|
| 18704 |
+
"step": 26650
|
| 18705 |
+
},
|
| 18706 |
+
{
|
| 18707 |
+
"epoch": 0.4947378432232189,
|
| 18708 |
+
"grad_norm": 35.0625,
|
| 18709 |
+
"learning_rate": 9.922697234410924e-06,
|
| 18710 |
+
"loss": 20.1211,
|
| 18711 |
+
"step": 26660
|
| 18712 |
+
},
|
| 18713 |
+
{
|
| 18714 |
+
"epoch": 0.4949234163076987,
|
| 18715 |
+
"grad_norm": 34.90625,
|
| 18716 |
+
"learning_rate": 9.922668238624882e-06,
|
| 18717 |
+
"loss": 20.5947,
|
| 18718 |
+
"step": 26670
|
| 18719 |
+
},
|
| 18720 |
+
{
|
| 18721 |
+
"epoch": 0.49510898939217857,
|
| 18722 |
+
"grad_norm": 34.3125,
|
| 18723 |
+
"learning_rate": 9.922639242838839e-06,
|
| 18724 |
+
"loss": 20.4556,
|
| 18725 |
+
"step": 26680
|
| 18726 |
+
},
|
| 18727 |
+
{
|
| 18728 |
+
"epoch": 0.4952945624766584,
|
| 18729 |
+
"grad_norm": 32.75,
|
| 18730 |
+
"learning_rate": 9.922610247052797e-06,
|
| 18731 |
+
"loss": 20.0388,
|
| 18732 |
+
"step": 26690
|
| 18733 |
+
},
|
| 18734 |
+
{
|
| 18735 |
+
"epoch": 0.49548013556113824,
|
| 18736 |
+
"grad_norm": 33.4375,
|
| 18737 |
+
"learning_rate": 9.922581251266754e-06,
|
| 18738 |
+
"loss": 20.3324,
|
| 18739 |
+
"step": 26700
|
| 18740 |
+
},
|
| 18741 |
+
{
|
| 18742 |
+
"epoch": 0.49566570864561804,
|
| 18743 |
+
"grad_norm": 35.09375,
|
| 18744 |
+
"learning_rate": 9.922552255480711e-06,
|
| 18745 |
+
"loss": 20.444,
|
| 18746 |
+
"step": 26710
|
| 18747 |
+
},
|
| 18748 |
+
{
|
| 18749 |
+
"epoch": 0.49585128173009785,
|
| 18750 |
+
"grad_norm": 35.15625,
|
| 18751 |
+
"learning_rate": 9.92252325969467e-06,
|
| 18752 |
+
"loss": 20.5939,
|
| 18753 |
+
"step": 26720
|
| 18754 |
+
},
|
| 18755 |
+
{
|
| 18756 |
+
"epoch": 0.4960368548145777,
|
| 18757 |
+
"grad_norm": 35.84375,
|
| 18758 |
+
"learning_rate": 9.922494263908628e-06,
|
| 18759 |
+
"loss": 20.3899,
|
| 18760 |
+
"step": 26730
|
| 18761 |
+
},
|
| 18762 |
+
{
|
| 18763 |
+
"epoch": 0.4962224278990575,
|
| 18764 |
+
"grad_norm": 34.8125,
|
| 18765 |
+
"learning_rate": 9.922465268122584e-06,
|
| 18766 |
+
"loss": 20.5074,
|
| 18767 |
+
"step": 26740
|
| 18768 |
+
},
|
| 18769 |
+
{
|
| 18770 |
+
"epoch": 0.4964080009835373,
|
| 18771 |
+
"grad_norm": 37.03125,
|
| 18772 |
+
"learning_rate": 9.922436272336543e-06,
|
| 18773 |
+
"loss": 20.4035,
|
| 18774 |
+
"step": 26750
|
| 18775 |
+
},
|
| 18776 |
+
{
|
| 18777 |
+
"epoch": 0.4965935740680172,
|
| 18778 |
+
"grad_norm": 34.5,
|
| 18779 |
+
"learning_rate": 9.9224072765505e-06,
|
| 18780 |
+
"loss": 21.0219,
|
| 18781 |
+
"step": 26760
|
| 18782 |
+
},
|
| 18783 |
+
{
|
| 18784 |
+
"epoch": 0.496779147152497,
|
| 18785 |
+
"grad_norm": 33.75,
|
| 18786 |
+
"learning_rate": 9.922378280764457e-06,
|
| 18787 |
+
"loss": 20.5443,
|
| 18788 |
+
"step": 26770
|
| 18789 |
+
},
|
| 18790 |
+
{
|
| 18791 |
+
"epoch": 0.49696472023697685,
|
| 18792 |
+
"grad_norm": 35.09375,
|
| 18793 |
+
"learning_rate": 9.922349284978415e-06,
|
| 18794 |
+
"loss": 20.6201,
|
| 18795 |
+
"step": 26780
|
| 18796 |
+
},
|
| 18797 |
+
{
|
| 18798 |
+
"epoch": 0.49715029332145666,
|
| 18799 |
+
"grad_norm": 33.40625,
|
| 18800 |
+
"learning_rate": 9.922320289192372e-06,
|
| 18801 |
+
"loss": 20.0897,
|
| 18802 |
+
"step": 26790
|
| 18803 |
+
},
|
| 18804 |
+
{
|
| 18805 |
+
"epoch": 0.49733586640593647,
|
| 18806 |
+
"grad_norm": 33.90625,
|
| 18807 |
+
"learning_rate": 9.92229129340633e-06,
|
| 18808 |
+
"loss": 20.9902,
|
| 18809 |
+
"step": 26800
|
| 18810 |
+
},
|
| 18811 |
+
{
|
| 18812 |
+
"epoch": 0.49752143949041633,
|
| 18813 |
+
"grad_norm": 34.3125,
|
| 18814 |
+
"learning_rate": 9.922262297620287e-06,
|
| 18815 |
+
"loss": 21.3532,
|
| 18816 |
+
"step": 26810
|
| 18817 |
+
},
|
| 18818 |
+
{
|
| 18819 |
+
"epoch": 0.49770701257489613,
|
| 18820 |
+
"grad_norm": 34.375,
|
| 18821 |
+
"learning_rate": 9.922233301834246e-06,
|
| 18822 |
+
"loss": 20.4501,
|
| 18823 |
+
"step": 26820
|
| 18824 |
+
},
|
| 18825 |
+
{
|
| 18826 |
+
"epoch": 0.49789258565937594,
|
| 18827 |
+
"grad_norm": 34.8125,
|
| 18828 |
+
"learning_rate": 9.922204306048204e-06,
|
| 18829 |
+
"loss": 20.4959,
|
| 18830 |
+
"step": 26830
|
| 18831 |
+
},
|
| 18832 |
+
{
|
| 18833 |
+
"epoch": 0.4980781587438558,
|
| 18834 |
+
"grad_norm": 35.75,
|
| 18835 |
+
"learning_rate": 9.922175310262161e-06,
|
| 18836 |
+
"loss": 20.5544,
|
| 18837 |
+
"step": 26840
|
| 18838 |
+
},
|
| 18839 |
+
{
|
| 18840 |
+
"epoch": 0.4982637318283356,
|
| 18841 |
+
"grad_norm": 33.0,
|
| 18842 |
+
"learning_rate": 9.922146314476118e-06,
|
| 18843 |
+
"loss": 20.6951,
|
| 18844 |
+
"step": 26850
|
| 18845 |
+
},
|
| 18846 |
+
{
|
| 18847 |
+
"epoch": 0.49844930491281547,
|
| 18848 |
+
"grad_norm": 34.625,
|
| 18849 |
+
"learning_rate": 9.922117318690076e-06,
|
| 18850 |
+
"loss": 20.583,
|
| 18851 |
+
"step": 26860
|
| 18852 |
+
},
|
| 18853 |
+
{
|
| 18854 |
+
"epoch": 0.4986348779972953,
|
| 18855 |
+
"grad_norm": 36.875,
|
| 18856 |
+
"learning_rate": 9.922088322904033e-06,
|
| 18857 |
+
"loss": 20.3859,
|
| 18858 |
+
"step": 26870
|
| 18859 |
+
},
|
| 18860 |
+
{
|
| 18861 |
+
"epoch": 0.4988204510817751,
|
| 18862 |
+
"grad_norm": 34.0,
|
| 18863 |
+
"learning_rate": 9.92205932711799e-06,
|
| 18864 |
+
"loss": 20.0682,
|
| 18865 |
+
"step": 26880
|
| 18866 |
+
},
|
| 18867 |
+
{
|
| 18868 |
+
"epoch": 0.49900602416625495,
|
| 18869 |
+
"grad_norm": 35.375,
|
| 18870 |
+
"learning_rate": 9.922030331331948e-06,
|
| 18871 |
+
"loss": 20.5781,
|
| 18872 |
+
"step": 26890
|
| 18873 |
+
},
|
| 18874 |
+
{
|
| 18875 |
+
"epoch": 0.49919159725073475,
|
| 18876 |
+
"grad_norm": 34.75,
|
| 18877 |
+
"learning_rate": 9.922001335545905e-06,
|
| 18878 |
+
"loss": 20.8437,
|
| 18879 |
+
"step": 26900
|
| 18880 |
+
},
|
| 18881 |
+
{
|
| 18882 |
+
"epoch": 0.49937717033521456,
|
| 18883 |
+
"grad_norm": 34.4375,
|
| 18884 |
+
"learning_rate": 9.921972339759863e-06,
|
| 18885 |
+
"loss": 20.4446,
|
| 18886 |
+
"step": 26910
|
| 18887 |
+
},
|
| 18888 |
+
{
|
| 18889 |
+
"epoch": 0.4995627434196944,
|
| 18890 |
+
"grad_norm": 34.71875,
|
| 18891 |
+
"learning_rate": 9.92194334397382e-06,
|
| 18892 |
+
"loss": 20.7822,
|
| 18893 |
+
"step": 26920
|
| 18894 |
+
},
|
| 18895 |
+
{
|
| 18896 |
+
"epoch": 0.49974831650417423,
|
| 18897 |
+
"grad_norm": 34.46875,
|
| 18898 |
+
"learning_rate": 9.92191434818778e-06,
|
| 18899 |
+
"loss": 20.6735,
|
| 18900 |
+
"step": 26930
|
| 18901 |
+
},
|
| 18902 |
+
{
|
| 18903 |
+
"epoch": 0.49993388958865403,
|
| 18904 |
+
"grad_norm": 34.21875,
|
| 18905 |
+
"learning_rate": 9.921885352401737e-06,
|
| 18906 |
+
"loss": 21.0007,
|
| 18907 |
+
"step": 26940
|
| 18908 |
+
},
|
| 18909 |
+
{
|
| 18910 |
+
"epoch": 0.5001194626731339,
|
| 18911 |
+
"grad_norm": 33.46875,
|
| 18912 |
+
"learning_rate": 9.921856356615694e-06,
|
| 18913 |
+
"loss": 20.2895,
|
| 18914 |
+
"step": 26950
|
| 18915 |
+
},
|
| 18916 |
+
{
|
| 18917 |
+
"epoch": 0.5003050357576138,
|
| 18918 |
+
"grad_norm": 36.625,
|
| 18919 |
+
"learning_rate": 9.921827360829652e-06,
|
| 18920 |
+
"loss": 20.3101,
|
| 18921 |
+
"step": 26960
|
| 18922 |
+
},
|
| 18923 |
+
{
|
| 18924 |
+
"epoch": 0.5004906088420935,
|
| 18925 |
+
"grad_norm": 36.53125,
|
| 18926 |
+
"learning_rate": 9.921798365043609e-06,
|
| 18927 |
+
"loss": 20.482,
|
| 18928 |
+
"step": 26970
|
| 18929 |
+
},
|
| 18930 |
+
{
|
| 18931 |
+
"epoch": 0.5006761819265734,
|
| 18932 |
+
"grad_norm": 33.84375,
|
| 18933 |
+
"learning_rate": 9.921769369257566e-06,
|
| 18934 |
+
"loss": 20.8419,
|
| 18935 |
+
"step": 26980
|
| 18936 |
+
},
|
| 18937 |
+
{
|
| 18938 |
+
"epoch": 0.5008617550110532,
|
| 18939 |
+
"grad_norm": 36.25,
|
| 18940 |
+
"learning_rate": 9.921740373471524e-06,
|
| 18941 |
+
"loss": 20.4644,
|
| 18942 |
+
"step": 26990
|
| 18943 |
+
},
|
| 18944 |
+
{
|
| 18945 |
+
"epoch": 0.501047328095533,
|
| 18946 |
+
"grad_norm": 34.625,
|
| 18947 |
+
"learning_rate": 9.921711377685483e-06,
|
| 18948 |
+
"loss": 20.5031,
|
| 18949 |
+
"step": 27000
|
| 18950 |
+
},
|
| 18951 |
+
{
|
| 18952 |
+
"epoch": 0.5012329011800128,
|
| 18953 |
+
"grad_norm": 34.65625,
|
| 18954 |
+
"learning_rate": 9.921682381899439e-06,
|
| 18955 |
+
"loss": 20.6603,
|
| 18956 |
+
"step": 27010
|
| 18957 |
+
},
|
| 18958 |
+
{
|
| 18959 |
+
"epoch": 0.5014184742644927,
|
| 18960 |
+
"grad_norm": 33.78125,
|
| 18961 |
+
"learning_rate": 9.921653386113396e-06,
|
| 18962 |
+
"loss": 20.7035,
|
| 18963 |
+
"step": 27020
|
| 18964 |
+
},
|
| 18965 |
+
{
|
| 18966 |
+
"epoch": 0.5016040473489725,
|
| 18967 |
+
"grad_norm": 34.65625,
|
| 18968 |
+
"learning_rate": 9.921624390327355e-06,
|
| 18969 |
+
"loss": 21.0303,
|
| 18970 |
+
"step": 27030
|
| 18971 |
+
},
|
| 18972 |
+
{
|
| 18973 |
+
"epoch": 0.5017896204334523,
|
| 18974 |
+
"grad_norm": 33.5,
|
| 18975 |
+
"learning_rate": 9.921595394541313e-06,
|
| 18976 |
+
"loss": 20.2431,
|
| 18977 |
+
"step": 27040
|
| 18978 |
+
},
|
| 18979 |
+
{
|
| 18980 |
+
"epoch": 0.5019751935179322,
|
| 18981 |
+
"grad_norm": 36.8125,
|
| 18982 |
+
"learning_rate": 9.92156639875527e-06,
|
| 18983 |
+
"loss": 20.9503,
|
| 18984 |
+
"step": 27050
|
| 18985 |
+
},
|
| 18986 |
+
{
|
| 18987 |
+
"epoch": 0.5021607666024119,
|
| 18988 |
+
"grad_norm": 36.25,
|
| 18989 |
+
"learning_rate": 9.921537402969227e-06,
|
| 18990 |
+
"loss": 20.5597,
|
| 18991 |
+
"step": 27060
|
| 18992 |
+
},
|
| 18993 |
+
{
|
| 18994 |
+
"epoch": 0.5023463396868918,
|
| 18995 |
+
"grad_norm": 35.53125,
|
| 18996 |
+
"learning_rate": 9.921508407183185e-06,
|
| 18997 |
+
"loss": 20.2874,
|
| 18998 |
+
"step": 27070
|
| 18999 |
+
},
|
| 19000 |
+
{
|
| 19001 |
+
"epoch": 0.5025319127713717,
|
| 19002 |
+
"grad_norm": 34.5625,
|
| 19003 |
+
"learning_rate": 9.921479411397142e-06,
|
| 19004 |
+
"loss": 20.2342,
|
| 19005 |
+
"step": 27080
|
| 19006 |
+
},
|
| 19007 |
+
{
|
| 19008 |
+
"epoch": 0.5027174858558515,
|
| 19009 |
+
"grad_norm": 35.625,
|
| 19010 |
+
"learning_rate": 9.9214504156111e-06,
|
| 19011 |
+
"loss": 20.2432,
|
| 19012 |
+
"step": 27090
|
| 19013 |
+
},
|
| 19014 |
+
{
|
| 19015 |
+
"epoch": 0.5029030589403313,
|
| 19016 |
+
"grad_norm": 36.71875,
|
| 19017 |
+
"learning_rate": 9.921421419825059e-06,
|
| 19018 |
+
"loss": 20.8367,
|
| 19019 |
+
"step": 27100
|
| 19020 |
+
},
|
| 19021 |
+
{
|
| 19022 |
+
"epoch": 0.5030886320248111,
|
| 19023 |
+
"grad_norm": 35.75,
|
| 19024 |
+
"learning_rate": 9.921392424039014e-06,
|
| 19025 |
+
"loss": 20.4609,
|
| 19026 |
+
"step": 27110
|
| 19027 |
+
},
|
| 19028 |
+
{
|
| 19029 |
+
"epoch": 0.503274205109291,
|
| 19030 |
+
"grad_norm": 35.5,
|
| 19031 |
+
"learning_rate": 9.921363428252972e-06,
|
| 19032 |
+
"loss": 20.4391,
|
| 19033 |
+
"step": 27120
|
| 19034 |
+
},
|
| 19035 |
+
{
|
| 19036 |
+
"epoch": 0.5034597781937707,
|
| 19037 |
+
"grad_norm": 34.0625,
|
| 19038 |
+
"learning_rate": 9.921334432466931e-06,
|
| 19039 |
+
"loss": 20.4591,
|
| 19040 |
+
"step": 27130
|
| 19041 |
+
},
|
| 19042 |
+
{
|
| 19043 |
+
"epoch": 0.5036453512782506,
|
| 19044 |
+
"grad_norm": 35.03125,
|
| 19045 |
+
"learning_rate": 9.921305436680888e-06,
|
| 19046 |
+
"loss": 20.8571,
|
| 19047 |
+
"step": 27140
|
| 19048 |
+
},
|
| 19049 |
+
{
|
| 19050 |
+
"epoch": 0.5038309243627305,
|
| 19051 |
+
"grad_norm": 38.15625,
|
| 19052 |
+
"learning_rate": 9.921276440894846e-06,
|
| 19053 |
+
"loss": 20.877,
|
| 19054 |
+
"step": 27150
|
| 19055 |
+
},
|
| 19056 |
+
{
|
| 19057 |
+
"epoch": 0.5040164974472102,
|
| 19058 |
+
"grad_norm": 35.125,
|
| 19059 |
+
"learning_rate": 9.921247445108803e-06,
|
| 19060 |
+
"loss": 20.7956,
|
| 19061 |
+
"step": 27160
|
| 19062 |
+
},
|
| 19063 |
+
{
|
| 19064 |
+
"epoch": 0.5042020705316901,
|
| 19065 |
+
"grad_norm": 35.46875,
|
| 19066 |
+
"learning_rate": 9.92121844932276e-06,
|
| 19067 |
+
"loss": 20.4186,
|
| 19068 |
+
"step": 27170
|
| 19069 |
+
},
|
| 19070 |
+
{
|
| 19071 |
+
"epoch": 0.50438764361617,
|
| 19072 |
+
"grad_norm": 35.25,
|
| 19073 |
+
"learning_rate": 9.921189453536718e-06,
|
| 19074 |
+
"loss": 20.5766,
|
| 19075 |
+
"step": 27180
|
| 19076 |
+
},
|
| 19077 |
+
{
|
| 19078 |
+
"epoch": 0.5045732167006497,
|
| 19079 |
+
"grad_norm": 35.40625,
|
| 19080 |
+
"learning_rate": 9.921160457750675e-06,
|
| 19081 |
+
"loss": 20.2852,
|
| 19082 |
+
"step": 27190
|
| 19083 |
+
},
|
| 19084 |
+
{
|
| 19085 |
+
"epoch": 0.5047587897851296,
|
| 19086 |
+
"grad_norm": 36.28125,
|
| 19087 |
+
"learning_rate": 9.921131461964634e-06,
|
| 19088 |
+
"loss": 20.5424,
|
| 19089 |
+
"step": 27200
|
| 19090 |
+
},
|
| 19091 |
+
{
|
| 19092 |
+
"epoch": 0.5049443628696094,
|
| 19093 |
+
"grad_norm": 35.4375,
|
| 19094 |
+
"learning_rate": 9.921102466178592e-06,
|
| 19095 |
+
"loss": 20.4194,
|
| 19096 |
+
"step": 27210
|
| 19097 |
+
},
|
| 19098 |
+
{
|
| 19099 |
+
"epoch": 0.5051299359540892,
|
| 19100 |
+
"grad_norm": 33.0625,
|
| 19101 |
+
"learning_rate": 9.921073470392548e-06,
|
| 19102 |
+
"loss": 20.407,
|
| 19103 |
+
"step": 27220
|
| 19104 |
+
},
|
| 19105 |
+
{
|
| 19106 |
+
"epoch": 0.505315509038569,
|
| 19107 |
+
"grad_norm": 35.65625,
|
| 19108 |
+
"learning_rate": 9.921044474606507e-06,
|
| 19109 |
+
"loss": 20.6512,
|
| 19110 |
+
"step": 27230
|
| 19111 |
+
},
|
| 19112 |
+
{
|
| 19113 |
+
"epoch": 0.5055010821230489,
|
| 19114 |
+
"grad_norm": 34.09375,
|
| 19115 |
+
"learning_rate": 9.921015478820464e-06,
|
| 19116 |
+
"loss": 20.7855,
|
| 19117 |
+
"step": 27240
|
| 19118 |
+
},
|
| 19119 |
+
{
|
| 19120 |
+
"epoch": 0.5056866552075286,
|
| 19121 |
+
"grad_norm": 36.6875,
|
| 19122 |
+
"learning_rate": 9.920986483034421e-06,
|
| 19123 |
+
"loss": 20.5273,
|
| 19124 |
+
"step": 27250
|
| 19125 |
+
},
|
| 19126 |
+
{
|
| 19127 |
+
"epoch": 0.5058722282920085,
|
| 19128 |
+
"grad_norm": 33.96875,
|
| 19129 |
+
"learning_rate": 9.920957487248379e-06,
|
| 19130 |
+
"loss": 20.1875,
|
| 19131 |
+
"step": 27260
|
| 19132 |
+
},
|
| 19133 |
+
{
|
| 19134 |
+
"epoch": 0.5060578013764884,
|
| 19135 |
+
"grad_norm": 32.96875,
|
| 19136 |
+
"learning_rate": 9.920928491462336e-06,
|
| 19137 |
+
"loss": 20.7281,
|
| 19138 |
+
"step": 27270
|
| 19139 |
+
},
|
| 19140 |
+
{
|
| 19141 |
+
"epoch": 0.5062433744609682,
|
| 19142 |
+
"grad_norm": 34.5,
|
| 19143 |
+
"learning_rate": 9.920899495676294e-06,
|
| 19144 |
+
"loss": 20.489,
|
| 19145 |
+
"step": 27280
|
| 19146 |
+
},
|
| 19147 |
+
{
|
| 19148 |
+
"epoch": 0.506428947545448,
|
| 19149 |
+
"grad_norm": 32.625,
|
| 19150 |
+
"learning_rate": 9.920870499890251e-06,
|
| 19151 |
+
"loss": 20.938,
|
| 19152 |
+
"step": 27290
|
| 19153 |
+
},
|
| 19154 |
+
{
|
| 19155 |
+
"epoch": 0.5066145206299278,
|
| 19156 |
+
"grad_norm": 34.46875,
|
| 19157 |
+
"learning_rate": 9.92084150410421e-06,
|
| 19158 |
+
"loss": 20.6521,
|
| 19159 |
+
"step": 27300
|
| 19160 |
+
},
|
| 19161 |
+
{
|
| 19162 |
+
"epoch": 0.5068000937144077,
|
| 19163 |
+
"grad_norm": 34.1875,
|
| 19164 |
+
"learning_rate": 9.920812508318168e-06,
|
| 19165 |
+
"loss": 20.7266,
|
| 19166 |
+
"step": 27310
|
| 19167 |
+
},
|
| 19168 |
+
{
|
| 19169 |
+
"epoch": 0.5069856667988875,
|
| 19170 |
+
"grad_norm": 35.125,
|
| 19171 |
+
"learning_rate": 9.920783512532125e-06,
|
| 19172 |
+
"loss": 20.4679,
|
| 19173 |
+
"step": 27320
|
| 19174 |
+
},
|
| 19175 |
+
{
|
| 19176 |
+
"epoch": 0.5071712398833673,
|
| 19177 |
+
"grad_norm": 34.25,
|
| 19178 |
+
"learning_rate": 9.920754516746082e-06,
|
| 19179 |
+
"loss": 20.5604,
|
| 19180 |
+
"step": 27330
|
| 19181 |
+
},
|
| 19182 |
+
{
|
| 19183 |
+
"epoch": 0.5073568129678472,
|
| 19184 |
+
"grad_norm": 35.90625,
|
| 19185 |
+
"learning_rate": 9.92072552096004e-06,
|
| 19186 |
+
"loss": 20.9174,
|
| 19187 |
+
"step": 27340
|
| 19188 |
+
},
|
| 19189 |
+
{
|
| 19190 |
+
"epoch": 0.5075423860523269,
|
| 19191 |
+
"grad_norm": 34.90625,
|
| 19192 |
+
"learning_rate": 9.920696525173997e-06,
|
| 19193 |
+
"loss": 20.5337,
|
| 19194 |
+
"step": 27350
|
| 19195 |
+
},
|
| 19196 |
+
{
|
| 19197 |
+
"epoch": 0.5077279591368068,
|
| 19198 |
+
"grad_norm": 35.375,
|
| 19199 |
+
"learning_rate": 9.920667529387955e-06,
|
| 19200 |
+
"loss": 20.4249,
|
| 19201 |
+
"step": 27360
|
| 19202 |
+
},
|
| 19203 |
+
{
|
| 19204 |
+
"epoch": 0.5079135322212867,
|
| 19205 |
+
"grad_norm": 34.125,
|
| 19206 |
+
"learning_rate": 9.920638533601912e-06,
|
| 19207 |
+
"loss": 19.9978,
|
| 19208 |
+
"step": 27370
|
| 19209 |
+
},
|
| 19210 |
+
{
|
| 19211 |
+
"epoch": 0.5080991053057664,
|
| 19212 |
+
"grad_norm": 32.4375,
|
| 19213 |
+
"learning_rate": 9.92060953781587e-06,
|
| 19214 |
+
"loss": 20.4063,
|
| 19215 |
+
"step": 27380
|
| 19216 |
+
},
|
| 19217 |
+
{
|
| 19218 |
+
"epoch": 0.5082846783902463,
|
| 19219 |
+
"grad_norm": 36.59375,
|
| 19220 |
+
"learning_rate": 9.920580542029827e-06,
|
| 19221 |
+
"loss": 20.2915,
|
| 19222 |
+
"step": 27390
|
| 19223 |
+
},
|
| 19224 |
+
{
|
| 19225 |
+
"epoch": 0.5084702514747261,
|
| 19226 |
+
"grad_norm": 34.90625,
|
| 19227 |
+
"learning_rate": 9.920551546243786e-06,
|
| 19228 |
+
"loss": 20.5562,
|
| 19229 |
+
"step": 27400
|
| 19230 |
+
},
|
| 19231 |
+
{
|
| 19232 |
+
"epoch": 0.5086558245592059,
|
| 19233 |
+
"grad_norm": 36.875,
|
| 19234 |
+
"learning_rate": 9.920522550457743e-06,
|
| 19235 |
+
"loss": 20.6528,
|
| 19236 |
+
"step": 27410
|
| 19237 |
+
},
|
| 19238 |
+
{
|
| 19239 |
+
"epoch": 0.5088413976436857,
|
| 19240 |
+
"grad_norm": 34.09375,
|
| 19241 |
+
"learning_rate": 9.9204935546717e-06,
|
| 19242 |
+
"loss": 20.4691,
|
| 19243 |
+
"step": 27420
|
| 19244 |
+
},
|
| 19245 |
+
{
|
| 19246 |
+
"epoch": 0.5090269707281656,
|
| 19247 |
+
"grad_norm": 35.09375,
|
| 19248 |
+
"learning_rate": 9.920464558885658e-06,
|
| 19249 |
+
"loss": 20.3996,
|
| 19250 |
+
"step": 27430
|
| 19251 |
+
},
|
| 19252 |
+
{
|
| 19253 |
+
"epoch": 0.5092125438126455,
|
| 19254 |
+
"grad_norm": 34.8125,
|
| 19255 |
+
"learning_rate": 9.920435563099616e-06,
|
| 19256 |
+
"loss": 20.4117,
|
| 19257 |
+
"step": 27440
|
| 19258 |
+
},
|
| 19259 |
+
{
|
| 19260 |
+
"epoch": 0.5093981168971252,
|
| 19261 |
+
"grad_norm": 36.125,
|
| 19262 |
+
"learning_rate": 9.920406567313573e-06,
|
| 19263 |
+
"loss": 20.6251,
|
| 19264 |
+
"step": 27450
|
| 19265 |
+
},
|
| 19266 |
+
{
|
| 19267 |
+
"epoch": 0.5095836899816051,
|
| 19268 |
+
"grad_norm": 35.84375,
|
| 19269 |
+
"learning_rate": 9.92037757152753e-06,
|
| 19270 |
+
"loss": 20.3157,
|
| 19271 |
+
"step": 27460
|
| 19272 |
+
},
|
| 19273 |
+
{
|
| 19274 |
+
"epoch": 0.5097692630660849,
|
| 19275 |
+
"grad_norm": 33.46875,
|
| 19276 |
+
"learning_rate": 9.920348575741488e-06,
|
| 19277 |
+
"loss": 20.4708,
|
| 19278 |
+
"step": 27470
|
| 19279 |
+
},
|
| 19280 |
+
{
|
| 19281 |
+
"epoch": 0.5099548361505647,
|
| 19282 |
+
"grad_norm": 34.375,
|
| 19283 |
+
"learning_rate": 9.920319579955447e-06,
|
| 19284 |
+
"loss": 20.7527,
|
| 19285 |
+
"step": 27480
|
| 19286 |
+
},
|
| 19287 |
+
{
|
| 19288 |
+
"epoch": 0.5101404092350446,
|
| 19289 |
+
"grad_norm": 34.25,
|
| 19290 |
+
"learning_rate": 9.920290584169403e-06,
|
| 19291 |
+
"loss": 20.7704,
|
| 19292 |
+
"step": 27490
|
| 19293 |
+
},
|
| 19294 |
+
{
|
| 19295 |
+
"epoch": 0.5103259823195244,
|
| 19296 |
+
"grad_norm": 35.90625,
|
| 19297 |
+
"learning_rate": 9.92026158838336e-06,
|
| 19298 |
+
"loss": 20.2382,
|
| 19299 |
+
"step": 27500
|
| 19300 |
+
},
|
| 19301 |
+
{
|
| 19302 |
+
"epoch": 0.5105115554040042,
|
| 19303 |
+
"grad_norm": 36.3125,
|
| 19304 |
+
"learning_rate": 9.920232592597319e-06,
|
| 19305 |
+
"loss": 20.698,
|
| 19306 |
+
"step": 27510
|
| 19307 |
+
},
|
| 19308 |
+
{
|
| 19309 |
+
"epoch": 0.510697128488484,
|
| 19310 |
+
"grad_norm": 33.9375,
|
| 19311 |
+
"learning_rate": 9.920203596811277e-06,
|
| 19312 |
+
"loss": 20.4556,
|
| 19313 |
+
"step": 27520
|
| 19314 |
+
},
|
| 19315 |
+
{
|
| 19316 |
+
"epoch": 0.5108827015729639,
|
| 19317 |
+
"grad_norm": 34.5625,
|
| 19318 |
+
"learning_rate": 9.920174601025234e-06,
|
| 19319 |
+
"loss": 20.5015,
|
| 19320 |
+
"step": 27530
|
| 19321 |
+
},
|
| 19322 |
+
{
|
| 19323 |
+
"epoch": 0.5110682746574436,
|
| 19324 |
+
"grad_norm": 36.5,
|
| 19325 |
+
"learning_rate": 9.920145605239191e-06,
|
| 19326 |
+
"loss": 20.179,
|
| 19327 |
+
"step": 27540
|
| 19328 |
+
},
|
| 19329 |
+
{
|
| 19330 |
+
"epoch": 0.5112538477419235,
|
| 19331 |
+
"grad_norm": 33.75,
|
| 19332 |
+
"learning_rate": 9.920116609453149e-06,
|
| 19333 |
+
"loss": 20.2646,
|
| 19334 |
+
"step": 27550
|
| 19335 |
+
},
|
| 19336 |
+
{
|
| 19337 |
+
"epoch": 0.5114394208264034,
|
| 19338 |
+
"grad_norm": 34.75,
|
| 19339 |
+
"learning_rate": 9.920087613667106e-06,
|
| 19340 |
+
"loss": 20.6251,
|
| 19341 |
+
"step": 27560
|
| 19342 |
+
},
|
| 19343 |
+
{
|
| 19344 |
+
"epoch": 0.5116249939108831,
|
| 19345 |
+
"grad_norm": 34.34375,
|
| 19346 |
+
"learning_rate": 9.920058617881064e-06,
|
| 19347 |
+
"loss": 19.9915,
|
| 19348 |
+
"step": 27570
|
| 19349 |
+
},
|
| 19350 |
+
{
|
| 19351 |
+
"epoch": 0.511810566995363,
|
| 19352 |
+
"grad_norm": 34.78125,
|
| 19353 |
+
"learning_rate": 9.920029622095023e-06,
|
| 19354 |
+
"loss": 20.6833,
|
| 19355 |
+
"step": 27580
|
| 19356 |
+
},
|
| 19357 |
+
{
|
| 19358 |
+
"epoch": 0.5119961400798428,
|
| 19359 |
+
"grad_norm": 32.09375,
|
| 19360 |
+
"learning_rate": 9.92000062630898e-06,
|
| 19361 |
+
"loss": 20.4652,
|
| 19362 |
+
"step": 27590
|
| 19363 |
+
},
|
| 19364 |
+
{
|
| 19365 |
+
"epoch": 0.5121817131643226,
|
| 19366 |
+
"grad_norm": 36.53125,
|
| 19367 |
+
"learning_rate": 9.919971630522936e-06,
|
| 19368 |
+
"loss": 20.854,
|
| 19369 |
+
"step": 27600
|
| 19370 |
+
},
|
| 19371 |
+
{
|
| 19372 |
+
"epoch": 0.5123672862488025,
|
| 19373 |
+
"grad_norm": 34.0,
|
| 19374 |
+
"learning_rate": 9.919942634736895e-06,
|
| 19375 |
+
"loss": 20.5599,
|
| 19376 |
+
"step": 27610
|
| 19377 |
+
},
|
| 19378 |
+
{
|
| 19379 |
+
"epoch": 0.5125528593332823,
|
| 19380 |
+
"grad_norm": 33.34375,
|
| 19381 |
+
"learning_rate": 9.919913638950852e-06,
|
| 19382 |
+
"loss": 20.8087,
|
| 19383 |
+
"step": 27620
|
| 19384 |
+
},
|
| 19385 |
+
{
|
| 19386 |
+
"epoch": 0.5127384324177622,
|
| 19387 |
+
"grad_norm": 34.59375,
|
| 19388 |
+
"learning_rate": 9.91988464316481e-06,
|
| 19389 |
+
"loss": 20.2152,
|
| 19390 |
+
"step": 27630
|
| 19391 |
+
},
|
| 19392 |
+
{
|
| 19393 |
+
"epoch": 0.5129240055022419,
|
| 19394 |
+
"grad_norm": 35.34375,
|
| 19395 |
+
"learning_rate": 9.919855647378767e-06,
|
| 19396 |
+
"loss": 20.4706,
|
| 19397 |
+
"step": 27640
|
| 19398 |
+
},
|
| 19399 |
+
{
|
| 19400 |
+
"epoch": 0.5131095785867218,
|
| 19401 |
+
"grad_norm": 33.71875,
|
| 19402 |
+
"learning_rate": 9.919826651592725e-06,
|
| 19403 |
+
"loss": 20.428,
|
| 19404 |
+
"step": 27650
|
| 19405 |
+
},
|
| 19406 |
+
{
|
| 19407 |
+
"epoch": 0.5132951516712017,
|
| 19408 |
+
"grad_norm": 34.25,
|
| 19409 |
+
"learning_rate": 9.919797655806682e-06,
|
| 19410 |
+
"loss": 20.24,
|
| 19411 |
+
"step": 27660
|
| 19412 |
+
},
|
| 19413 |
+
{
|
| 19414 |
+
"epoch": 0.5134807247556814,
|
| 19415 |
+
"grad_norm": 35.21875,
|
| 19416 |
+
"learning_rate": 9.91976866002064e-06,
|
| 19417 |
+
"loss": 20.7069,
|
| 19418 |
+
"step": 27670
|
| 19419 |
+
},
|
| 19420 |
+
{
|
| 19421 |
+
"epoch": 0.5136662978401613,
|
| 19422 |
+
"grad_norm": 36.5,
|
| 19423 |
+
"learning_rate": 9.919739664234598e-06,
|
| 19424 |
+
"loss": 20.4921,
|
| 19425 |
+
"step": 27680
|
| 19426 |
+
},
|
| 19427 |
+
{
|
| 19428 |
+
"epoch": 0.5138518709246411,
|
| 19429 |
+
"grad_norm": 34.21875,
|
| 19430 |
+
"learning_rate": 9.919710668448556e-06,
|
| 19431 |
+
"loss": 20.3458,
|
| 19432 |
+
"step": 27690
|
| 19433 |
+
},
|
| 19434 |
+
{
|
| 19435 |
+
"epoch": 0.5140374440091209,
|
| 19436 |
+
"grad_norm": 34.0,
|
| 19437 |
+
"learning_rate": 9.919681672662512e-06,
|
| 19438 |
+
"loss": 20.4302,
|
| 19439 |
+
"step": 27700
|
| 19440 |
+
},
|
| 19441 |
+
{
|
| 19442 |
+
"epoch": 0.5142230170936007,
|
| 19443 |
+
"grad_norm": 33.96875,
|
| 19444 |
+
"learning_rate": 9.91965267687647e-06,
|
| 19445 |
+
"loss": 20.6798,
|
| 19446 |
+
"step": 27710
|
| 19447 |
+
},
|
| 19448 |
+
{
|
| 19449 |
+
"epoch": 0.5144085901780806,
|
| 19450 |
+
"grad_norm": 36.96875,
|
| 19451 |
+
"learning_rate": 9.919623681090428e-06,
|
| 19452 |
+
"loss": 20.571,
|
| 19453 |
+
"step": 27720
|
| 19454 |
+
},
|
| 19455 |
+
{
|
| 19456 |
+
"epoch": 0.5145941632625604,
|
| 19457 |
+
"grad_norm": 34.59375,
|
| 19458 |
+
"learning_rate": 9.919594685304385e-06,
|
| 19459 |
+
"loss": 19.6917,
|
| 19460 |
+
"step": 27730
|
| 19461 |
+
},
|
| 19462 |
+
{
|
| 19463 |
+
"epoch": 0.5147797363470402,
|
| 19464 |
+
"grad_norm": 34.40625,
|
| 19465 |
+
"learning_rate": 9.919565689518343e-06,
|
| 19466 |
+
"loss": 20.6663,
|
| 19467 |
+
"step": 27740
|
| 19468 |
+
},
|
| 19469 |
+
{
|
| 19470 |
+
"epoch": 0.5149653094315201,
|
| 19471 |
+
"grad_norm": 35.71875,
|
| 19472 |
+
"learning_rate": 9.919536693732302e-06,
|
| 19473 |
+
"loss": 20.3554,
|
| 19474 |
+
"step": 27750
|
| 19475 |
+
},
|
| 19476 |
+
{
|
| 19477 |
+
"epoch": 0.5151508825159998,
|
| 19478 |
+
"grad_norm": 34.6875,
|
| 19479 |
+
"learning_rate": 9.919507697946258e-06,
|
| 19480 |
+
"loss": 20.623,
|
| 19481 |
+
"step": 27760
|
| 19482 |
+
},
|
| 19483 |
+
{
|
| 19484 |
+
"epoch": 0.5153364556004797,
|
| 19485 |
+
"grad_norm": 34.0625,
|
| 19486 |
+
"learning_rate": 9.919478702160215e-06,
|
| 19487 |
+
"loss": 20.3847,
|
| 19488 |
+
"step": 27770
|
| 19489 |
+
},
|
| 19490 |
+
{
|
| 19491 |
+
"epoch": 0.5155220286849596,
|
| 19492 |
+
"grad_norm": 38.625,
|
| 19493 |
+
"learning_rate": 9.919449706374174e-06,
|
| 19494 |
+
"loss": 20.4449,
|
| 19495 |
+
"step": 27780
|
| 19496 |
+
},
|
| 19497 |
+
{
|
| 19498 |
+
"epoch": 0.5157076017694393,
|
| 19499 |
+
"grad_norm": 34.03125,
|
| 19500 |
+
"learning_rate": 9.919420710588132e-06,
|
| 19501 |
+
"loss": 20.3852,
|
| 19502 |
+
"step": 27790
|
| 19503 |
+
},
|
| 19504 |
+
{
|
| 19505 |
+
"epoch": 0.5158931748539192,
|
| 19506 |
+
"grad_norm": 36.0625,
|
| 19507 |
+
"learning_rate": 9.919391714802089e-06,
|
| 19508 |
+
"loss": 20.4759,
|
| 19509 |
+
"step": 27800
|
| 19510 |
+
},
|
| 19511 |
+
{
|
| 19512 |
+
"epoch": 0.516078747938399,
|
| 19513 |
+
"grad_norm": 34.4375,
|
| 19514 |
+
"learning_rate": 9.919362719016046e-06,
|
| 19515 |
+
"loss": 20.3306,
|
| 19516 |
+
"step": 27810
|
| 19517 |
+
},
|
| 19518 |
+
{
|
| 19519 |
+
"epoch": 0.5162643210228789,
|
| 19520 |
+
"grad_norm": 32.90625,
|
| 19521 |
+
"learning_rate": 9.919333723230004e-06,
|
| 19522 |
+
"loss": 20.121,
|
| 19523 |
+
"step": 27820
|
| 19524 |
+
},
|
| 19525 |
+
{
|
| 19526 |
+
"epoch": 0.5164498941073586,
|
| 19527 |
+
"grad_norm": 35.15625,
|
| 19528 |
+
"learning_rate": 9.919304727443961e-06,
|
| 19529 |
+
"loss": 20.4499,
|
| 19530 |
+
"step": 27830
|
| 19531 |
+
},
|
| 19532 |
+
{
|
| 19533 |
+
"epoch": 0.5166354671918385,
|
| 19534 |
+
"grad_norm": 37.40625,
|
| 19535 |
+
"learning_rate": 9.919275731657919e-06,
|
| 19536 |
+
"loss": 20.4448,
|
| 19537 |
+
"step": 27840
|
| 19538 |
+
},
|
| 19539 |
+
{
|
| 19540 |
+
"epoch": 0.5168210402763184,
|
| 19541 |
+
"grad_norm": 34.96875,
|
| 19542 |
+
"learning_rate": 9.919246735871878e-06,
|
| 19543 |
+
"loss": 20.3247,
|
| 19544 |
+
"step": 27850
|
| 19545 |
+
},
|
| 19546 |
+
{
|
| 19547 |
+
"epoch": 0.5170066133607981,
|
| 19548 |
+
"grad_norm": 36.6875,
|
| 19549 |
+
"learning_rate": 9.919217740085833e-06,
|
| 19550 |
+
"loss": 20.2849,
|
| 19551 |
+
"step": 27860
|
| 19552 |
+
},
|
| 19553 |
+
{
|
| 19554 |
+
"epoch": 0.517192186445278,
|
| 19555 |
+
"grad_norm": 34.21875,
|
| 19556 |
+
"learning_rate": 9.919188744299791e-06,
|
| 19557 |
+
"loss": 20.6523,
|
| 19558 |
+
"step": 27870
|
| 19559 |
+
},
|
| 19560 |
+
{
|
| 19561 |
+
"epoch": 0.5173777595297578,
|
| 19562 |
+
"grad_norm": 34.59375,
|
| 19563 |
+
"learning_rate": 9.91915974851375e-06,
|
| 19564 |
+
"loss": 20.3804,
|
| 19565 |
+
"step": 27880
|
| 19566 |
+
},
|
| 19567 |
+
{
|
| 19568 |
+
"epoch": 0.5175633326142376,
|
| 19569 |
+
"grad_norm": 35.15625,
|
| 19570 |
+
"learning_rate": 9.919130752727707e-06,
|
| 19571 |
+
"loss": 20.2975,
|
| 19572 |
+
"step": 27890
|
| 19573 |
+
},
|
| 19574 |
+
{
|
| 19575 |
+
"epoch": 0.5177489056987175,
|
| 19576 |
+
"grad_norm": 35.84375,
|
| 19577 |
+
"learning_rate": 9.919101756941665e-06,
|
| 19578 |
+
"loss": 20.4741,
|
| 19579 |
+
"step": 27900
|
| 19580 |
+
},
|
| 19581 |
+
{
|
| 19582 |
+
"epoch": 0.5179344787831973,
|
| 19583 |
+
"grad_norm": 35.3125,
|
| 19584 |
+
"learning_rate": 9.919072761155622e-06,
|
| 19585 |
+
"loss": 20.7598,
|
| 19586 |
+
"step": 27910
|
| 19587 |
+
},
|
| 19588 |
+
{
|
| 19589 |
+
"epoch": 0.5181200518676771,
|
| 19590 |
+
"grad_norm": 34.9375,
|
| 19591 |
+
"learning_rate": 9.91904376536958e-06,
|
| 19592 |
+
"loss": 19.9323,
|
| 19593 |
+
"step": 27920
|
| 19594 |
+
},
|
| 19595 |
+
{
|
| 19596 |
+
"epoch": 0.5183056249521569,
|
| 19597 |
+
"grad_norm": 35.09375,
|
| 19598 |
+
"learning_rate": 9.919014769583537e-06,
|
| 19599 |
+
"loss": 20.2018,
|
| 19600 |
+
"step": 27930
|
| 19601 |
+
},
|
| 19602 |
+
{
|
| 19603 |
+
"epoch": 0.5184911980366368,
|
| 19604 |
+
"grad_norm": 38.0,
|
| 19605 |
+
"learning_rate": 9.918985773797494e-06,
|
| 19606 |
+
"loss": 20.4047,
|
| 19607 |
+
"step": 27940
|
| 19608 |
+
},
|
| 19609 |
+
{
|
| 19610 |
+
"epoch": 0.5186767711211165,
|
| 19611 |
+
"grad_norm": 34.34375,
|
| 19612 |
+
"learning_rate": 9.918956778011452e-06,
|
| 19613 |
+
"loss": 20.4592,
|
| 19614 |
+
"step": 27950
|
| 19615 |
+
},
|
| 19616 |
+
{
|
| 19617 |
+
"epoch": 0.5188623442055964,
|
| 19618 |
+
"grad_norm": 34.15625,
|
| 19619 |
+
"learning_rate": 9.918927782225411e-06,
|
| 19620 |
+
"loss": 20.2424,
|
| 19621 |
+
"step": 27960
|
| 19622 |
+
},
|
| 19623 |
+
{
|
| 19624 |
+
"epoch": 0.5190479172900763,
|
| 19625 |
+
"grad_norm": 36.0625,
|
| 19626 |
+
"learning_rate": 9.918898786439367e-06,
|
| 19627 |
+
"loss": 20.5012,
|
| 19628 |
+
"step": 27970
|
| 19629 |
+
},
|
| 19630 |
+
{
|
| 19631 |
+
"epoch": 0.5192334903745561,
|
| 19632 |
+
"grad_norm": 34.53125,
|
| 19633 |
+
"learning_rate": 9.918869790653324e-06,
|
| 19634 |
+
"loss": 20.6693,
|
| 19635 |
+
"step": 27980
|
| 19636 |
+
},
|
| 19637 |
+
{
|
| 19638 |
+
"epoch": 0.5194190634590359,
|
| 19639 |
+
"grad_norm": 35.78125,
|
| 19640 |
+
"learning_rate": 9.918840794867283e-06,
|
| 19641 |
+
"loss": 20.6029,
|
| 19642 |
+
"step": 27990
|
| 19643 |
+
},
|
| 19644 |
+
{
|
| 19645 |
+
"epoch": 0.5196046365435157,
|
| 19646 |
+
"grad_norm": 33.6875,
|
| 19647 |
+
"learning_rate": 9.91881179908124e-06,
|
| 19648 |
+
"loss": 20.3617,
|
| 19649 |
+
"step": 28000
|
| 19650 |
+
},
|
| 19651 |
+
{
|
| 19652 |
+
"epoch": 0.5197902096279956,
|
| 19653 |
+
"grad_norm": 32.34375,
|
| 19654 |
+
"learning_rate": 9.918782803295198e-06,
|
| 19655 |
+
"loss": 20.85,
|
| 19656 |
+
"step": 28010
|
| 19657 |
+
},
|
| 19658 |
+
{
|
| 19659 |
+
"epoch": 0.5199757827124754,
|
| 19660 |
+
"grad_norm": 36.375,
|
| 19661 |
+
"learning_rate": 9.918753807509155e-06,
|
| 19662 |
+
"loss": 20.1941,
|
| 19663 |
+
"step": 28020
|
| 19664 |
+
},
|
| 19665 |
+
{
|
| 19666 |
+
"epoch": 0.5201613557969552,
|
| 19667 |
+
"grad_norm": 35.78125,
|
| 19668 |
+
"learning_rate": 9.918724811723113e-06,
|
| 19669 |
+
"loss": 20.3447,
|
| 19670 |
+
"step": 28030
|
| 19671 |
+
},
|
| 19672 |
+
{
|
| 19673 |
+
"epoch": 0.5203469288814351,
|
| 19674 |
+
"grad_norm": 35.0625,
|
| 19675 |
+
"learning_rate": 9.91869581593707e-06,
|
| 19676 |
+
"loss": 20.8577,
|
| 19677 |
+
"step": 28040
|
| 19678 |
+
},
|
| 19679 |
+
{
|
| 19680 |
+
"epoch": 0.5205325019659148,
|
| 19681 |
+
"grad_norm": 36.0625,
|
| 19682 |
+
"learning_rate": 9.918666820151028e-06,
|
| 19683 |
+
"loss": 20.098,
|
| 19684 |
+
"step": 28050
|
| 19685 |
+
},
|
| 19686 |
+
{
|
| 19687 |
+
"epoch": 0.5207180750503947,
|
| 19688 |
+
"grad_norm": 34.90625,
|
| 19689 |
+
"learning_rate": 9.918637824364987e-06,
|
| 19690 |
+
"loss": 20.6016,
|
| 19691 |
+
"step": 28060
|
| 19692 |
+
},
|
| 19693 |
+
{
|
| 19694 |
+
"epoch": 0.5209036481348746,
|
| 19695 |
+
"grad_norm": 32.3125,
|
| 19696 |
+
"learning_rate": 9.918608828578944e-06,
|
| 19697 |
+
"loss": 20.3424,
|
| 19698 |
+
"step": 28070
|
| 19699 |
+
},
|
| 19700 |
+
{
|
| 19701 |
+
"epoch": 0.5210892212193543,
|
| 19702 |
+
"grad_norm": 36.46875,
|
| 19703 |
+
"learning_rate": 9.9185798327929e-06,
|
| 19704 |
+
"loss": 20.371,
|
| 19705 |
+
"step": 28080
|
| 19706 |
+
},
|
| 19707 |
+
{
|
| 19708 |
+
"epoch": 0.5212747943038342,
|
| 19709 |
+
"grad_norm": 35.96875,
|
| 19710 |
+
"learning_rate": 9.918550837006859e-06,
|
| 19711 |
+
"loss": 20.4307,
|
| 19712 |
+
"step": 28090
|
| 19713 |
+
},
|
| 19714 |
+
{
|
| 19715 |
+
"epoch": 0.521460367388314,
|
| 19716 |
+
"grad_norm": 35.34375,
|
| 19717 |
+
"learning_rate": 9.918521841220816e-06,
|
| 19718 |
+
"loss": 20.6163,
|
| 19719 |
+
"step": 28100
|
| 19720 |
+
},
|
| 19721 |
+
{
|
| 19722 |
+
"epoch": 0.5216459404727938,
|
| 19723 |
+
"grad_norm": 34.25,
|
| 19724 |
+
"learning_rate": 9.918492845434774e-06,
|
| 19725 |
+
"loss": 20.8442,
|
| 19726 |
+
"step": 28110
|
| 19727 |
+
},
|
| 19728 |
+
{
|
| 19729 |
+
"epoch": 0.5218315135572736,
|
| 19730 |
+
"grad_norm": 34.5,
|
| 19731 |
+
"learning_rate": 9.918463849648731e-06,
|
| 19732 |
+
"loss": 20.2093,
|
| 19733 |
+
"step": 28120
|
| 19734 |
+
},
|
| 19735 |
+
{
|
| 19736 |
+
"epoch": 0.5220170866417535,
|
| 19737 |
+
"grad_norm": 34.8125,
|
| 19738 |
+
"learning_rate": 9.918434853862689e-06,
|
| 19739 |
+
"loss": 20.2828,
|
| 19740 |
+
"step": 28130
|
| 19741 |
+
},
|
| 19742 |
+
{
|
| 19743 |
+
"epoch": 0.5222026597262333,
|
| 19744 |
+
"grad_norm": 37.0625,
|
| 19745 |
+
"learning_rate": 9.918405858076646e-06,
|
| 19746 |
+
"loss": 19.9799,
|
| 19747 |
+
"step": 28140
|
| 19748 |
+
},
|
| 19749 |
+
{
|
| 19750 |
+
"epoch": 0.5223882328107131,
|
| 19751 |
+
"grad_norm": 34.5,
|
| 19752 |
+
"learning_rate": 9.918376862290603e-06,
|
| 19753 |
+
"loss": 20.411,
|
| 19754 |
+
"step": 28150
|
| 19755 |
+
},
|
| 19756 |
+
{
|
| 19757 |
+
"epoch": 0.522573805895193,
|
| 19758 |
+
"grad_norm": 34.6875,
|
| 19759 |
+
"learning_rate": 9.918347866504562e-06,
|
| 19760 |
+
"loss": 20.6861,
|
| 19761 |
+
"step": 28160
|
| 19762 |
+
},
|
| 19763 |
+
{
|
| 19764 |
+
"epoch": 0.5227593789796728,
|
| 19765 |
+
"grad_norm": 36.09375,
|
| 19766 |
+
"learning_rate": 9.91831887071852e-06,
|
| 19767 |
+
"loss": 20.4573,
|
| 19768 |
+
"step": 28170
|
| 19769 |
+
},
|
| 19770 |
+
{
|
| 19771 |
+
"epoch": 0.5229449520641526,
|
| 19772 |
+
"grad_norm": 33.90625,
|
| 19773 |
+
"learning_rate": 9.918289874932477e-06,
|
| 19774 |
+
"loss": 20.773,
|
| 19775 |
+
"step": 28180
|
| 19776 |
+
},
|
| 19777 |
+
{
|
| 19778 |
+
"epoch": 0.5231305251486325,
|
| 19779 |
+
"grad_norm": 36.25,
|
| 19780 |
+
"learning_rate": 9.918260879146435e-06,
|
| 19781 |
+
"loss": 20.5433,
|
| 19782 |
+
"step": 28190
|
| 19783 |
+
},
|
| 19784 |
+
{
|
| 19785 |
+
"epoch": 0.5233160982331123,
|
| 19786 |
+
"grad_norm": 34.5625,
|
| 19787 |
+
"learning_rate": 9.918231883360392e-06,
|
| 19788 |
+
"loss": 20.2934,
|
| 19789 |
+
"step": 28200
|
| 19790 |
+
},
|
| 19791 |
+
{
|
| 19792 |
+
"epoch": 0.5235016713175921,
|
| 19793 |
+
"grad_norm": 38.0625,
|
| 19794 |
+
"learning_rate": 9.91820288757435e-06,
|
| 19795 |
+
"loss": 20.8898,
|
| 19796 |
+
"step": 28210
|
| 19797 |
+
},
|
| 19798 |
+
{
|
| 19799 |
+
"epoch": 0.5236872444020719,
|
| 19800 |
+
"grad_norm": 33.90625,
|
| 19801 |
+
"learning_rate": 9.918173891788307e-06,
|
| 19802 |
+
"loss": 20.4065,
|
| 19803 |
+
"step": 28220
|
| 19804 |
+
},
|
| 19805 |
+
{
|
| 19806 |
+
"epoch": 0.5238728174865518,
|
| 19807 |
+
"grad_norm": 35.375,
|
| 19808 |
+
"learning_rate": 9.918144896002266e-06,
|
| 19809 |
+
"loss": 20.3046,
|
| 19810 |
+
"step": 28230
|
| 19811 |
+
},
|
| 19812 |
+
{
|
| 19813 |
+
"epoch": 0.5240583905710315,
|
| 19814 |
+
"grad_norm": 33.71875,
|
| 19815 |
+
"learning_rate": 9.918115900216222e-06,
|
| 19816 |
+
"loss": 19.9315,
|
| 19817 |
+
"step": 28240
|
| 19818 |
+
},
|
| 19819 |
+
{
|
| 19820 |
+
"epoch": 0.5242439636555114,
|
| 19821 |
+
"grad_norm": 35.71875,
|
| 19822 |
+
"learning_rate": 9.918086904430179e-06,
|
| 19823 |
+
"loss": 20.2579,
|
| 19824 |
+
"step": 28250
|
| 19825 |
+
},
|
| 19826 |
+
{
|
| 19827 |
+
"epoch": 0.5244295367399913,
|
| 19828 |
+
"grad_norm": 35.09375,
|
| 19829 |
+
"learning_rate": 9.918057908644138e-06,
|
| 19830 |
+
"loss": 20.2974,
|
| 19831 |
+
"step": 28260
|
| 19832 |
+
},
|
| 19833 |
+
{
|
| 19834 |
+
"epoch": 0.524615109824471,
|
| 19835 |
+
"grad_norm": 34.5,
|
| 19836 |
+
"learning_rate": 9.918028912858096e-06,
|
| 19837 |
+
"loss": 20.3451,
|
| 19838 |
+
"step": 28270
|
| 19839 |
+
},
|
| 19840 |
+
{
|
| 19841 |
+
"epoch": 0.5248006829089509,
|
| 19842 |
+
"grad_norm": 32.625,
|
| 19843 |
+
"learning_rate": 9.917999917072053e-06,
|
| 19844 |
+
"loss": 20.5492,
|
| 19845 |
+
"step": 28280
|
| 19846 |
+
},
|
| 19847 |
+
{
|
| 19848 |
+
"epoch": 0.5249862559934307,
|
| 19849 |
+
"grad_norm": 35.34375,
|
| 19850 |
+
"learning_rate": 9.91797092128601e-06,
|
| 19851 |
+
"loss": 20.4729,
|
| 19852 |
+
"step": 28290
|
| 19853 |
+
},
|
| 19854 |
+
{
|
| 19855 |
+
"epoch": 0.5251718290779105,
|
| 19856 |
+
"grad_norm": 34.84375,
|
| 19857 |
+
"learning_rate": 9.917941925499968e-06,
|
| 19858 |
+
"loss": 20.4557,
|
| 19859 |
+
"step": 28300
|
| 19860 |
+
},
|
| 19861 |
+
{
|
| 19862 |
+
"epoch": 0.5253574021623904,
|
| 19863 |
+
"grad_norm": 34.4375,
|
| 19864 |
+
"learning_rate": 9.917912929713925e-06,
|
| 19865 |
+
"loss": 20.4742,
|
| 19866 |
+
"step": 28310
|
| 19867 |
+
},
|
| 19868 |
+
{
|
| 19869 |
+
"epoch": 0.5255429752468702,
|
| 19870 |
+
"grad_norm": 36.59375,
|
| 19871 |
+
"learning_rate": 9.917883933927883e-06,
|
| 19872 |
+
"loss": 20.2819,
|
| 19873 |
+
"step": 28320
|
| 19874 |
+
},
|
| 19875 |
+
{
|
| 19876 |
+
"epoch": 0.5257285483313501,
|
| 19877 |
+
"grad_norm": 31.46875,
|
| 19878 |
+
"learning_rate": 9.917854938141842e-06,
|
| 19879 |
+
"loss": 20.7438,
|
| 19880 |
+
"step": 28330
|
| 19881 |
+
},
|
| 19882 |
+
{
|
| 19883 |
+
"epoch": 0.5259141214158298,
|
| 19884 |
+
"grad_norm": 34.6875,
|
| 19885 |
+
"learning_rate": 9.9178259423558e-06,
|
| 19886 |
+
"loss": 20.2566,
|
| 19887 |
+
"step": 28340
|
| 19888 |
+
},
|
| 19889 |
+
{
|
| 19890 |
+
"epoch": 0.5260996945003097,
|
| 19891 |
+
"grad_norm": 34.375,
|
| 19892 |
+
"learning_rate": 9.917796946569755e-06,
|
| 19893 |
+
"loss": 20.168,
|
| 19894 |
+
"step": 28350
|
| 19895 |
+
},
|
| 19896 |
+
{
|
| 19897 |
+
"epoch": 0.5262852675847896,
|
| 19898 |
+
"grad_norm": 35.46875,
|
| 19899 |
+
"learning_rate": 9.917767950783714e-06,
|
| 19900 |
+
"loss": 19.9336,
|
| 19901 |
+
"step": 28360
|
| 19902 |
+
},
|
| 19903 |
+
{
|
| 19904 |
+
"epoch": 0.5264708406692693,
|
| 19905 |
+
"grad_norm": 35.15625,
|
| 19906 |
+
"learning_rate": 9.917738954997671e-06,
|
| 19907 |
+
"loss": 20.2843,
|
| 19908 |
+
"step": 28370
|
| 19909 |
+
},
|
| 19910 |
+
{
|
| 19911 |
+
"epoch": 0.5266564137537492,
|
| 19912 |
+
"grad_norm": 35.53125,
|
| 19913 |
+
"learning_rate": 9.917709959211629e-06,
|
| 19914 |
+
"loss": 20.5998,
|
| 19915 |
+
"step": 28380
|
| 19916 |
+
},
|
| 19917 |
+
{
|
| 19918 |
+
"epoch": 0.526841986838229,
|
| 19919 |
+
"grad_norm": 34.71875,
|
| 19920 |
+
"learning_rate": 9.917680963425586e-06,
|
| 19921 |
+
"loss": 20.7026,
|
| 19922 |
+
"step": 28390
|
| 19923 |
+
},
|
| 19924 |
+
{
|
| 19925 |
+
"epoch": 0.5270275599227088,
|
| 19926 |
+
"grad_norm": 36.125,
|
| 19927 |
+
"learning_rate": 9.917651967639544e-06,
|
| 19928 |
+
"loss": 20.4475,
|
| 19929 |
+
"step": 28400
|
| 19930 |
+
},
|
| 19931 |
+
{
|
| 19932 |
+
"epoch": 0.5272131330071886,
|
| 19933 |
+
"grad_norm": 34.5,
|
| 19934 |
+
"learning_rate": 9.917622971853501e-06,
|
| 19935 |
+
"loss": 20.56,
|
| 19936 |
+
"step": 28410
|
| 19937 |
+
},
|
| 19938 |
+
{
|
| 19939 |
+
"epoch": 0.5273987060916685,
|
| 19940 |
+
"grad_norm": 35.09375,
|
| 19941 |
+
"learning_rate": 9.917593976067458e-06,
|
| 19942 |
+
"loss": 20.4937,
|
| 19943 |
+
"step": 28420
|
| 19944 |
+
},
|
| 19945 |
+
{
|
| 19946 |
+
"epoch": 0.5275842791761483,
|
| 19947 |
+
"grad_norm": 33.53125,
|
| 19948 |
+
"learning_rate": 9.917564980281416e-06,
|
| 19949 |
+
"loss": 20.5536,
|
| 19950 |
+
"step": 28430
|
| 19951 |
+
},
|
| 19952 |
+
{
|
| 19953 |
+
"epoch": 0.5277698522606281,
|
| 19954 |
+
"grad_norm": 34.25,
|
| 19955 |
+
"learning_rate": 9.917535984495375e-06,
|
| 19956 |
+
"loss": 19.9744,
|
| 19957 |
+
"step": 28440
|
| 19958 |
+
},
|
| 19959 |
+
{
|
| 19960 |
+
"epoch": 0.527955425345108,
|
| 19961 |
+
"grad_norm": 33.25,
|
| 19962 |
+
"learning_rate": 9.91750698870933e-06,
|
| 19963 |
+
"loss": 20.3831,
|
| 19964 |
+
"step": 28450
|
| 19965 |
+
},
|
| 19966 |
+
{
|
| 19967 |
+
"epoch": 0.5281409984295877,
|
| 19968 |
+
"grad_norm": 35.6875,
|
| 19969 |
+
"learning_rate": 9.91747799292329e-06,
|
| 19970 |
+
"loss": 20.5233,
|
| 19971 |
+
"step": 28460
|
| 19972 |
+
},
|
| 19973 |
+
{
|
| 19974 |
+
"epoch": 0.5283265715140676,
|
| 19975 |
+
"grad_norm": 35.4375,
|
| 19976 |
+
"learning_rate": 9.917448997137247e-06,
|
| 19977 |
+
"loss": 20.7939,
|
| 19978 |
+
"step": 28470
|
| 19979 |
+
},
|
| 19980 |
+
{
|
| 19981 |
+
"epoch": 0.5285121445985475,
|
| 19982 |
+
"grad_norm": 34.65625,
|
| 19983 |
+
"learning_rate": 9.917420001351205e-06,
|
| 19984 |
+
"loss": 20.3844,
|
| 19985 |
+
"step": 28480
|
| 19986 |
+
},
|
| 19987 |
+
{
|
| 19988 |
+
"epoch": 0.5286977176830272,
|
| 19989 |
+
"grad_norm": 34.90625,
|
| 19990 |
+
"learning_rate": 9.917391005565162e-06,
|
| 19991 |
+
"loss": 20.6473,
|
| 19992 |
+
"step": 28490
|
| 19993 |
+
},
|
| 19994 |
+
{
|
| 19995 |
+
"epoch": 0.5288832907675071,
|
| 19996 |
+
"grad_norm": 34.65625,
|
| 19997 |
+
"learning_rate": 9.91736200977912e-06,
|
| 19998 |
+
"loss": 20.1855,
|
| 19999 |
+
"step": 28500
|
| 20000 |
+
},
|
| 20001 |
+
{
|
| 20002 |
+
"epoch": 0.5290688638519869,
|
| 20003 |
+
"grad_norm": 35.6875,
|
| 20004 |
+
"learning_rate": 9.917333013993077e-06,
|
| 20005 |
+
"loss": 20.6092,
|
| 20006 |
+
"step": 28510
|
| 20007 |
+
},
|
| 20008 |
+
{
|
| 20009 |
+
"epoch": 0.5292544369364668,
|
| 20010 |
+
"grad_norm": 34.4375,
|
| 20011 |
+
"learning_rate": 9.917304018207034e-06,
|
| 20012 |
+
"loss": 20.351,
|
| 20013 |
+
"step": 28520
|
| 20014 |
+
},
|
| 20015 |
+
{
|
| 20016 |
+
"epoch": 0.5294400100209465,
|
| 20017 |
+
"grad_norm": 33.59375,
|
| 20018 |
+
"learning_rate": 9.917275022420992e-06,
|
| 20019 |
+
"loss": 19.9355,
|
| 20020 |
+
"step": 28530
|
| 20021 |
+
},
|
| 20022 |
+
{
|
| 20023 |
+
"epoch": 0.5296255831054264,
|
| 20024 |
+
"grad_norm": 35.84375,
|
| 20025 |
+
"learning_rate": 9.91724602663495e-06,
|
| 20026 |
+
"loss": 20.3667,
|
| 20027 |
+
"step": 28540
|
| 20028 |
+
},
|
| 20029 |
+
{
|
| 20030 |
+
"epoch": 0.5298111561899063,
|
| 20031 |
+
"grad_norm": 33.34375,
|
| 20032 |
+
"learning_rate": 9.917217030848908e-06,
|
| 20033 |
+
"loss": 20.1995,
|
| 20034 |
+
"step": 28550
|
| 20035 |
+
},
|
| 20036 |
+
{
|
| 20037 |
+
"epoch": 0.529996729274386,
|
| 20038 |
+
"grad_norm": 35.65625,
|
| 20039 |
+
"learning_rate": 9.917188035062864e-06,
|
| 20040 |
+
"loss": 20.2276,
|
| 20041 |
+
"step": 28560
|
| 20042 |
+
},
|
| 20043 |
+
{
|
| 20044 |
+
"epoch": 0.5301823023588659,
|
| 20045 |
+
"grad_norm": 35.75,
|
| 20046 |
+
"learning_rate": 9.917159039276823e-06,
|
| 20047 |
+
"loss": 20.5546,
|
| 20048 |
+
"step": 28570
|
| 20049 |
+
},
|
| 20050 |
+
{
|
| 20051 |
+
"epoch": 0.5303678754433457,
|
| 20052 |
+
"grad_norm": 36.9375,
|
| 20053 |
+
"learning_rate": 9.91713004349078e-06,
|
| 20054 |
+
"loss": 20.7107,
|
| 20055 |
+
"step": 28580
|
| 20056 |
+
},
|
| 20057 |
+
{
|
| 20058 |
+
"epoch": 0.5305534485278255,
|
| 20059 |
+
"grad_norm": 35.875,
|
| 20060 |
+
"learning_rate": 9.917101047704738e-06,
|
| 20061 |
+
"loss": 20.6607,
|
| 20062 |
+
"step": 28590
|
| 20063 |
+
},
|
| 20064 |
+
{
|
| 20065 |
+
"epoch": 0.5307390216123054,
|
| 20066 |
+
"grad_norm": 33.90625,
|
| 20067 |
+
"learning_rate": 9.917072051918695e-06,
|
| 20068 |
+
"loss": 20.3141,
|
| 20069 |
+
"step": 28600
|
| 20070 |
+
},
|
| 20071 |
+
{
|
| 20072 |
+
"epoch": 0.5309245946967852,
|
| 20073 |
+
"grad_norm": 37.34375,
|
| 20074 |
+
"learning_rate": 9.917043056132653e-06,
|
| 20075 |
+
"loss": 20.4263,
|
| 20076 |
+
"step": 28610
|
| 20077 |
+
},
|
| 20078 |
+
{
|
| 20079 |
+
"epoch": 0.531110167781265,
|
| 20080 |
+
"grad_norm": 34.15625,
|
| 20081 |
+
"learning_rate": 9.91701406034661e-06,
|
| 20082 |
+
"loss": 20.6688,
|
| 20083 |
+
"step": 28620
|
| 20084 |
+
},
|
| 20085 |
+
{
|
| 20086 |
+
"epoch": 0.5312957408657448,
|
| 20087 |
+
"grad_norm": 33.65625,
|
| 20088 |
+
"learning_rate": 9.916985064560567e-06,
|
| 20089 |
+
"loss": 20.6133,
|
| 20090 |
+
"step": 28630
|
| 20091 |
+
},
|
| 20092 |
+
{
|
| 20093 |
+
"epoch": 0.5314813139502247,
|
| 20094 |
+
"grad_norm": 33.625,
|
| 20095 |
+
"learning_rate": 9.916956068774526e-06,
|
| 20096 |
+
"loss": 20.1227,
|
| 20097 |
+
"step": 28640
|
| 20098 |
+
},
|
| 20099 |
+
{
|
| 20100 |
+
"epoch": 0.5316668870347044,
|
| 20101 |
+
"grad_norm": 36.28125,
|
| 20102 |
+
"learning_rate": 9.916927072988484e-06,
|
| 20103 |
+
"loss": 20.2326,
|
| 20104 |
+
"step": 28650
|
| 20105 |
+
},
|
| 20106 |
+
{
|
| 20107 |
+
"epoch": 0.5318524601191843,
|
| 20108 |
+
"grad_norm": 35.8125,
|
| 20109 |
+
"learning_rate": 9.916898077202441e-06,
|
| 20110 |
+
"loss": 20.3055,
|
| 20111 |
+
"step": 28660
|
| 20112 |
+
},
|
| 20113 |
+
{
|
| 20114 |
+
"epoch": 0.5320380332036642,
|
| 20115 |
+
"grad_norm": 34.21875,
|
| 20116 |
+
"learning_rate": 9.916869081416399e-06,
|
| 20117 |
+
"loss": 20.5771,
|
| 20118 |
+
"step": 28670
|
| 20119 |
+
},
|
| 20120 |
+
{
|
| 20121 |
+
"epoch": 0.5322236062881439,
|
| 20122 |
+
"grad_norm": 33.96875,
|
| 20123 |
+
"learning_rate": 9.916840085630356e-06,
|
| 20124 |
+
"loss": 20.036,
|
| 20125 |
+
"step": 28680
|
| 20126 |
+
},
|
| 20127 |
+
{
|
| 20128 |
+
"epoch": 0.5324091793726238,
|
| 20129 |
+
"grad_norm": 34.0,
|
| 20130 |
+
"learning_rate": 9.916811089844314e-06,
|
| 20131 |
+
"loss": 20.4672,
|
| 20132 |
+
"step": 28690
|
| 20133 |
+
},
|
| 20134 |
+
{
|
| 20135 |
+
"epoch": 0.5325947524571036,
|
| 20136 |
+
"grad_norm": 35.90625,
|
| 20137 |
+
"learning_rate": 9.916782094058271e-06,
|
| 20138 |
+
"loss": 20.2936,
|
| 20139 |
+
"step": 28700
|
| 20140 |
+
},
|
| 20141 |
+
{
|
| 20142 |
+
"epoch": 0.5327803255415835,
|
| 20143 |
+
"grad_norm": 34.4375,
|
| 20144 |
+
"learning_rate": 9.91675309827223e-06,
|
| 20145 |
+
"loss": 20.5766,
|
| 20146 |
+
"step": 28710
|
| 20147 |
+
},
|
| 20148 |
+
{
|
| 20149 |
+
"epoch": 0.5329658986260633,
|
| 20150 |
+
"grad_norm": 32.40625,
|
| 20151 |
+
"learning_rate": 9.916724102486186e-06,
|
| 20152 |
+
"loss": 20.6091,
|
| 20153 |
+
"step": 28720
|
| 20154 |
+
},
|
| 20155 |
+
{
|
| 20156 |
+
"epoch": 0.5331514717105431,
|
| 20157 |
+
"grad_norm": 34.5,
|
| 20158 |
+
"learning_rate": 9.916695106700143e-06,
|
| 20159 |
+
"loss": 20.1864,
|
| 20160 |
+
"step": 28730
|
| 20161 |
+
},
|
| 20162 |
+
{
|
| 20163 |
+
"epoch": 0.533337044795023,
|
| 20164 |
+
"grad_norm": 37.6875,
|
| 20165 |
+
"learning_rate": 9.916666110914102e-06,
|
| 20166 |
+
"loss": 20.4298,
|
| 20167 |
+
"step": 28740
|
| 20168 |
+
},
|
| 20169 |
+
{
|
| 20170 |
+
"epoch": 0.5335226178795027,
|
| 20171 |
+
"grad_norm": 34.9375,
|
| 20172 |
+
"learning_rate": 9.91663711512806e-06,
|
| 20173 |
+
"loss": 20.1732,
|
| 20174 |
+
"step": 28750
|
| 20175 |
+
},
|
| 20176 |
+
{
|
| 20177 |
+
"epoch": 0.5337081909639826,
|
| 20178 |
+
"grad_norm": 35.5,
|
| 20179 |
+
"learning_rate": 9.916608119342017e-06,
|
| 20180 |
+
"loss": 20.4916,
|
| 20181 |
+
"step": 28760
|
| 20182 |
+
},
|
| 20183 |
+
{
|
| 20184 |
+
"epoch": 0.5338937640484624,
|
| 20185 |
+
"grad_norm": 35.125,
|
| 20186 |
+
"learning_rate": 9.916579123555974e-06,
|
| 20187 |
+
"loss": 20.4197,
|
| 20188 |
+
"step": 28770
|
| 20189 |
+
},
|
| 20190 |
+
{
|
| 20191 |
+
"epoch": 0.5340793371329422,
|
| 20192 |
+
"grad_norm": 37.21875,
|
| 20193 |
+
"learning_rate": 9.916550127769932e-06,
|
| 20194 |
+
"loss": 20.1008,
|
| 20195 |
+
"step": 28780
|
| 20196 |
+
},
|
| 20197 |
+
{
|
| 20198 |
+
"epoch": 0.5342649102174221,
|
| 20199 |
+
"grad_norm": 35.96875,
|
| 20200 |
+
"learning_rate": 9.91652113198389e-06,
|
| 20201 |
+
"loss": 20.4591,
|
| 20202 |
+
"step": 28790
|
| 20203 |
+
},
|
| 20204 |
+
{
|
| 20205 |
+
"epoch": 0.5344504833019019,
|
| 20206 |
+
"grad_norm": 35.03125,
|
| 20207 |
+
"learning_rate": 9.916492136197847e-06,
|
| 20208 |
+
"loss": 20.3487,
|
| 20209 |
+
"step": 28800
|
| 20210 |
+
},
|
| 20211 |
+
{
|
| 20212 |
+
"epoch": 0.5346360563863817,
|
| 20213 |
+
"grad_norm": 33.84375,
|
| 20214 |
+
"learning_rate": 9.916463140411806e-06,
|
| 20215 |
+
"loss": 20.0548,
|
| 20216 |
+
"step": 28810
|
| 20217 |
+
},
|
| 20218 |
+
{
|
| 20219 |
+
"epoch": 0.5348216294708615,
|
| 20220 |
+
"grad_norm": 34.375,
|
| 20221 |
+
"learning_rate": 9.916434144625763e-06,
|
| 20222 |
+
"loss": 20.0794,
|
| 20223 |
+
"step": 28820
|
| 20224 |
+
},
|
| 20225 |
+
{
|
| 20226 |
+
"epoch": 0.5350072025553414,
|
| 20227 |
+
"grad_norm": 36.84375,
|
| 20228 |
+
"learning_rate": 9.916405148839719e-06,
|
| 20229 |
+
"loss": 20.7802,
|
| 20230 |
+
"step": 28830
|
| 20231 |
+
},
|
| 20232 |
+
{
|
| 20233 |
+
"epoch": 0.5351927756398211,
|
| 20234 |
+
"grad_norm": 35.03125,
|
| 20235 |
+
"learning_rate": 9.916376153053678e-06,
|
| 20236 |
+
"loss": 20.5046,
|
| 20237 |
+
"step": 28840
|
| 20238 |
+
},
|
| 20239 |
+
{
|
| 20240 |
+
"epoch": 0.535378348724301,
|
| 20241 |
+
"grad_norm": 36.15625,
|
| 20242 |
+
"learning_rate": 9.916347157267635e-06,
|
| 20243 |
+
"loss": 20.5881,
|
| 20244 |
+
"step": 28850
|
| 20245 |
+
},
|
| 20246 |
+
{
|
| 20247 |
+
"epoch": 0.5355639218087809,
|
| 20248 |
+
"grad_norm": 36.375,
|
| 20249 |
+
"learning_rate": 9.916318161481593e-06,
|
| 20250 |
+
"loss": 20.5849,
|
| 20251 |
+
"step": 28860
|
| 20252 |
+
},
|
| 20253 |
+
{
|
| 20254 |
+
"epoch": 0.5357494948932607,
|
| 20255 |
+
"grad_norm": 35.71875,
|
| 20256 |
+
"learning_rate": 9.91628916569555e-06,
|
| 20257 |
+
"loss": 20.5164,
|
| 20258 |
+
"step": 28870
|
| 20259 |
+
},
|
| 20260 |
+
{
|
| 20261 |
+
"epoch": 0.5359350679777405,
|
| 20262 |
+
"grad_norm": 36.28125,
|
| 20263 |
+
"learning_rate": 9.916260169909508e-06,
|
| 20264 |
+
"loss": 20.0847,
|
| 20265 |
+
"step": 28880
|
| 20266 |
+
},
|
| 20267 |
+
{
|
| 20268 |
+
"epoch": 0.5361206410622203,
|
| 20269 |
+
"grad_norm": 34.21875,
|
| 20270 |
+
"learning_rate": 9.916231174123465e-06,
|
| 20271 |
+
"loss": 19.9122,
|
| 20272 |
+
"step": 28890
|
| 20273 |
+
},
|
| 20274 |
+
{
|
| 20275 |
+
"epoch": 0.5363062141467002,
|
| 20276 |
+
"grad_norm": 34.0,
|
| 20277 |
+
"learning_rate": 9.916202178337422e-06,
|
| 20278 |
+
"loss": 20.5583,
|
| 20279 |
+
"step": 28900
|
| 20280 |
+
},
|
| 20281 |
+
{
|
| 20282 |
+
"epoch": 0.53649178723118,
|
| 20283 |
+
"grad_norm": 34.1875,
|
| 20284 |
+
"learning_rate": 9.916173182551382e-06,
|
| 20285 |
+
"loss": 20.2385,
|
| 20286 |
+
"step": 28910
|
| 20287 |
+
},
|
| 20288 |
+
{
|
| 20289 |
+
"epoch": 0.5366773603156598,
|
| 20290 |
+
"grad_norm": 33.5,
|
| 20291 |
+
"learning_rate": 9.916144186765339e-06,
|
| 20292 |
+
"loss": 20.233,
|
| 20293 |
+
"step": 28920
|
| 20294 |
+
},
|
| 20295 |
+
{
|
| 20296 |
+
"epoch": 0.5368629334001397,
|
| 20297 |
+
"grad_norm": 35.6875,
|
| 20298 |
+
"learning_rate": 9.916115190979296e-06,
|
| 20299 |
+
"loss": 20.5665,
|
| 20300 |
+
"step": 28930
|
| 20301 |
+
},
|
| 20302 |
+
{
|
| 20303 |
+
"epoch": 0.5370485064846194,
|
| 20304 |
+
"grad_norm": 34.53125,
|
| 20305 |
+
"learning_rate": 9.916086195193254e-06,
|
| 20306 |
+
"loss": 20.1071,
|
| 20307 |
+
"step": 28940
|
| 20308 |
+
},
|
| 20309 |
+
{
|
| 20310 |
+
"epoch": 0.5372340795690993,
|
| 20311 |
+
"grad_norm": 35.34375,
|
| 20312 |
+
"learning_rate": 9.916057199407211e-06,
|
| 20313 |
+
"loss": 20.5049,
|
| 20314 |
+
"step": 28950
|
| 20315 |
+
},
|
| 20316 |
+
{
|
| 20317 |
+
"epoch": 0.5374196526535792,
|
| 20318 |
+
"grad_norm": 34.96875,
|
| 20319 |
+
"learning_rate": 9.916028203621169e-06,
|
| 20320 |
+
"loss": 20.1108,
|
| 20321 |
+
"step": 28960
|
| 20322 |
+
},
|
| 20323 |
+
{
|
| 20324 |
+
"epoch": 0.5376052257380589,
|
| 20325 |
+
"grad_norm": 37.625,
|
| 20326 |
+
"learning_rate": 9.915999207835126e-06,
|
| 20327 |
+
"loss": 20.7692,
|
| 20328 |
+
"step": 28970
|
| 20329 |
+
},
|
| 20330 |
+
{
|
| 20331 |
+
"epoch": 0.5377907988225388,
|
| 20332 |
+
"grad_norm": 37.0625,
|
| 20333 |
+
"learning_rate": 9.915970212049083e-06,
|
| 20334 |
+
"loss": 20.4374,
|
| 20335 |
+
"step": 28980
|
| 20336 |
+
},
|
| 20337 |
+
{
|
| 20338 |
+
"epoch": 0.5379763719070186,
|
| 20339 |
+
"grad_norm": 34.28125,
|
| 20340 |
+
"learning_rate": 9.91594121626304e-06,
|
| 20341 |
+
"loss": 20.5759,
|
| 20342 |
+
"step": 28990
|
| 20343 |
+
},
|
| 20344 |
+
{
|
| 20345 |
+
"epoch": 0.5381619449914984,
|
| 20346 |
+
"grad_norm": 34.6875,
|
| 20347 |
+
"learning_rate": 9.915912220476998e-06,
|
| 20348 |
+
"loss": 20.2427,
|
| 20349 |
+
"step": 29000
|
| 20350 |
+
},
|
| 20351 |
+
{
|
| 20352 |
+
"epoch": 0.5383475180759782,
|
| 20353 |
+
"grad_norm": 36.25,
|
| 20354 |
+
"learning_rate": 9.915883224690956e-06,
|
| 20355 |
+
"loss": 20.5409,
|
| 20356 |
+
"step": 29010
|
| 20357 |
+
},
|
| 20358 |
+
{
|
| 20359 |
+
"epoch": 0.5385330911604581,
|
| 20360 |
+
"grad_norm": 35.25,
|
| 20361 |
+
"learning_rate": 9.915854228904915e-06,
|
| 20362 |
+
"loss": 20.6585,
|
| 20363 |
+
"step": 29020
|
| 20364 |
+
},
|
| 20365 |
+
{
|
| 20366 |
+
"epoch": 0.5387186642449379,
|
| 20367 |
+
"grad_norm": 35.0,
|
| 20368 |
+
"learning_rate": 9.915825233118872e-06,
|
| 20369 |
+
"loss": 20.4239,
|
| 20370 |
+
"step": 29030
|
| 20371 |
+
},
|
| 20372 |
+
{
|
| 20373 |
+
"epoch": 0.5389042373294177,
|
| 20374 |
+
"grad_norm": 35.5,
|
| 20375 |
+
"learning_rate": 9.91579623733283e-06,
|
| 20376 |
+
"loss": 20.3294,
|
| 20377 |
+
"step": 29040
|
| 20378 |
+
},
|
| 20379 |
+
{
|
| 20380 |
+
"epoch": 0.5390898104138976,
|
| 20381 |
+
"grad_norm": 32.21875,
|
| 20382 |
+
"learning_rate": 9.915767241546787e-06,
|
| 20383 |
+
"loss": 20.3876,
|
| 20384 |
+
"step": 29050
|
| 20385 |
+
},
|
| 20386 |
+
{
|
| 20387 |
+
"epoch": 0.5392753834983774,
|
| 20388 |
+
"grad_norm": 34.3125,
|
| 20389 |
+
"learning_rate": 9.915738245760744e-06,
|
| 20390 |
+
"loss": 20.8428,
|
| 20391 |
+
"step": 29060
|
| 20392 |
+
},
|
| 20393 |
+
{
|
| 20394 |
+
"epoch": 0.5394609565828572,
|
| 20395 |
+
"grad_norm": 35.3125,
|
| 20396 |
+
"learning_rate": 9.915709249974702e-06,
|
| 20397 |
+
"loss": 20.0861,
|
| 20398 |
+
"step": 29070
|
| 20399 |
+
},
|
| 20400 |
+
{
|
| 20401 |
+
"epoch": 0.5396465296673371,
|
| 20402 |
+
"grad_norm": 35.9375,
|
| 20403 |
+
"learning_rate": 9.91568025418866e-06,
|
| 20404 |
+
"loss": 20.6351,
|
| 20405 |
+
"step": 29080
|
| 20406 |
+
},
|
| 20407 |
+
{
|
| 20408 |
+
"epoch": 0.5398321027518169,
|
| 20409 |
+
"grad_norm": 34.75,
|
| 20410 |
+
"learning_rate": 9.915651258402618e-06,
|
| 20411 |
+
"loss": 20.3223,
|
| 20412 |
+
"step": 29090
|
| 20413 |
+
},
|
| 20414 |
+
{
|
| 20415 |
+
"epoch": 0.5400176758362967,
|
| 20416 |
+
"grad_norm": 34.90625,
|
| 20417 |
+
"learning_rate": 9.915622262616574e-06,
|
| 20418 |
+
"loss": 20.0895,
|
| 20419 |
+
"step": 29100
|
| 20420 |
+
},
|
| 20421 |
+
{
|
| 20422 |
+
"epoch": 0.5402032489207765,
|
| 20423 |
+
"grad_norm": 33.9375,
|
| 20424 |
+
"learning_rate": 9.915593266830531e-06,
|
| 20425 |
+
"loss": 20.2776,
|
| 20426 |
+
"step": 29110
|
| 20427 |
+
},
|
| 20428 |
+
{
|
| 20429 |
+
"epoch": 0.5403888220052564,
|
| 20430 |
+
"grad_norm": 35.75,
|
| 20431 |
+
"learning_rate": 9.91556427104449e-06,
|
| 20432 |
+
"loss": 20.3674,
|
| 20433 |
+
"step": 29120
|
| 20434 |
+
},
|
| 20435 |
+
{
|
| 20436 |
+
"epoch": 0.5405743950897361,
|
| 20437 |
+
"grad_norm": 33.75,
|
| 20438 |
+
"learning_rate": 9.915535275258448e-06,
|
| 20439 |
+
"loss": 20.5777,
|
| 20440 |
+
"step": 29130
|
| 20441 |
+
},
|
| 20442 |
+
{
|
| 20443 |
+
"epoch": 0.540759968174216,
|
| 20444 |
+
"grad_norm": 34.84375,
|
| 20445 |
+
"learning_rate": 9.915506279472405e-06,
|
| 20446 |
+
"loss": 20.3793,
|
| 20447 |
+
"step": 29140
|
| 20448 |
+
},
|
| 20449 |
+
{
|
| 20450 |
+
"epoch": 0.5409455412586959,
|
| 20451 |
+
"grad_norm": 35.78125,
|
| 20452 |
+
"learning_rate": 9.915477283686363e-06,
|
| 20453 |
+
"loss": 20.5047,
|
| 20454 |
+
"step": 29150
|
| 20455 |
+
},
|
| 20456 |
+
{
|
| 20457 |
+
"epoch": 0.5411311143431756,
|
| 20458 |
+
"grad_norm": 35.46875,
|
| 20459 |
+
"learning_rate": 9.91544828790032e-06,
|
| 20460 |
+
"loss": 20.7663,
|
| 20461 |
+
"step": 29160
|
| 20462 |
+
},
|
| 20463 |
+
{
|
| 20464 |
+
"epoch": 0.5413166874276555,
|
| 20465 |
+
"grad_norm": 35.84375,
|
| 20466 |
+
"learning_rate": 9.915419292114278e-06,
|
| 20467 |
+
"loss": 20.3021,
|
| 20468 |
+
"step": 29170
|
| 20469 |
+
},
|
| 20470 |
+
{
|
| 20471 |
+
"epoch": 0.5415022605121353,
|
| 20472 |
+
"grad_norm": 35.21875,
|
| 20473 |
+
"learning_rate": 9.915390296328235e-06,
|
| 20474 |
+
"loss": 20.4529,
|
| 20475 |
+
"step": 29180
|
| 20476 |
+
},
|
| 20477 |
+
{
|
| 20478 |
+
"epoch": 0.5416878335966151,
|
| 20479 |
+
"grad_norm": 34.8125,
|
| 20480 |
+
"learning_rate": 9.915361300542194e-06,
|
| 20481 |
+
"loss": 20.3839,
|
| 20482 |
+
"step": 29190
|
| 20483 |
+
},
|
| 20484 |
+
{
|
| 20485 |
+
"epoch": 0.541873406681095,
|
| 20486 |
+
"grad_norm": 33.0625,
|
| 20487 |
+
"learning_rate": 9.91533230475615e-06,
|
| 20488 |
+
"loss": 20.3712,
|
| 20489 |
+
"step": 29200
|
| 20490 |
+
},
|
| 20491 |
+
{
|
| 20492 |
+
"epoch": 0.5420589797655748,
|
| 20493 |
+
"grad_norm": 34.1875,
|
| 20494 |
+
"learning_rate": 9.915303308970107e-06,
|
| 20495 |
+
"loss": 20.1465,
|
| 20496 |
+
"step": 29210
|
| 20497 |
+
},
|
| 20498 |
+
{
|
| 20499 |
+
"epoch": 0.5422445528500546,
|
| 20500 |
+
"grad_norm": 36.90625,
|
| 20501 |
+
"learning_rate": 9.915274313184066e-06,
|
| 20502 |
+
"loss": 20.0052,
|
| 20503 |
+
"step": 29220
|
| 20504 |
+
},
|
| 20505 |
+
{
|
| 20506 |
+
"epoch": 0.5424301259345344,
|
| 20507 |
+
"grad_norm": 36.6875,
|
| 20508 |
+
"learning_rate": 9.915245317398024e-06,
|
| 20509 |
+
"loss": 20.3256,
|
| 20510 |
+
"step": 29230
|
| 20511 |
+
},
|
| 20512 |
+
{
|
| 20513 |
+
"epoch": 0.5426156990190143,
|
| 20514 |
+
"grad_norm": 36.125,
|
| 20515 |
+
"learning_rate": 9.915216321611981e-06,
|
| 20516 |
+
"loss": 20.0692,
|
| 20517 |
+
"step": 29240
|
| 20518 |
+
},
|
| 20519 |
+
{
|
| 20520 |
+
"epoch": 0.5428012721034942,
|
| 20521 |
+
"grad_norm": 34.46875,
|
| 20522 |
+
"learning_rate": 9.915187325825938e-06,
|
| 20523 |
+
"loss": 20.5439,
|
| 20524 |
+
"step": 29250
|
| 20525 |
+
},
|
| 20526 |
+
{
|
| 20527 |
+
"epoch": 0.5429868451879739,
|
| 20528 |
+
"grad_norm": 34.78125,
|
| 20529 |
+
"learning_rate": 9.915158330039896e-06,
|
| 20530 |
+
"loss": 20.3697,
|
| 20531 |
+
"step": 29260
|
| 20532 |
+
},
|
| 20533 |
+
{
|
| 20534 |
+
"epoch": 0.5431724182724538,
|
| 20535 |
+
"grad_norm": 34.0625,
|
| 20536 |
+
"learning_rate": 9.915129334253853e-06,
|
| 20537 |
+
"loss": 20.0717,
|
| 20538 |
+
"step": 29270
|
| 20539 |
+
},
|
| 20540 |
+
{
|
| 20541 |
+
"epoch": 0.5433579913569336,
|
| 20542 |
+
"grad_norm": 33.84375,
|
| 20543 |
+
"learning_rate": 9.91510033846781e-06,
|
| 20544 |
+
"loss": 20.1454,
|
| 20545 |
+
"step": 29280
|
| 20546 |
+
},
|
| 20547 |
+
{
|
| 20548 |
+
"epoch": 0.5435435644414134,
|
| 20549 |
+
"grad_norm": 33.78125,
|
| 20550 |
+
"learning_rate": 9.91507134268177e-06,
|
| 20551 |
+
"loss": 20.4484,
|
| 20552 |
+
"step": 29290
|
| 20553 |
+
},
|
| 20554 |
+
{
|
| 20555 |
+
"epoch": 0.5437291375258932,
|
| 20556 |
+
"grad_norm": 34.0625,
|
| 20557 |
+
"learning_rate": 9.915042346895727e-06,
|
| 20558 |
+
"loss": 20.3169,
|
| 20559 |
+
"step": 29300
|
| 20560 |
+
},
|
| 20561 |
+
{
|
| 20562 |
+
"epoch": 0.5439147106103731,
|
| 20563 |
+
"grad_norm": 33.25,
|
| 20564 |
+
"learning_rate": 9.915013351109683e-06,
|
| 20565 |
+
"loss": 20.0393,
|
| 20566 |
+
"step": 29310
|
| 20567 |
+
},
|
| 20568 |
+
{
|
| 20569 |
+
"epoch": 0.5441002836948529,
|
| 20570 |
+
"grad_norm": 35.625,
|
| 20571 |
+
"learning_rate": 9.914984355323642e-06,
|
| 20572 |
+
"loss": 20.2945,
|
| 20573 |
+
"step": 29320
|
| 20574 |
+
},
|
| 20575 |
+
{
|
| 20576 |
+
"epoch": 0.5442858567793327,
|
| 20577 |
+
"grad_norm": 33.96875,
|
| 20578 |
+
"learning_rate": 9.9149553595376e-06,
|
| 20579 |
+
"loss": 20.3528,
|
| 20580 |
+
"step": 29330
|
| 20581 |
+
},
|
| 20582 |
+
{
|
| 20583 |
+
"epoch": 0.5444714298638126,
|
| 20584 |
+
"grad_norm": 35.03125,
|
| 20585 |
+
"learning_rate": 9.914926363751557e-06,
|
| 20586 |
+
"loss": 20.6093,
|
| 20587 |
+
"step": 29340
|
| 20588 |
+
},
|
| 20589 |
+
{
|
| 20590 |
+
"epoch": 0.5446570029482923,
|
| 20591 |
+
"grad_norm": 34.5,
|
| 20592 |
+
"learning_rate": 9.914897367965514e-06,
|
| 20593 |
+
"loss": 20.2775,
|
| 20594 |
+
"step": 29350
|
| 20595 |
+
},
|
| 20596 |
+
{
|
| 20597 |
+
"epoch": 0.5448425760327722,
|
| 20598 |
+
"grad_norm": 35.5,
|
| 20599 |
+
"learning_rate": 9.914868372179473e-06,
|
| 20600 |
+
"loss": 21.0042,
|
| 20601 |
+
"step": 29360
|
| 20602 |
+
},
|
| 20603 |
+
{
|
| 20604 |
+
"epoch": 0.5450281491172521,
|
| 20605 |
+
"grad_norm": 35.65625,
|
| 20606 |
+
"learning_rate": 9.914839376393429e-06,
|
| 20607 |
+
"loss": 20.6964,
|
| 20608 |
+
"step": 29370
|
| 20609 |
+
},
|
| 20610 |
+
{
|
| 20611 |
+
"epoch": 0.5452137222017318,
|
| 20612 |
+
"grad_norm": 35.8125,
|
| 20613 |
+
"learning_rate": 9.914810380607386e-06,
|
| 20614 |
+
"loss": 20.6275,
|
| 20615 |
+
"step": 29380
|
| 20616 |
+
},
|
| 20617 |
+
{
|
| 20618 |
+
"epoch": 0.5453992952862117,
|
| 20619 |
+
"grad_norm": 35.09375,
|
| 20620 |
+
"learning_rate": 9.914781384821346e-06,
|
| 20621 |
+
"loss": 19.9444,
|
| 20622 |
+
"step": 29390
|
| 20623 |
+
},
|
| 20624 |
+
{
|
| 20625 |
+
"epoch": 0.5455848683706915,
|
| 20626 |
+
"grad_norm": 35.21875,
|
| 20627 |
+
"learning_rate": 9.914752389035303e-06,
|
| 20628 |
+
"loss": 20.543,
|
| 20629 |
+
"step": 29400
|
| 20630 |
+
},
|
| 20631 |
+
{
|
| 20632 |
+
"epoch": 0.5457704414551714,
|
| 20633 |
+
"grad_norm": 37.21875,
|
| 20634 |
+
"learning_rate": 9.91472339324926e-06,
|
| 20635 |
+
"loss": 20.303,
|
| 20636 |
+
"step": 29410
|
| 20637 |
+
},
|
| 20638 |
+
{
|
| 20639 |
+
"epoch": 0.5459560145396511,
|
| 20640 |
+
"grad_norm": 34.9375,
|
| 20641 |
+
"learning_rate": 9.914694397463218e-06,
|
| 20642 |
+
"loss": 20.1036,
|
| 20643 |
+
"step": 29420
|
| 20644 |
+
},
|
| 20645 |
+
{
|
| 20646 |
+
"epoch": 0.546141587624131,
|
| 20647 |
+
"grad_norm": 33.78125,
|
| 20648 |
+
"learning_rate": 9.914665401677175e-06,
|
| 20649 |
+
"loss": 20.4787,
|
| 20650 |
+
"step": 29430
|
| 20651 |
+
},
|
| 20652 |
+
{
|
| 20653 |
+
"epoch": 0.5463271607086109,
|
| 20654 |
+
"grad_norm": 34.5,
|
| 20655 |
+
"learning_rate": 9.914636405891133e-06,
|
| 20656 |
+
"loss": 20.1483,
|
| 20657 |
+
"step": 29440
|
| 20658 |
+
},
|
| 20659 |
+
{
|
| 20660 |
+
"epoch": 0.5465127337930906,
|
| 20661 |
+
"grad_norm": 34.53125,
|
| 20662 |
+
"learning_rate": 9.91460741010509e-06,
|
| 20663 |
+
"loss": 20.1377,
|
| 20664 |
+
"step": 29450
|
| 20665 |
+
},
|
| 20666 |
+
{
|
| 20667 |
+
"epoch": 0.5466983068775705,
|
| 20668 |
+
"grad_norm": 32.65625,
|
| 20669 |
+
"learning_rate": 9.914578414319047e-06,
|
| 20670 |
+
"loss": 20.1915,
|
| 20671 |
+
"step": 29460
|
| 20672 |
+
},
|
| 20673 |
+
{
|
| 20674 |
+
"epoch": 0.5468838799620503,
|
| 20675 |
+
"grad_norm": 36.3125,
|
| 20676 |
+
"learning_rate": 9.914549418533005e-06,
|
| 20677 |
+
"loss": 20.0376,
|
| 20678 |
+
"step": 29470
|
| 20679 |
+
},
|
| 20680 |
+
{
|
| 20681 |
+
"epoch": 0.5470694530465301,
|
| 20682 |
+
"grad_norm": 36.34375,
|
| 20683 |
+
"learning_rate": 9.914520422746962e-06,
|
| 20684 |
+
"loss": 20.5274,
|
| 20685 |
+
"step": 29480
|
| 20686 |
+
},
|
| 20687 |
+
{
|
| 20688 |
+
"epoch": 0.54725502613101,
|
| 20689 |
+
"grad_norm": 37.5625,
|
| 20690 |
+
"learning_rate": 9.914491426960921e-06,
|
| 20691 |
+
"loss": 20.618,
|
| 20692 |
+
"step": 29490
|
| 20693 |
+
},
|
| 20694 |
+
{
|
| 20695 |
+
"epoch": 0.5474405992154898,
|
| 20696 |
+
"grad_norm": 34.78125,
|
| 20697 |
+
"learning_rate": 9.914462431174879e-06,
|
| 20698 |
+
"loss": 20.2392,
|
| 20699 |
+
"step": 29500
|
| 20700 |
+
},
|
| 20701 |
+
{
|
| 20702 |
+
"epoch": 0.5476261722999696,
|
| 20703 |
+
"grad_norm": 35.5,
|
| 20704 |
+
"learning_rate": 9.914433435388836e-06,
|
| 20705 |
+
"loss": 20.0363,
|
| 20706 |
+
"step": 29510
|
| 20707 |
+
},
|
| 20708 |
+
{
|
| 20709 |
+
"epoch": 0.5478117453844494,
|
| 20710 |
+
"grad_norm": 35.6875,
|
| 20711 |
+
"learning_rate": 9.914404439602794e-06,
|
| 20712 |
+
"loss": 19.957,
|
| 20713 |
+
"step": 29520
|
| 20714 |
+
},
|
| 20715 |
+
{
|
| 20716 |
+
"epoch": 0.5479973184689293,
|
| 20717 |
+
"grad_norm": 33.3125,
|
| 20718 |
+
"learning_rate": 9.914375443816751e-06,
|
| 20719 |
+
"loss": 20.0681,
|
| 20720 |
+
"step": 29530
|
| 20721 |
+
},
|
| 20722 |
+
{
|
| 20723 |
+
"epoch": 0.548182891553409,
|
| 20724 |
+
"grad_norm": 32.8125,
|
| 20725 |
+
"learning_rate": 9.914346448030708e-06,
|
| 20726 |
+
"loss": 20.4819,
|
| 20727 |
+
"step": 29540
|
| 20728 |
+
},
|
| 20729 |
+
{
|
| 20730 |
+
"epoch": 0.5483684646378889,
|
| 20731 |
+
"grad_norm": 35.1875,
|
| 20732 |
+
"learning_rate": 9.914317452244666e-06,
|
| 20733 |
+
"loss": 20.1795,
|
| 20734 |
+
"step": 29550
|
| 20735 |
+
},
|
| 20736 |
+
{
|
| 20737 |
+
"epoch": 0.5485540377223688,
|
| 20738 |
+
"grad_norm": 34.25,
|
| 20739 |
+
"learning_rate": 9.914288456458623e-06,
|
| 20740 |
+
"loss": 20.3919,
|
| 20741 |
+
"step": 29560
|
| 20742 |
+
},
|
| 20743 |
+
{
|
| 20744 |
+
"epoch": 0.5487396108068485,
|
| 20745 |
+
"grad_norm": 35.03125,
|
| 20746 |
+
"learning_rate": 9.914259460672582e-06,
|
| 20747 |
+
"loss": 20.1524,
|
| 20748 |
+
"step": 29570
|
| 20749 |
+
},
|
| 20750 |
+
{
|
| 20751 |
+
"epoch": 0.5489251838913284,
|
| 20752 |
+
"grad_norm": 33.4375,
|
| 20753 |
+
"learning_rate": 9.914230464886538e-06,
|
| 20754 |
+
"loss": 20.2308,
|
| 20755 |
+
"step": 29580
|
| 20756 |
+
},
|
| 20757 |
+
{
|
| 20758 |
+
"epoch": 0.5491107569758082,
|
| 20759 |
+
"grad_norm": 35.4375,
|
| 20760 |
+
"learning_rate": 9.914201469100495e-06,
|
| 20761 |
+
"loss": 19.9178,
|
| 20762 |
+
"step": 29590
|
| 20763 |
+
},
|
| 20764 |
+
{
|
| 20765 |
+
"epoch": 0.5492963300602881,
|
| 20766 |
+
"grad_norm": 34.8125,
|
| 20767 |
+
"learning_rate": 9.914172473314454e-06,
|
| 20768 |
+
"loss": 20.6476,
|
| 20769 |
+
"step": 29600
|
| 20770 |
+
},
|
| 20771 |
+
{
|
| 20772 |
+
"epoch": 0.5494819031447679,
|
| 20773 |
+
"grad_norm": 34.40625,
|
| 20774 |
+
"learning_rate": 9.914143477528412e-06,
|
| 20775 |
+
"loss": 20.2983,
|
| 20776 |
+
"step": 29610
|
| 20777 |
+
},
|
| 20778 |
+
{
|
| 20779 |
+
"epoch": 0.5496674762292477,
|
| 20780 |
+
"grad_norm": 35.09375,
|
| 20781 |
+
"learning_rate": 9.91411448174237e-06,
|
| 20782 |
+
"loss": 20.609,
|
| 20783 |
+
"step": 29620
|
| 20784 |
+
},
|
| 20785 |
+
{
|
| 20786 |
+
"epoch": 0.5498530493137276,
|
| 20787 |
+
"grad_norm": 35.375,
|
| 20788 |
+
"learning_rate": 9.914085485956327e-06,
|
| 20789 |
+
"loss": 20.1885,
|
| 20790 |
+
"step": 29630
|
| 20791 |
+
},
|
| 20792 |
+
{
|
| 20793 |
+
"epoch": 0.5500386223982073,
|
| 20794 |
+
"grad_norm": 34.1875,
|
| 20795 |
+
"learning_rate": 9.914056490170284e-06,
|
| 20796 |
+
"loss": 19.9611,
|
| 20797 |
+
"step": 29640
|
| 20798 |
+
},
|
| 20799 |
+
{
|
| 20800 |
+
"epoch": 0.5502241954826872,
|
| 20801 |
+
"grad_norm": 33.9375,
|
| 20802 |
+
"learning_rate": 9.914027494384242e-06,
|
| 20803 |
+
"loss": 19.8009,
|
| 20804 |
+
"step": 29650
|
| 20805 |
+
},
|
| 20806 |
+
{
|
| 20807 |
+
"epoch": 0.5504097685671671,
|
| 20808 |
+
"grad_norm": 34.34375,
|
| 20809 |
+
"learning_rate": 9.913998498598199e-06,
|
| 20810 |
+
"loss": 20.1612,
|
| 20811 |
+
"step": 29660
|
| 20812 |
+
},
|
| 20813 |
+
{
|
| 20814 |
+
"epoch": 0.5505953416516468,
|
| 20815 |
+
"grad_norm": 35.1875,
|
| 20816 |
+
"learning_rate": 9.913969502812158e-06,
|
| 20817 |
+
"loss": 20.1682,
|
| 20818 |
+
"step": 29670
|
| 20819 |
+
},
|
| 20820 |
+
{
|
| 20821 |
+
"epoch": 0.5507809147361267,
|
| 20822 |
+
"grad_norm": 32.9375,
|
| 20823 |
+
"learning_rate": 9.913940507026115e-06,
|
| 20824 |
+
"loss": 20.3603,
|
| 20825 |
+
"step": 29680
|
| 20826 |
+
},
|
| 20827 |
+
{
|
| 20828 |
+
"epoch": 0.5509664878206065,
|
| 20829 |
+
"grad_norm": 34.09375,
|
| 20830 |
+
"learning_rate": 9.913911511240071e-06,
|
| 20831 |
+
"loss": 20.2844,
|
| 20832 |
+
"step": 29690
|
| 20833 |
+
},
|
| 20834 |
+
{
|
| 20835 |
+
"epoch": 0.5511520609050863,
|
| 20836 |
+
"grad_norm": 35.78125,
|
| 20837 |
+
"learning_rate": 9.91388251545403e-06,
|
| 20838 |
+
"loss": 20.8106,
|
| 20839 |
+
"step": 29700
|
| 20840 |
+
},
|
| 20841 |
+
{
|
| 20842 |
+
"epoch": 0.5513376339895661,
|
| 20843 |
+
"grad_norm": 35.25,
|
| 20844 |
+
"learning_rate": 9.913853519667988e-06,
|
| 20845 |
+
"loss": 20.042,
|
| 20846 |
+
"step": 29710
|
| 20847 |
+
},
|
| 20848 |
+
{
|
| 20849 |
+
"epoch": 0.551523207074046,
|
| 20850 |
+
"grad_norm": 35.15625,
|
| 20851 |
+
"learning_rate": 9.913824523881945e-06,
|
| 20852 |
+
"loss": 20.7616,
|
| 20853 |
+
"step": 29720
|
| 20854 |
+
},
|
| 20855 |
+
{
|
| 20856 |
+
"epoch": 0.5517087801585258,
|
| 20857 |
+
"grad_norm": 37.375,
|
| 20858 |
+
"learning_rate": 9.913795528095902e-06,
|
| 20859 |
+
"loss": 20.4981,
|
| 20860 |
+
"step": 29730
|
| 20861 |
+
},
|
| 20862 |
+
{
|
| 20863 |
+
"epoch": 0.5518943532430056,
|
| 20864 |
+
"grad_norm": 34.34375,
|
| 20865 |
+
"learning_rate": 9.91376653230986e-06,
|
| 20866 |
+
"loss": 20.4069,
|
| 20867 |
+
"step": 29740
|
| 20868 |
+
},
|
| 20869 |
+
{
|
| 20870 |
+
"epoch": 0.5520799263274855,
|
| 20871 |
+
"grad_norm": 34.25,
|
| 20872 |
+
"learning_rate": 9.913737536523817e-06,
|
| 20873 |
+
"loss": 20.3144,
|
| 20874 |
+
"step": 29750
|
| 20875 |
+
},
|
| 20876 |
+
{
|
| 20877 |
+
"epoch": 0.5522654994119653,
|
| 20878 |
+
"grad_norm": 35.28125,
|
| 20879 |
+
"learning_rate": 9.913708540737775e-06,
|
| 20880 |
+
"loss": 20.3102,
|
| 20881 |
+
"step": 29760
|
| 20882 |
+
},
|
| 20883 |
+
{
|
| 20884 |
+
"epoch": 0.5524510724964451,
|
| 20885 |
+
"grad_norm": 34.09375,
|
| 20886 |
+
"learning_rate": 9.913679544951734e-06,
|
| 20887 |
+
"loss": 20.3573,
|
| 20888 |
+
"step": 29770
|
| 20889 |
+
},
|
| 20890 |
+
{
|
| 20891 |
+
"epoch": 0.552636645580925,
|
| 20892 |
+
"grad_norm": 37.34375,
|
| 20893 |
+
"learning_rate": 9.913650549165691e-06,
|
| 20894 |
+
"loss": 19.9171,
|
| 20895 |
+
"step": 29780
|
| 20896 |
+
},
|
| 20897 |
+
{
|
| 20898 |
+
"epoch": 0.5528222186654048,
|
| 20899 |
+
"grad_norm": 32.78125,
|
| 20900 |
+
"learning_rate": 9.913621553379647e-06,
|
| 20901 |
+
"loss": 20.3493,
|
| 20902 |
+
"step": 29790
|
| 20903 |
+
},
|
| 20904 |
+
{
|
| 20905 |
+
"epoch": 0.5530077917498846,
|
| 20906 |
+
"grad_norm": 35.28125,
|
| 20907 |
+
"learning_rate": 9.913592557593606e-06,
|
| 20908 |
+
"loss": 20.3546,
|
| 20909 |
+
"step": 29800
|
| 20910 |
+
},
|
| 20911 |
+
{
|
| 20912 |
+
"epoch": 0.5531933648343644,
|
| 20913 |
+
"grad_norm": 33.09375,
|
| 20914 |
+
"learning_rate": 9.913563561807563e-06,
|
| 20915 |
+
"loss": 20.5089,
|
| 20916 |
+
"step": 29810
|
| 20917 |
+
},
|
| 20918 |
+
{
|
| 20919 |
+
"epoch": 0.5533789379188443,
|
| 20920 |
+
"grad_norm": 35.03125,
|
| 20921 |
+
"learning_rate": 9.91353456602152e-06,
|
| 20922 |
+
"loss": 20.2858,
|
| 20923 |
+
"step": 29820
|
| 20924 |
+
},
|
| 20925 |
+
{
|
| 20926 |
+
"epoch": 0.553564511003324,
|
| 20927 |
+
"grad_norm": 33.9375,
|
| 20928 |
+
"learning_rate": 9.913505570235478e-06,
|
| 20929 |
+
"loss": 20.5666,
|
| 20930 |
+
"step": 29830
|
| 20931 |
+
},
|
| 20932 |
+
{
|
| 20933 |
+
"epoch": 0.5537500840878039,
|
| 20934 |
+
"grad_norm": 33.3125,
|
| 20935 |
+
"learning_rate": 9.913476574449437e-06,
|
| 20936 |
+
"loss": 20.3237,
|
| 20937 |
+
"step": 29840
|
| 20938 |
+
},
|
| 20939 |
+
{
|
| 20940 |
+
"epoch": 0.5539356571722838,
|
| 20941 |
+
"grad_norm": 33.4375,
|
| 20942 |
+
"learning_rate": 9.913447578663393e-06,
|
| 20943 |
+
"loss": 20.3019,
|
| 20944 |
+
"step": 29850
|
| 20945 |
+
},
|
| 20946 |
+
{
|
| 20947 |
+
"epoch": 0.5541212302567635,
|
| 20948 |
+
"grad_norm": 34.4375,
|
| 20949 |
+
"learning_rate": 9.91341858287735e-06,
|
| 20950 |
+
"loss": 20.2433,
|
| 20951 |
+
"step": 29860
|
| 20952 |
+
},
|
| 20953 |
+
{
|
| 20954 |
+
"epoch": 0.5543068033412434,
|
| 20955 |
+
"grad_norm": 37.40625,
|
| 20956 |
+
"learning_rate": 9.91338958709131e-06,
|
| 20957 |
+
"loss": 20.648,
|
| 20958 |
+
"step": 29870
|
| 20959 |
+
},
|
| 20960 |
+
{
|
| 20961 |
+
"epoch": 0.5544923764257232,
|
| 20962 |
+
"grad_norm": 34.28125,
|
| 20963 |
+
"learning_rate": 9.913360591305267e-06,
|
| 20964 |
+
"loss": 20.0626,
|
| 20965 |
+
"step": 29880
|
| 20966 |
+
},
|
| 20967 |
+
{
|
| 20968 |
+
"epoch": 0.554677949510203,
|
| 20969 |
+
"grad_norm": 37.0,
|
| 20970 |
+
"learning_rate": 9.913331595519224e-06,
|
| 20971 |
+
"loss": 19.7648,
|
| 20972 |
+
"step": 29890
|
| 20973 |
+
},
|
| 20974 |
+
{
|
| 20975 |
+
"epoch": 0.5548635225946829,
|
| 20976 |
+
"grad_norm": 35.71875,
|
| 20977 |
+
"learning_rate": 9.913302599733182e-06,
|
| 20978 |
+
"loss": 20.2157,
|
| 20979 |
+
"step": 29900
|
| 20980 |
+
},
|
| 20981 |
+
{
|
| 20982 |
+
"epoch": 0.5550490956791627,
|
| 20983 |
+
"grad_norm": 35.125,
|
| 20984 |
+
"learning_rate": 9.91327360394714e-06,
|
| 20985 |
+
"loss": 20.0799,
|
| 20986 |
+
"step": 29910
|
| 20987 |
+
},
|
| 20988 |
+
{
|
| 20989 |
+
"epoch": 0.5552346687636425,
|
| 20990 |
+
"grad_norm": 36.6875,
|
| 20991 |
+
"learning_rate": 9.913244608161097e-06,
|
| 20992 |
+
"loss": 20.2929,
|
| 20993 |
+
"step": 29920
|
| 20994 |
+
},
|
| 20995 |
+
{
|
| 20996 |
+
"epoch": 0.5554202418481223,
|
| 20997 |
+
"grad_norm": 34.25,
|
| 20998 |
+
"learning_rate": 9.913215612375054e-06,
|
| 20999 |
+
"loss": 20.0606,
|
| 21000 |
+
"step": 29930
|
| 21001 |
+
},
|
| 21002 |
+
{
|
| 21003 |
+
"epoch": 0.5556058149326022,
|
| 21004 |
+
"grad_norm": 35.3125,
|
| 21005 |
+
"learning_rate": 9.913186616589011e-06,
|
| 21006 |
+
"loss": 20.3643,
|
| 21007 |
+
"step": 29940
|
| 21008 |
+
},
|
| 21009 |
+
{
|
| 21010 |
+
"epoch": 0.555791388017082,
|
| 21011 |
+
"grad_norm": 34.84375,
|
| 21012 |
+
"learning_rate": 9.91315762080297e-06,
|
| 21013 |
+
"loss": 20.3112,
|
| 21014 |
+
"step": 29950
|
| 21015 |
+
},
|
| 21016 |
+
{
|
| 21017 |
+
"epoch": 0.5559769611015618,
|
| 21018 |
+
"grad_norm": 34.46875,
|
| 21019 |
+
"learning_rate": 9.913128625016926e-06,
|
| 21020 |
+
"loss": 20.1955,
|
| 21021 |
+
"step": 29960
|
| 21022 |
+
},
|
| 21023 |
+
{
|
| 21024 |
+
"epoch": 0.5561625341860417,
|
| 21025 |
+
"grad_norm": 34.4375,
|
| 21026 |
+
"learning_rate": 9.913099629230885e-06,
|
| 21027 |
+
"loss": 20.1614,
|
| 21028 |
+
"step": 29970
|
| 21029 |
+
},
|
| 21030 |
+
{
|
| 21031 |
+
"epoch": 0.5563481072705215,
|
| 21032 |
+
"grad_norm": 35.625,
|
| 21033 |
+
"learning_rate": 9.913070633444843e-06,
|
| 21034 |
+
"loss": 20.1444,
|
| 21035 |
+
"step": 29980
|
| 21036 |
+
},
|
| 21037 |
+
{
|
| 21038 |
+
"epoch": 0.5565336803550013,
|
| 21039 |
+
"grad_norm": 34.34375,
|
| 21040 |
+
"learning_rate": 9.9130416376588e-06,
|
| 21041 |
+
"loss": 20.4401,
|
| 21042 |
+
"step": 29990
|
| 21043 |
+
},
|
| 21044 |
+
{
|
| 21045 |
+
"epoch": 0.5567192534394811,
|
| 21046 |
+
"grad_norm": 35.65625,
|
| 21047 |
+
"learning_rate": 9.913012641872758e-06,
|
| 21048 |
+
"loss": 20.0201,
|
| 21049 |
+
"step": 30000
|
| 21050 |
+
},
|
| 21051 |
+
{
|
| 21052 |
+
"epoch": 0.5567192534394811,
|
| 21053 |
+
"eval_loss": 2.5262889862060547,
|
| 21054 |
+
"eval_runtime": 455.282,
|
| 21055 |
+
"eval_samples_per_second": 3189.489,
|
| 21056 |
+
"eval_steps_per_second": 49.837,
|
| 21057 |
+
"step": 30000
|
| 21058 |
}
|
| 21059 |
],
|
| 21060 |
"logging_steps": 10,
|
|
|
|
| 21074 |
"attributes": {}
|
| 21075 |
}
|
| 21076 |
},
|
| 21077 |
+
"total_flos": 5.236568277123072e+18,
|
| 21078 |
"train_batch_size": 8,
|
| 21079 |
"trial_name": null,
|
| 21080 |
"trial_params": null
|