Training in progress, step 15000, 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 sha256:6f373582976d1956b2adacaa9790d35a15e2ac8bfa5e191ca43010f1c2968b03
|
| 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 sha256:5c0bb19542ae52096c047421eb4fe0a1f9474910bfbb60e796399a92ff1ef2e8
|
| 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 sha256:0c461c9d337dfc684e9352ec72bfa344e2f5d377f7cfc4475de9acae294dca89
|
| 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:fae392ec6232cbf9da21d6ed12bc8247d0d24e7f3a3606acd23be00f3e8bbfc5
|
| 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:cbf3e7ca9991a58b0b16574a3c653483c551c270aa05aba06c162ea593f7b0f2
|
| 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:c695bebf6bcb75cbe26378bfe0ab7e2a33c49f713b9d6e4d10632b24322977e7
|
| 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:d5ebb13c71265c5464c9aa9bb9b66f07764d73befe6cd63a2aaf8e781bf0a374
|
| 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 sha256:12cc6e245e189be568c8dfd43a4dd8f04bb3dbd9f17f41458107935d2c2a6a9d
|
| 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 sha256:36086646e9a8f76fea69f8a227112e83bb63524964ccdfb82f4cdad88b90e5e4
|
| 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 sha256:9b44153bacf860d0ca6ce4c6b9380a199feab8a72ca613e6745bfb671b02c4e4
|
| 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 sha256:8f2df6d70b5170ad5f7fbfdd1443cce32f9e75727c4f2973f33ec9362abb0dd0
|
| 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,
|
|
@@ -7023,6 +7023,3514 @@
|
|
| 7023 |
"eval_samples_per_second": 3192.564,
|
| 7024 |
"eval_steps_per_second": 49.885,
|
| 7025 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7026 |
}
|
| 7027 |
],
|
| 7028 |
"logging_steps": 10,
|
|
@@ -7042,7 +10550,7 @@
|
|
| 7042 |
"attributes": {}
|
| 7043 |
}
|
| 7044 |
},
|
| 7045 |
-
"total_flos":
|
| 7046 |
"train_batch_size": 8,
|
| 7047 |
"trial_name": null,
|
| 7048 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.27835962671974057,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 15000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7023 |
"eval_samples_per_second": 3192.564,
|
| 7024 |
"eval_steps_per_second": 49.885,
|
| 7025 |
"step": 10000
|
| 7026 |
+
},
|
| 7027 |
+
{
|
| 7028 |
+
"epoch": 0.18575865756430687,
|
| 7029 |
+
"grad_norm": 34.6875,
|
| 7030 |
+
"learning_rate": 9.970975218171543e-06,
|
| 7031 |
+
"loss": 24.2771,
|
| 7032 |
+
"step": 10010
|
| 7033 |
+
},
|
| 7034 |
+
{
|
| 7035 |
+
"epoch": 0.1859442306487867,
|
| 7036 |
+
"grad_norm": 35.9375,
|
| 7037 |
+
"learning_rate": 9.9709462223855e-06,
|
| 7038 |
+
"loss": 23.6828,
|
| 7039 |
+
"step": 10020
|
| 7040 |
+
},
|
| 7041 |
+
{
|
| 7042 |
+
"epoch": 0.1861298037332665,
|
| 7043 |
+
"grad_norm": 34.8125,
|
| 7044 |
+
"learning_rate": 9.97091722659946e-06,
|
| 7045 |
+
"loss": 23.3928,
|
| 7046 |
+
"step": 10030
|
| 7047 |
+
},
|
| 7048 |
+
{
|
| 7049 |
+
"epoch": 0.18631537681774635,
|
| 7050 |
+
"grad_norm": 35.125,
|
| 7051 |
+
"learning_rate": 9.970888230813417e-06,
|
| 7052 |
+
"loss": 23.5511,
|
| 7053 |
+
"step": 10040
|
| 7054 |
+
},
|
| 7055 |
+
{
|
| 7056 |
+
"epoch": 0.18650094990222618,
|
| 7057 |
+
"grad_norm": 35.46875,
|
| 7058 |
+
"learning_rate": 9.970859235027375e-06,
|
| 7059 |
+
"loss": 23.9068,
|
| 7060 |
+
"step": 10050
|
| 7061 |
+
},
|
| 7062 |
+
{
|
| 7063 |
+
"epoch": 0.18668652298670602,
|
| 7064 |
+
"grad_norm": 34.25,
|
| 7065 |
+
"learning_rate": 9.970830239241332e-06,
|
| 7066 |
+
"loss": 23.9531,
|
| 7067 |
+
"step": 10060
|
| 7068 |
+
},
|
| 7069 |
+
{
|
| 7070 |
+
"epoch": 0.18687209607118582,
|
| 7071 |
+
"grad_norm": 35.53125,
|
| 7072 |
+
"learning_rate": 9.97080124345529e-06,
|
| 7073 |
+
"loss": 23.9918,
|
| 7074 |
+
"step": 10070
|
| 7075 |
+
},
|
| 7076 |
+
{
|
| 7077 |
+
"epoch": 0.18705766915566566,
|
| 7078 |
+
"grad_norm": 35.03125,
|
| 7079 |
+
"learning_rate": 9.970772247669247e-06,
|
| 7080 |
+
"loss": 23.7854,
|
| 7081 |
+
"step": 10080
|
| 7082 |
+
},
|
| 7083 |
+
{
|
| 7084 |
+
"epoch": 0.1872432422401455,
|
| 7085 |
+
"grad_norm": 35.0625,
|
| 7086 |
+
"learning_rate": 9.970743251883204e-06,
|
| 7087 |
+
"loss": 23.7965,
|
| 7088 |
+
"step": 10090
|
| 7089 |
+
},
|
| 7090 |
+
{
|
| 7091 |
+
"epoch": 0.18742881532462533,
|
| 7092 |
+
"grad_norm": 35.25,
|
| 7093 |
+
"learning_rate": 9.970714256097163e-06,
|
| 7094 |
+
"loss": 23.72,
|
| 7095 |
+
"step": 10100
|
| 7096 |
+
},
|
| 7097 |
+
{
|
| 7098 |
+
"epoch": 0.18761438840910513,
|
| 7099 |
+
"grad_norm": 34.71875,
|
| 7100 |
+
"learning_rate": 9.970685260311119e-06,
|
| 7101 |
+
"loss": 23.7046,
|
| 7102 |
+
"step": 10110
|
| 7103 |
+
},
|
| 7104 |
+
{
|
| 7105 |
+
"epoch": 0.18779996149358497,
|
| 7106 |
+
"grad_norm": 34.75,
|
| 7107 |
+
"learning_rate": 9.970656264525076e-06,
|
| 7108 |
+
"loss": 24.0583,
|
| 7109 |
+
"step": 10120
|
| 7110 |
+
},
|
| 7111 |
+
{
|
| 7112 |
+
"epoch": 0.1879855345780648,
|
| 7113 |
+
"grad_norm": 35.21875,
|
| 7114 |
+
"learning_rate": 9.970627268739036e-06,
|
| 7115 |
+
"loss": 23.5673,
|
| 7116 |
+
"step": 10130
|
| 7117 |
+
},
|
| 7118 |
+
{
|
| 7119 |
+
"epoch": 0.18817110766254463,
|
| 7120 |
+
"grad_norm": 35.0625,
|
| 7121 |
+
"learning_rate": 9.970598272952993e-06,
|
| 7122 |
+
"loss": 23.5273,
|
| 7123 |
+
"step": 10140
|
| 7124 |
+
},
|
| 7125 |
+
{
|
| 7126 |
+
"epoch": 0.18835668074702444,
|
| 7127 |
+
"grad_norm": 35.625,
|
| 7128 |
+
"learning_rate": 9.97056927716695e-06,
|
| 7129 |
+
"loss": 23.8089,
|
| 7130 |
+
"step": 10150
|
| 7131 |
+
},
|
| 7132 |
+
{
|
| 7133 |
+
"epoch": 0.18854225383150428,
|
| 7134 |
+
"grad_norm": 35.0,
|
| 7135 |
+
"learning_rate": 9.970540281380908e-06,
|
| 7136 |
+
"loss": 23.5046,
|
| 7137 |
+
"step": 10160
|
| 7138 |
+
},
|
| 7139 |
+
{
|
| 7140 |
+
"epoch": 0.1887278269159841,
|
| 7141 |
+
"grad_norm": 35.75,
|
| 7142 |
+
"learning_rate": 9.970511285594865e-06,
|
| 7143 |
+
"loss": 23.674,
|
| 7144 |
+
"step": 10170
|
| 7145 |
+
},
|
| 7146 |
+
{
|
| 7147 |
+
"epoch": 0.18891340000046394,
|
| 7148 |
+
"grad_norm": 33.59375,
|
| 7149 |
+
"learning_rate": 9.970482289808823e-06,
|
| 7150 |
+
"loss": 23.2634,
|
| 7151 |
+
"step": 10180
|
| 7152 |
+
},
|
| 7153 |
+
{
|
| 7154 |
+
"epoch": 0.18909897308494375,
|
| 7155 |
+
"grad_norm": 36.53125,
|
| 7156 |
+
"learning_rate": 9.97045329402278e-06,
|
| 7157 |
+
"loss": 23.6953,
|
| 7158 |
+
"step": 10190
|
| 7159 |
+
},
|
| 7160 |
+
{
|
| 7161 |
+
"epoch": 0.18928454616942358,
|
| 7162 |
+
"grad_norm": 36.4375,
|
| 7163 |
+
"learning_rate": 9.970424298236739e-06,
|
| 7164 |
+
"loss": 23.4264,
|
| 7165 |
+
"step": 10200
|
| 7166 |
+
},
|
| 7167 |
+
{
|
| 7168 |
+
"epoch": 0.18947011925390342,
|
| 7169 |
+
"grad_norm": 35.3125,
|
| 7170 |
+
"learning_rate": 9.970395302450696e-06,
|
| 7171 |
+
"loss": 23.1557,
|
| 7172 |
+
"step": 10210
|
| 7173 |
+
},
|
| 7174 |
+
{
|
| 7175 |
+
"epoch": 0.18965569233838325,
|
| 7176 |
+
"grad_norm": 34.90625,
|
| 7177 |
+
"learning_rate": 9.970366306664652e-06,
|
| 7178 |
+
"loss": 23.539,
|
| 7179 |
+
"step": 10220
|
| 7180 |
+
},
|
| 7181 |
+
{
|
| 7182 |
+
"epoch": 0.18984126542286306,
|
| 7183 |
+
"grad_norm": 34.09375,
|
| 7184 |
+
"learning_rate": 9.970337310878611e-06,
|
| 7185 |
+
"loss": 23.8074,
|
| 7186 |
+
"step": 10230
|
| 7187 |
+
},
|
| 7188 |
+
{
|
| 7189 |
+
"epoch": 0.1900268385073429,
|
| 7190 |
+
"grad_norm": 37.5625,
|
| 7191 |
+
"learning_rate": 9.970308315092569e-06,
|
| 7192 |
+
"loss": 23.6492,
|
| 7193 |
+
"step": 10240
|
| 7194 |
+
},
|
| 7195 |
+
{
|
| 7196 |
+
"epoch": 0.19021241159182273,
|
| 7197 |
+
"grad_norm": 33.40625,
|
| 7198 |
+
"learning_rate": 9.970279319306526e-06,
|
| 7199 |
+
"loss": 23.5333,
|
| 7200 |
+
"step": 10250
|
| 7201 |
+
},
|
| 7202 |
+
{
|
| 7203 |
+
"epoch": 0.19039798467630256,
|
| 7204 |
+
"grad_norm": 35.15625,
|
| 7205 |
+
"learning_rate": 9.970250323520484e-06,
|
| 7206 |
+
"loss": 23.7081,
|
| 7207 |
+
"step": 10260
|
| 7208 |
+
},
|
| 7209 |
+
{
|
| 7210 |
+
"epoch": 0.19058355776078237,
|
| 7211 |
+
"grad_norm": 36.21875,
|
| 7212 |
+
"learning_rate": 9.970221327734441e-06,
|
| 7213 |
+
"loss": 23.0549,
|
| 7214 |
+
"step": 10270
|
| 7215 |
+
},
|
| 7216 |
+
{
|
| 7217 |
+
"epoch": 0.1907691308452622,
|
| 7218 |
+
"grad_norm": 37.1875,
|
| 7219 |
+
"learning_rate": 9.970192331948398e-06,
|
| 7220 |
+
"loss": 23.529,
|
| 7221 |
+
"step": 10280
|
| 7222 |
+
},
|
| 7223 |
+
{
|
| 7224 |
+
"epoch": 0.19095470392974204,
|
| 7225 |
+
"grad_norm": 35.6875,
|
| 7226 |
+
"learning_rate": 9.970163336162356e-06,
|
| 7227 |
+
"loss": 23.6286,
|
| 7228 |
+
"step": 10290
|
| 7229 |
+
},
|
| 7230 |
+
{
|
| 7231 |
+
"epoch": 0.19114027701422187,
|
| 7232 |
+
"grad_norm": 36.9375,
|
| 7233 |
+
"learning_rate": 9.970134340376315e-06,
|
| 7234 |
+
"loss": 23.9015,
|
| 7235 |
+
"step": 10300
|
| 7236 |
+
},
|
| 7237 |
+
{
|
| 7238 |
+
"epoch": 0.19132585009870168,
|
| 7239 |
+
"grad_norm": 35.1875,
|
| 7240 |
+
"learning_rate": 9.970105344590272e-06,
|
| 7241 |
+
"loss": 23.5723,
|
| 7242 |
+
"step": 10310
|
| 7243 |
+
},
|
| 7244 |
+
{
|
| 7245 |
+
"epoch": 0.1915114231831815,
|
| 7246 |
+
"grad_norm": 33.78125,
|
| 7247 |
+
"learning_rate": 9.97007634880423e-06,
|
| 7248 |
+
"loss": 23.5897,
|
| 7249 |
+
"step": 10320
|
| 7250 |
+
},
|
| 7251 |
+
{
|
| 7252 |
+
"epoch": 0.19169699626766135,
|
| 7253 |
+
"grad_norm": 33.90625,
|
| 7254 |
+
"learning_rate": 9.970047353018187e-06,
|
| 7255 |
+
"loss": 24.0946,
|
| 7256 |
+
"step": 10330
|
| 7257 |
+
},
|
| 7258 |
+
{
|
| 7259 |
+
"epoch": 0.19188256935214115,
|
| 7260 |
+
"grad_norm": 35.15625,
|
| 7261 |
+
"learning_rate": 9.970018357232144e-06,
|
| 7262 |
+
"loss": 23.9352,
|
| 7263 |
+
"step": 10340
|
| 7264 |
+
},
|
| 7265 |
+
{
|
| 7266 |
+
"epoch": 0.192068142436621,
|
| 7267 |
+
"grad_norm": 37.9375,
|
| 7268 |
+
"learning_rate": 9.969989361446102e-06,
|
| 7269 |
+
"loss": 23.9665,
|
| 7270 |
+
"step": 10350
|
| 7271 |
+
},
|
| 7272 |
+
{
|
| 7273 |
+
"epoch": 0.19225371552110082,
|
| 7274 |
+
"grad_norm": 33.53125,
|
| 7275 |
+
"learning_rate": 9.96996036566006e-06,
|
| 7276 |
+
"loss": 23.3916,
|
| 7277 |
+
"step": 10360
|
| 7278 |
+
},
|
| 7279 |
+
{
|
| 7280 |
+
"epoch": 0.19243928860558066,
|
| 7281 |
+
"grad_norm": 34.3125,
|
| 7282 |
+
"learning_rate": 9.969931369874017e-06,
|
| 7283 |
+
"loss": 23.5357,
|
| 7284 |
+
"step": 10370
|
| 7285 |
+
},
|
| 7286 |
+
{
|
| 7287 |
+
"epoch": 0.19262486169006046,
|
| 7288 |
+
"grad_norm": 32.90625,
|
| 7289 |
+
"learning_rate": 9.969902374087974e-06,
|
| 7290 |
+
"loss": 24.2098,
|
| 7291 |
+
"step": 10380
|
| 7292 |
+
},
|
| 7293 |
+
{
|
| 7294 |
+
"epoch": 0.1928104347745403,
|
| 7295 |
+
"grad_norm": 36.4375,
|
| 7296 |
+
"learning_rate": 9.969873378301931e-06,
|
| 7297 |
+
"loss": 23.3536,
|
| 7298 |
+
"step": 10390
|
| 7299 |
+
},
|
| 7300 |
+
{
|
| 7301 |
+
"epoch": 0.19299600785902013,
|
| 7302 |
+
"grad_norm": 34.96875,
|
| 7303 |
+
"learning_rate": 9.969844382515889e-06,
|
| 7304 |
+
"loss": 23.5672,
|
| 7305 |
+
"step": 10400
|
| 7306 |
+
},
|
| 7307 |
+
{
|
| 7308 |
+
"epoch": 0.19318158094349996,
|
| 7309 |
+
"grad_norm": 35.21875,
|
| 7310 |
+
"learning_rate": 9.969815386729848e-06,
|
| 7311 |
+
"loss": 23.262,
|
| 7312 |
+
"step": 10410
|
| 7313 |
+
},
|
| 7314 |
+
{
|
| 7315 |
+
"epoch": 0.19336715402797977,
|
| 7316 |
+
"grad_norm": 34.9375,
|
| 7317 |
+
"learning_rate": 9.969786390943805e-06,
|
| 7318 |
+
"loss": 23.7718,
|
| 7319 |
+
"step": 10420
|
| 7320 |
+
},
|
| 7321 |
+
{
|
| 7322 |
+
"epoch": 0.1935527271124596,
|
| 7323 |
+
"grad_norm": 35.1875,
|
| 7324 |
+
"learning_rate": 9.969757395157763e-06,
|
| 7325 |
+
"loss": 23.3401,
|
| 7326 |
+
"step": 10430
|
| 7327 |
+
},
|
| 7328 |
+
{
|
| 7329 |
+
"epoch": 0.19373830019693944,
|
| 7330 |
+
"grad_norm": 33.28125,
|
| 7331 |
+
"learning_rate": 9.96972839937172e-06,
|
| 7332 |
+
"loss": 23.7594,
|
| 7333 |
+
"step": 10440
|
| 7334 |
+
},
|
| 7335 |
+
{
|
| 7336 |
+
"epoch": 0.19392387328141927,
|
| 7337 |
+
"grad_norm": 35.8125,
|
| 7338 |
+
"learning_rate": 9.969699403585678e-06,
|
| 7339 |
+
"loss": 23.7115,
|
| 7340 |
+
"step": 10450
|
| 7341 |
+
},
|
| 7342 |
+
{
|
| 7343 |
+
"epoch": 0.19410944636589908,
|
| 7344 |
+
"grad_norm": 35.6875,
|
| 7345 |
+
"learning_rate": 9.969670407799635e-06,
|
| 7346 |
+
"loss": 23.6501,
|
| 7347 |
+
"step": 10460
|
| 7348 |
+
},
|
| 7349 |
+
{
|
| 7350 |
+
"epoch": 0.19429501945037891,
|
| 7351 |
+
"grad_norm": 34.15625,
|
| 7352 |
+
"learning_rate": 9.969641412013592e-06,
|
| 7353 |
+
"loss": 23.6335,
|
| 7354 |
+
"step": 10470
|
| 7355 |
+
},
|
| 7356 |
+
{
|
| 7357 |
+
"epoch": 0.19448059253485875,
|
| 7358 |
+
"grad_norm": 33.75,
|
| 7359 |
+
"learning_rate": 9.969612416227552e-06,
|
| 7360 |
+
"loss": 24.3264,
|
| 7361 |
+
"step": 10480
|
| 7362 |
+
},
|
| 7363 |
+
{
|
| 7364 |
+
"epoch": 0.19466616561933858,
|
| 7365 |
+
"grad_norm": 34.46875,
|
| 7366 |
+
"learning_rate": 9.969583420441507e-06,
|
| 7367 |
+
"loss": 23.7675,
|
| 7368 |
+
"step": 10490
|
| 7369 |
+
},
|
| 7370 |
+
{
|
| 7371 |
+
"epoch": 0.1948517387038184,
|
| 7372 |
+
"grad_norm": 35.3125,
|
| 7373 |
+
"learning_rate": 9.969554424655465e-06,
|
| 7374 |
+
"loss": 23.1986,
|
| 7375 |
+
"step": 10500
|
| 7376 |
+
},
|
| 7377 |
+
{
|
| 7378 |
+
"epoch": 0.19503731178829822,
|
| 7379 |
+
"grad_norm": 33.8125,
|
| 7380 |
+
"learning_rate": 9.969525428869424e-06,
|
| 7381 |
+
"loss": 23.6049,
|
| 7382 |
+
"step": 10510
|
| 7383 |
+
},
|
| 7384 |
+
{
|
| 7385 |
+
"epoch": 0.19522288487277806,
|
| 7386 |
+
"grad_norm": 34.34375,
|
| 7387 |
+
"learning_rate": 9.969496433083381e-06,
|
| 7388 |
+
"loss": 23.5584,
|
| 7389 |
+
"step": 10520
|
| 7390 |
+
},
|
| 7391 |
+
{
|
| 7392 |
+
"epoch": 0.1954084579572579,
|
| 7393 |
+
"grad_norm": 35.65625,
|
| 7394 |
+
"learning_rate": 9.969467437297339e-06,
|
| 7395 |
+
"loss": 23.4357,
|
| 7396 |
+
"step": 10530
|
| 7397 |
+
},
|
| 7398 |
+
{
|
| 7399 |
+
"epoch": 0.1955940310417377,
|
| 7400 |
+
"grad_norm": 32.65625,
|
| 7401 |
+
"learning_rate": 9.969438441511296e-06,
|
| 7402 |
+
"loss": 23.708,
|
| 7403 |
+
"step": 10540
|
| 7404 |
+
},
|
| 7405 |
+
{
|
| 7406 |
+
"epoch": 0.19577960412621753,
|
| 7407 |
+
"grad_norm": 34.25,
|
| 7408 |
+
"learning_rate": 9.969409445725253e-06,
|
| 7409 |
+
"loss": 23.0793,
|
| 7410 |
+
"step": 10550
|
| 7411 |
+
},
|
| 7412 |
+
{
|
| 7413 |
+
"epoch": 0.19596517721069737,
|
| 7414 |
+
"grad_norm": 35.15625,
|
| 7415 |
+
"learning_rate": 9.96938044993921e-06,
|
| 7416 |
+
"loss": 23.6577,
|
| 7417 |
+
"step": 10560
|
| 7418 |
+
},
|
| 7419 |
+
{
|
| 7420 |
+
"epoch": 0.1961507502951772,
|
| 7421 |
+
"grad_norm": 34.03125,
|
| 7422 |
+
"learning_rate": 9.969351454153168e-06,
|
| 7423 |
+
"loss": 23.9597,
|
| 7424 |
+
"step": 10570
|
| 7425 |
+
},
|
| 7426 |
+
{
|
| 7427 |
+
"epoch": 0.196336323379657,
|
| 7428 |
+
"grad_norm": 34.15625,
|
| 7429 |
+
"learning_rate": 9.969322458367127e-06,
|
| 7430 |
+
"loss": 23.5803,
|
| 7431 |
+
"step": 10580
|
| 7432 |
+
},
|
| 7433 |
+
{
|
| 7434 |
+
"epoch": 0.19652189646413684,
|
| 7435 |
+
"grad_norm": 31.96875,
|
| 7436 |
+
"learning_rate": 9.969293462581085e-06,
|
| 7437 |
+
"loss": 23.3744,
|
| 7438 |
+
"step": 10590
|
| 7439 |
+
},
|
| 7440 |
+
{
|
| 7441 |
+
"epoch": 0.19670746954861668,
|
| 7442 |
+
"grad_norm": 34.9375,
|
| 7443 |
+
"learning_rate": 9.96926446679504e-06,
|
| 7444 |
+
"loss": 23.2254,
|
| 7445 |
+
"step": 10600
|
| 7446 |
+
},
|
| 7447 |
+
{
|
| 7448 |
+
"epoch": 0.19689304263309648,
|
| 7449 |
+
"grad_norm": 35.9375,
|
| 7450 |
+
"learning_rate": 9.969235471009e-06,
|
| 7451 |
+
"loss": 23.7611,
|
| 7452 |
+
"step": 10610
|
| 7453 |
+
},
|
| 7454 |
+
{
|
| 7455 |
+
"epoch": 0.19707861571757632,
|
| 7456 |
+
"grad_norm": 36.28125,
|
| 7457 |
+
"learning_rate": 9.969206475222957e-06,
|
| 7458 |
+
"loss": 23.9757,
|
| 7459 |
+
"step": 10620
|
| 7460 |
+
},
|
| 7461 |
+
{
|
| 7462 |
+
"epoch": 0.19726418880205615,
|
| 7463 |
+
"grad_norm": 35.53125,
|
| 7464 |
+
"learning_rate": 9.969177479436914e-06,
|
| 7465 |
+
"loss": 23.0914,
|
| 7466 |
+
"step": 10630
|
| 7467 |
+
},
|
| 7468 |
+
{
|
| 7469 |
+
"epoch": 0.19744976188653598,
|
| 7470 |
+
"grad_norm": 37.25,
|
| 7471 |
+
"learning_rate": 9.969148483650872e-06,
|
| 7472 |
+
"loss": 23.5468,
|
| 7473 |
+
"step": 10640
|
| 7474 |
+
},
|
| 7475 |
+
{
|
| 7476 |
+
"epoch": 0.1976353349710158,
|
| 7477 |
+
"grad_norm": 33.375,
|
| 7478 |
+
"learning_rate": 9.96911948786483e-06,
|
| 7479 |
+
"loss": 23.446,
|
| 7480 |
+
"step": 10650
|
| 7481 |
+
},
|
| 7482 |
+
{
|
| 7483 |
+
"epoch": 0.19782090805549563,
|
| 7484 |
+
"grad_norm": 35.09375,
|
| 7485 |
+
"learning_rate": 9.969090492078787e-06,
|
| 7486 |
+
"loss": 23.2178,
|
| 7487 |
+
"step": 10660
|
| 7488 |
+
},
|
| 7489 |
+
{
|
| 7490 |
+
"epoch": 0.19800648113997546,
|
| 7491 |
+
"grad_norm": 36.5,
|
| 7492 |
+
"learning_rate": 9.969061496292744e-06,
|
| 7493 |
+
"loss": 23.965,
|
| 7494 |
+
"step": 10670
|
| 7495 |
+
},
|
| 7496 |
+
{
|
| 7497 |
+
"epoch": 0.1981920542244553,
|
| 7498 |
+
"grad_norm": 34.21875,
|
| 7499 |
+
"learning_rate": 9.969032500506703e-06,
|
| 7500 |
+
"loss": 23.7035,
|
| 7501 |
+
"step": 10680
|
| 7502 |
+
},
|
| 7503 |
+
{
|
| 7504 |
+
"epoch": 0.1983776273089351,
|
| 7505 |
+
"grad_norm": 33.21875,
|
| 7506 |
+
"learning_rate": 9.96900350472066e-06,
|
| 7507 |
+
"loss": 23.758,
|
| 7508 |
+
"step": 10690
|
| 7509 |
+
},
|
| 7510 |
+
{
|
| 7511 |
+
"epoch": 0.19856320039341493,
|
| 7512 |
+
"grad_norm": 34.84375,
|
| 7513 |
+
"learning_rate": 9.968974508934616e-06,
|
| 7514 |
+
"loss": 23.4742,
|
| 7515 |
+
"step": 10700
|
| 7516 |
+
},
|
| 7517 |
+
{
|
| 7518 |
+
"epoch": 0.19874877347789477,
|
| 7519 |
+
"grad_norm": 34.46875,
|
| 7520 |
+
"learning_rate": 9.968945513148575e-06,
|
| 7521 |
+
"loss": 23.8078,
|
| 7522 |
+
"step": 10710
|
| 7523 |
+
},
|
| 7524 |
+
{
|
| 7525 |
+
"epoch": 0.1989343465623746,
|
| 7526 |
+
"grad_norm": 34.4375,
|
| 7527 |
+
"learning_rate": 9.968916517362533e-06,
|
| 7528 |
+
"loss": 23.9313,
|
| 7529 |
+
"step": 10720
|
| 7530 |
+
},
|
| 7531 |
+
{
|
| 7532 |
+
"epoch": 0.1991199196468544,
|
| 7533 |
+
"grad_norm": 34.59375,
|
| 7534 |
+
"learning_rate": 9.96888752157649e-06,
|
| 7535 |
+
"loss": 24.0837,
|
| 7536 |
+
"step": 10730
|
| 7537 |
+
},
|
| 7538 |
+
{
|
| 7539 |
+
"epoch": 0.19930549273133424,
|
| 7540 |
+
"grad_norm": 35.1875,
|
| 7541 |
+
"learning_rate": 9.968858525790448e-06,
|
| 7542 |
+
"loss": 23.7037,
|
| 7543 |
+
"step": 10740
|
| 7544 |
+
},
|
| 7545 |
+
{
|
| 7546 |
+
"epoch": 0.19949106581581408,
|
| 7547 |
+
"grad_norm": 33.5625,
|
| 7548 |
+
"learning_rate": 9.968829530004407e-06,
|
| 7549 |
+
"loss": 23.7664,
|
| 7550 |
+
"step": 10750
|
| 7551 |
+
},
|
| 7552 |
+
{
|
| 7553 |
+
"epoch": 0.1996766389002939,
|
| 7554 |
+
"grad_norm": 33.9375,
|
| 7555 |
+
"learning_rate": 9.968800534218362e-06,
|
| 7556 |
+
"loss": 23.1647,
|
| 7557 |
+
"step": 10760
|
| 7558 |
+
},
|
| 7559 |
+
{
|
| 7560 |
+
"epoch": 0.19986221198477372,
|
| 7561 |
+
"grad_norm": 33.8125,
|
| 7562 |
+
"learning_rate": 9.96877153843232e-06,
|
| 7563 |
+
"loss": 23.4258,
|
| 7564 |
+
"step": 10770
|
| 7565 |
+
},
|
| 7566 |
+
{
|
| 7567 |
+
"epoch": 0.20004778506925355,
|
| 7568 |
+
"grad_norm": 34.09375,
|
| 7569 |
+
"learning_rate": 9.968742542646279e-06,
|
| 7570 |
+
"loss": 23.2949,
|
| 7571 |
+
"step": 10780
|
| 7572 |
+
},
|
| 7573 |
+
{
|
| 7574 |
+
"epoch": 0.2002333581537334,
|
| 7575 |
+
"grad_norm": 35.25,
|
| 7576 |
+
"learning_rate": 9.968713546860236e-06,
|
| 7577 |
+
"loss": 23.2747,
|
| 7578 |
+
"step": 10790
|
| 7579 |
+
},
|
| 7580 |
+
{
|
| 7581 |
+
"epoch": 0.20041893123821322,
|
| 7582 |
+
"grad_norm": 34.4375,
|
| 7583 |
+
"learning_rate": 9.968684551074194e-06,
|
| 7584 |
+
"loss": 23.3498,
|
| 7585 |
+
"step": 10800
|
| 7586 |
+
},
|
| 7587 |
+
{
|
| 7588 |
+
"epoch": 0.20060450432269303,
|
| 7589 |
+
"grad_norm": 34.6875,
|
| 7590 |
+
"learning_rate": 9.968655555288151e-06,
|
| 7591 |
+
"loss": 23.1578,
|
| 7592 |
+
"step": 10810
|
| 7593 |
+
},
|
| 7594 |
+
{
|
| 7595 |
+
"epoch": 0.20079007740717286,
|
| 7596 |
+
"grad_norm": 35.1875,
|
| 7597 |
+
"learning_rate": 9.968626559502108e-06,
|
| 7598 |
+
"loss": 23.6403,
|
| 7599 |
+
"step": 10820
|
| 7600 |
+
},
|
| 7601 |
+
{
|
| 7602 |
+
"epoch": 0.2009756504916527,
|
| 7603 |
+
"grad_norm": 38.40625,
|
| 7604 |
+
"learning_rate": 9.968597563716066e-06,
|
| 7605 |
+
"loss": 23.2563,
|
| 7606 |
+
"step": 10830
|
| 7607 |
+
},
|
| 7608 |
+
{
|
| 7609 |
+
"epoch": 0.20116122357613253,
|
| 7610 |
+
"grad_norm": 33.375,
|
| 7611 |
+
"learning_rate": 9.968568567930023e-06,
|
| 7612 |
+
"loss": 23.6925,
|
| 7613 |
+
"step": 10840
|
| 7614 |
+
},
|
| 7615 |
+
{
|
| 7616 |
+
"epoch": 0.20134679666061234,
|
| 7617 |
+
"grad_norm": 36.0,
|
| 7618 |
+
"learning_rate": 9.96853957214398e-06,
|
| 7619 |
+
"loss": 23.2151,
|
| 7620 |
+
"step": 10850
|
| 7621 |
+
},
|
| 7622 |
+
{
|
| 7623 |
+
"epoch": 0.20153236974509217,
|
| 7624 |
+
"grad_norm": 34.09375,
|
| 7625 |
+
"learning_rate": 9.968510576357938e-06,
|
| 7626 |
+
"loss": 23.1649,
|
| 7627 |
+
"step": 10860
|
| 7628 |
+
},
|
| 7629 |
+
{
|
| 7630 |
+
"epoch": 0.201717942829572,
|
| 7631 |
+
"grad_norm": 35.3125,
|
| 7632 |
+
"learning_rate": 9.968481580571896e-06,
|
| 7633 |
+
"loss": 23.6179,
|
| 7634 |
+
"step": 10870
|
| 7635 |
+
},
|
| 7636 |
+
{
|
| 7637 |
+
"epoch": 0.2019035159140518,
|
| 7638 |
+
"grad_norm": 32.59375,
|
| 7639 |
+
"learning_rate": 9.968452584785855e-06,
|
| 7640 |
+
"loss": 23.1651,
|
| 7641 |
+
"step": 10880
|
| 7642 |
+
},
|
| 7643 |
+
{
|
| 7644 |
+
"epoch": 0.20208908899853165,
|
| 7645 |
+
"grad_norm": 35.21875,
|
| 7646 |
+
"learning_rate": 9.968423588999812e-06,
|
| 7647 |
+
"loss": 23.2298,
|
| 7648 |
+
"step": 10890
|
| 7649 |
+
},
|
| 7650 |
+
{
|
| 7651 |
+
"epoch": 0.20227466208301148,
|
| 7652 |
+
"grad_norm": 36.4375,
|
| 7653 |
+
"learning_rate": 9.96839459321377e-06,
|
| 7654 |
+
"loss": 23.0713,
|
| 7655 |
+
"step": 10900
|
| 7656 |
+
},
|
| 7657 |
+
{
|
| 7658 |
+
"epoch": 0.20246023516749131,
|
| 7659 |
+
"grad_norm": 34.8125,
|
| 7660 |
+
"learning_rate": 9.968365597427727e-06,
|
| 7661 |
+
"loss": 23.2793,
|
| 7662 |
+
"step": 10910
|
| 7663 |
+
},
|
| 7664 |
+
{
|
| 7665 |
+
"epoch": 0.20264580825197112,
|
| 7666 |
+
"grad_norm": 34.53125,
|
| 7667 |
+
"learning_rate": 9.968336601641684e-06,
|
| 7668 |
+
"loss": 23.8128,
|
| 7669 |
+
"step": 10920
|
| 7670 |
+
},
|
| 7671 |
+
{
|
| 7672 |
+
"epoch": 0.20283138133645096,
|
| 7673 |
+
"grad_norm": 35.59375,
|
| 7674 |
+
"learning_rate": 9.968307605855642e-06,
|
| 7675 |
+
"loss": 23.302,
|
| 7676 |
+
"step": 10930
|
| 7677 |
+
},
|
| 7678 |
+
{
|
| 7679 |
+
"epoch": 0.2030169544209308,
|
| 7680 |
+
"grad_norm": 34.5,
|
| 7681 |
+
"learning_rate": 9.968278610069599e-06,
|
| 7682 |
+
"loss": 23.4782,
|
| 7683 |
+
"step": 10940
|
| 7684 |
+
},
|
| 7685 |
+
{
|
| 7686 |
+
"epoch": 0.20320252750541062,
|
| 7687 |
+
"grad_norm": 35.875,
|
| 7688 |
+
"learning_rate": 9.968249614283556e-06,
|
| 7689 |
+
"loss": 23.7409,
|
| 7690 |
+
"step": 10950
|
| 7691 |
+
},
|
| 7692 |
+
{
|
| 7693 |
+
"epoch": 0.20338810058989043,
|
| 7694 |
+
"grad_norm": 37.84375,
|
| 7695 |
+
"learning_rate": 9.968220618497516e-06,
|
| 7696 |
+
"loss": 23.5444,
|
| 7697 |
+
"step": 10960
|
| 7698 |
+
},
|
| 7699 |
+
{
|
| 7700 |
+
"epoch": 0.20357367367437026,
|
| 7701 |
+
"grad_norm": 31.90625,
|
| 7702 |
+
"learning_rate": 9.968191622711471e-06,
|
| 7703 |
+
"loss": 23.3963,
|
| 7704 |
+
"step": 10970
|
| 7705 |
+
},
|
| 7706 |
+
{
|
| 7707 |
+
"epoch": 0.2037592467588501,
|
| 7708 |
+
"grad_norm": 34.125,
|
| 7709 |
+
"learning_rate": 9.968162626925429e-06,
|
| 7710 |
+
"loss": 22.8245,
|
| 7711 |
+
"step": 10980
|
| 7712 |
+
},
|
| 7713 |
+
{
|
| 7714 |
+
"epoch": 0.20394481984332993,
|
| 7715 |
+
"grad_norm": 33.375,
|
| 7716 |
+
"learning_rate": 9.968133631139388e-06,
|
| 7717 |
+
"loss": 23.3189,
|
| 7718 |
+
"step": 10990
|
| 7719 |
+
},
|
| 7720 |
+
{
|
| 7721 |
+
"epoch": 0.20413039292780974,
|
| 7722 |
+
"grad_norm": 35.28125,
|
| 7723 |
+
"learning_rate": 9.968104635353345e-06,
|
| 7724 |
+
"loss": 23.5008,
|
| 7725 |
+
"step": 11000
|
| 7726 |
+
},
|
| 7727 |
+
{
|
| 7728 |
+
"epoch": 0.20431596601228957,
|
| 7729 |
+
"grad_norm": 35.75,
|
| 7730 |
+
"learning_rate": 9.968075639567303e-06,
|
| 7731 |
+
"loss": 23.534,
|
| 7732 |
+
"step": 11010
|
| 7733 |
+
},
|
| 7734 |
+
{
|
| 7735 |
+
"epoch": 0.2045015390967694,
|
| 7736 |
+
"grad_norm": 32.78125,
|
| 7737 |
+
"learning_rate": 9.96804664378126e-06,
|
| 7738 |
+
"loss": 23.4768,
|
| 7739 |
+
"step": 11020
|
| 7740 |
+
},
|
| 7741 |
+
{
|
| 7742 |
+
"epoch": 0.20468711218124924,
|
| 7743 |
+
"grad_norm": 36.46875,
|
| 7744 |
+
"learning_rate": 9.968017647995217e-06,
|
| 7745 |
+
"loss": 23.6789,
|
| 7746 |
+
"step": 11030
|
| 7747 |
+
},
|
| 7748 |
+
{
|
| 7749 |
+
"epoch": 0.20487268526572905,
|
| 7750 |
+
"grad_norm": 32.9375,
|
| 7751 |
+
"learning_rate": 9.967988652209175e-06,
|
| 7752 |
+
"loss": 23.4918,
|
| 7753 |
+
"step": 11040
|
| 7754 |
+
},
|
| 7755 |
+
{
|
| 7756 |
+
"epoch": 0.20505825835020888,
|
| 7757 |
+
"grad_norm": 35.03125,
|
| 7758 |
+
"learning_rate": 9.967959656423132e-06,
|
| 7759 |
+
"loss": 22.996,
|
| 7760 |
+
"step": 11050
|
| 7761 |
+
},
|
| 7762 |
+
{
|
| 7763 |
+
"epoch": 0.20524383143468872,
|
| 7764 |
+
"grad_norm": 34.53125,
|
| 7765 |
+
"learning_rate": 9.967930660637091e-06,
|
| 7766 |
+
"loss": 23.3215,
|
| 7767 |
+
"step": 11060
|
| 7768 |
+
},
|
| 7769 |
+
{
|
| 7770 |
+
"epoch": 0.20542940451916855,
|
| 7771 |
+
"grad_norm": 35.0,
|
| 7772 |
+
"learning_rate": 9.967901664851049e-06,
|
| 7773 |
+
"loss": 22.972,
|
| 7774 |
+
"step": 11070
|
| 7775 |
+
},
|
| 7776 |
+
{
|
| 7777 |
+
"epoch": 0.20561497760364836,
|
| 7778 |
+
"grad_norm": 36.0,
|
| 7779 |
+
"learning_rate": 9.967872669065004e-06,
|
| 7780 |
+
"loss": 22.7126,
|
| 7781 |
+
"step": 11080
|
| 7782 |
+
},
|
| 7783 |
+
{
|
| 7784 |
+
"epoch": 0.2058005506881282,
|
| 7785 |
+
"grad_norm": 35.875,
|
| 7786 |
+
"learning_rate": 9.967843673278964e-06,
|
| 7787 |
+
"loss": 23.8109,
|
| 7788 |
+
"step": 11090
|
| 7789 |
+
},
|
| 7790 |
+
{
|
| 7791 |
+
"epoch": 0.20598612377260803,
|
| 7792 |
+
"grad_norm": 32.96875,
|
| 7793 |
+
"learning_rate": 9.967814677492921e-06,
|
| 7794 |
+
"loss": 23.2342,
|
| 7795 |
+
"step": 11100
|
| 7796 |
+
},
|
| 7797 |
+
{
|
| 7798 |
+
"epoch": 0.20617169685708786,
|
| 7799 |
+
"grad_norm": 37.65625,
|
| 7800 |
+
"learning_rate": 9.967785681706878e-06,
|
| 7801 |
+
"loss": 23.8215,
|
| 7802 |
+
"step": 11110
|
| 7803 |
+
},
|
| 7804 |
+
{
|
| 7805 |
+
"epoch": 0.20635726994156767,
|
| 7806 |
+
"grad_norm": 37.1875,
|
| 7807 |
+
"learning_rate": 9.967756685920836e-06,
|
| 7808 |
+
"loss": 23.4363,
|
| 7809 |
+
"step": 11120
|
| 7810 |
+
},
|
| 7811 |
+
{
|
| 7812 |
+
"epoch": 0.2065428430260475,
|
| 7813 |
+
"grad_norm": 34.5,
|
| 7814 |
+
"learning_rate": 9.967727690134793e-06,
|
| 7815 |
+
"loss": 23.4552,
|
| 7816 |
+
"step": 11130
|
| 7817 |
+
},
|
| 7818 |
+
{
|
| 7819 |
+
"epoch": 0.20672841611052734,
|
| 7820 |
+
"grad_norm": 34.125,
|
| 7821 |
+
"learning_rate": 9.96769869434875e-06,
|
| 7822 |
+
"loss": 23.5466,
|
| 7823 |
+
"step": 11140
|
| 7824 |
+
},
|
| 7825 |
+
{
|
| 7826 |
+
"epoch": 0.20691398919500717,
|
| 7827 |
+
"grad_norm": 34.5625,
|
| 7828 |
+
"learning_rate": 9.967669698562708e-06,
|
| 7829 |
+
"loss": 23.4444,
|
| 7830 |
+
"step": 11150
|
| 7831 |
+
},
|
| 7832 |
+
{
|
| 7833 |
+
"epoch": 0.20709956227948698,
|
| 7834 |
+
"grad_norm": 35.28125,
|
| 7835 |
+
"learning_rate": 9.967640702776667e-06,
|
| 7836 |
+
"loss": 23.0774,
|
| 7837 |
+
"step": 11160
|
| 7838 |
+
},
|
| 7839 |
+
{
|
| 7840 |
+
"epoch": 0.2072851353639668,
|
| 7841 |
+
"grad_norm": 35.53125,
|
| 7842 |
+
"learning_rate": 9.967611706990624e-06,
|
| 7843 |
+
"loss": 23.1711,
|
| 7844 |
+
"step": 11170
|
| 7845 |
+
},
|
| 7846 |
+
{
|
| 7847 |
+
"epoch": 0.20747070844844664,
|
| 7848 |
+
"grad_norm": 33.75,
|
| 7849 |
+
"learning_rate": 9.967582711204582e-06,
|
| 7850 |
+
"loss": 23.2173,
|
| 7851 |
+
"step": 11180
|
| 7852 |
+
},
|
| 7853 |
+
{
|
| 7854 |
+
"epoch": 0.20765628153292645,
|
| 7855 |
+
"grad_norm": 35.84375,
|
| 7856 |
+
"learning_rate": 9.96755371541854e-06,
|
| 7857 |
+
"loss": 23.6226,
|
| 7858 |
+
"step": 11190
|
| 7859 |
+
},
|
| 7860 |
+
{
|
| 7861 |
+
"epoch": 0.20784185461740629,
|
| 7862 |
+
"grad_norm": 34.96875,
|
| 7863 |
+
"learning_rate": 9.967524719632497e-06,
|
| 7864 |
+
"loss": 22.9892,
|
| 7865 |
+
"step": 11200
|
| 7866 |
+
},
|
| 7867 |
+
{
|
| 7868 |
+
"epoch": 0.20802742770188612,
|
| 7869 |
+
"grad_norm": 34.0625,
|
| 7870 |
+
"learning_rate": 9.967495723846454e-06,
|
| 7871 |
+
"loss": 23.2622,
|
| 7872 |
+
"step": 11210
|
| 7873 |
+
},
|
| 7874 |
+
{
|
| 7875 |
+
"epoch": 0.20821300078636595,
|
| 7876 |
+
"grad_norm": 34.03125,
|
| 7877 |
+
"learning_rate": 9.967466728060412e-06,
|
| 7878 |
+
"loss": 22.9742,
|
| 7879 |
+
"step": 11220
|
| 7880 |
+
},
|
| 7881 |
+
{
|
| 7882 |
+
"epoch": 0.20839857387084576,
|
| 7883 |
+
"grad_norm": 34.84375,
|
| 7884 |
+
"learning_rate": 9.96743773227437e-06,
|
| 7885 |
+
"loss": 23.3746,
|
| 7886 |
+
"step": 11230
|
| 7887 |
+
},
|
| 7888 |
+
{
|
| 7889 |
+
"epoch": 0.2085841469553256,
|
| 7890 |
+
"grad_norm": 33.71875,
|
| 7891 |
+
"learning_rate": 9.967408736488326e-06,
|
| 7892 |
+
"loss": 23.0632,
|
| 7893 |
+
"step": 11240
|
| 7894 |
+
},
|
| 7895 |
+
{
|
| 7896 |
+
"epoch": 0.20876972003980543,
|
| 7897 |
+
"grad_norm": 36.875,
|
| 7898 |
+
"learning_rate": 9.967379740702284e-06,
|
| 7899 |
+
"loss": 23.3394,
|
| 7900 |
+
"step": 11250
|
| 7901 |
+
},
|
| 7902 |
+
{
|
| 7903 |
+
"epoch": 0.20895529312428526,
|
| 7904 |
+
"grad_norm": 34.65625,
|
| 7905 |
+
"learning_rate": 9.967350744916243e-06,
|
| 7906 |
+
"loss": 23.298,
|
| 7907 |
+
"step": 11260
|
| 7908 |
+
},
|
| 7909 |
+
{
|
| 7910 |
+
"epoch": 0.20914086620876507,
|
| 7911 |
+
"grad_norm": 36.0,
|
| 7912 |
+
"learning_rate": 9.9673217491302e-06,
|
| 7913 |
+
"loss": 23.015,
|
| 7914 |
+
"step": 11270
|
| 7915 |
+
},
|
| 7916 |
+
{
|
| 7917 |
+
"epoch": 0.2093264392932449,
|
| 7918 |
+
"grad_norm": 35.34375,
|
| 7919 |
+
"learning_rate": 9.967292753344158e-06,
|
| 7920 |
+
"loss": 23.2773,
|
| 7921 |
+
"step": 11280
|
| 7922 |
+
},
|
| 7923 |
+
{
|
| 7924 |
+
"epoch": 0.20951201237772474,
|
| 7925 |
+
"grad_norm": 34.65625,
|
| 7926 |
+
"learning_rate": 9.967263757558115e-06,
|
| 7927 |
+
"loss": 23.3175,
|
| 7928 |
+
"step": 11290
|
| 7929 |
+
},
|
| 7930 |
+
{
|
| 7931 |
+
"epoch": 0.20969758546220457,
|
| 7932 |
+
"grad_norm": 35.96875,
|
| 7933 |
+
"learning_rate": 9.967234761772072e-06,
|
| 7934 |
+
"loss": 23.3183,
|
| 7935 |
+
"step": 11300
|
| 7936 |
+
},
|
| 7937 |
+
{
|
| 7938 |
+
"epoch": 0.20988315854668438,
|
| 7939 |
+
"grad_norm": 33.53125,
|
| 7940 |
+
"learning_rate": 9.96720576598603e-06,
|
| 7941 |
+
"loss": 23.4582,
|
| 7942 |
+
"step": 11310
|
| 7943 |
+
},
|
| 7944 |
+
{
|
| 7945 |
+
"epoch": 0.2100687316311642,
|
| 7946 |
+
"grad_norm": 34.78125,
|
| 7947 |
+
"learning_rate": 9.967176770199987e-06,
|
| 7948 |
+
"loss": 23.4951,
|
| 7949 |
+
"step": 11320
|
| 7950 |
+
},
|
| 7951 |
+
{
|
| 7952 |
+
"epoch": 0.21025430471564405,
|
| 7953 |
+
"grad_norm": 35.09375,
|
| 7954 |
+
"learning_rate": 9.967147774413945e-06,
|
| 7955 |
+
"loss": 23.4455,
|
| 7956 |
+
"step": 11330
|
| 7957 |
+
},
|
| 7958 |
+
{
|
| 7959 |
+
"epoch": 0.21043987780012388,
|
| 7960 |
+
"grad_norm": 34.59375,
|
| 7961 |
+
"learning_rate": 9.967118778627904e-06,
|
| 7962 |
+
"loss": 22.9397,
|
| 7963 |
+
"step": 11340
|
| 7964 |
+
},
|
| 7965 |
+
{
|
| 7966 |
+
"epoch": 0.2106254508846037,
|
| 7967 |
+
"grad_norm": 37.4375,
|
| 7968 |
+
"learning_rate": 9.96708978284186e-06,
|
| 7969 |
+
"loss": 23.5124,
|
| 7970 |
+
"step": 11350
|
| 7971 |
+
},
|
| 7972 |
+
{
|
| 7973 |
+
"epoch": 0.21081102396908352,
|
| 7974 |
+
"grad_norm": 33.75,
|
| 7975 |
+
"learning_rate": 9.967060787055819e-06,
|
| 7976 |
+
"loss": 23.2777,
|
| 7977 |
+
"step": 11360
|
| 7978 |
+
},
|
| 7979 |
+
{
|
| 7980 |
+
"epoch": 0.21099659705356336,
|
| 7981 |
+
"grad_norm": 36.375,
|
| 7982 |
+
"learning_rate": 9.967031791269776e-06,
|
| 7983 |
+
"loss": 23.5096,
|
| 7984 |
+
"step": 11370
|
| 7985 |
+
},
|
| 7986 |
+
{
|
| 7987 |
+
"epoch": 0.2111821701380432,
|
| 7988 |
+
"grad_norm": 35.40625,
|
| 7989 |
+
"learning_rate": 9.967002795483733e-06,
|
| 7990 |
+
"loss": 23.5407,
|
| 7991 |
+
"step": 11380
|
| 7992 |
+
},
|
| 7993 |
+
{
|
| 7994 |
+
"epoch": 0.211367743222523,
|
| 7995 |
+
"grad_norm": 35.375,
|
| 7996 |
+
"learning_rate": 9.96697379969769e-06,
|
| 7997 |
+
"loss": 23.0631,
|
| 7998 |
+
"step": 11390
|
| 7999 |
+
},
|
| 8000 |
+
{
|
| 8001 |
+
"epoch": 0.21155331630700283,
|
| 8002 |
+
"grad_norm": 33.71875,
|
| 8003 |
+
"learning_rate": 9.966944803911648e-06,
|
| 8004 |
+
"loss": 23.4863,
|
| 8005 |
+
"step": 11400
|
| 8006 |
+
},
|
| 8007 |
+
{
|
| 8008 |
+
"epoch": 0.21173888939148267,
|
| 8009 |
+
"grad_norm": 33.65625,
|
| 8010 |
+
"learning_rate": 9.966915808125606e-06,
|
| 8011 |
+
"loss": 22.7694,
|
| 8012 |
+
"step": 11410
|
| 8013 |
+
},
|
| 8014 |
+
{
|
| 8015 |
+
"epoch": 0.2119244624759625,
|
| 8016 |
+
"grad_norm": 33.84375,
|
| 8017 |
+
"learning_rate": 9.966886812339563e-06,
|
| 8018 |
+
"loss": 23.4398,
|
| 8019 |
+
"step": 11420
|
| 8020 |
+
},
|
| 8021 |
+
{
|
| 8022 |
+
"epoch": 0.2121100355604423,
|
| 8023 |
+
"grad_norm": 36.125,
|
| 8024 |
+
"learning_rate": 9.96685781655352e-06,
|
| 8025 |
+
"loss": 23.2906,
|
| 8026 |
+
"step": 11430
|
| 8027 |
+
},
|
| 8028 |
+
{
|
| 8029 |
+
"epoch": 0.21229560864492214,
|
| 8030 |
+
"grad_norm": 33.3125,
|
| 8031 |
+
"learning_rate": 9.96682882076748e-06,
|
| 8032 |
+
"loss": 23.335,
|
| 8033 |
+
"step": 11440
|
| 8034 |
+
},
|
| 8035 |
+
{
|
| 8036 |
+
"epoch": 0.21248118172940197,
|
| 8037 |
+
"grad_norm": 33.4375,
|
| 8038 |
+
"learning_rate": 9.966799824981435e-06,
|
| 8039 |
+
"loss": 23.3474,
|
| 8040 |
+
"step": 11450
|
| 8041 |
+
},
|
| 8042 |
+
{
|
| 8043 |
+
"epoch": 0.21266675481388178,
|
| 8044 |
+
"grad_norm": 35.6875,
|
| 8045 |
+
"learning_rate": 9.966770829195393e-06,
|
| 8046 |
+
"loss": 23.5648,
|
| 8047 |
+
"step": 11460
|
| 8048 |
+
},
|
| 8049 |
+
{
|
| 8050 |
+
"epoch": 0.21285232789836162,
|
| 8051 |
+
"grad_norm": 36.21875,
|
| 8052 |
+
"learning_rate": 9.966741833409352e-06,
|
| 8053 |
+
"loss": 23.1292,
|
| 8054 |
+
"step": 11470
|
| 8055 |
+
},
|
| 8056 |
+
{
|
| 8057 |
+
"epoch": 0.21303790098284145,
|
| 8058 |
+
"grad_norm": 36.78125,
|
| 8059 |
+
"learning_rate": 9.96671283762331e-06,
|
| 8060 |
+
"loss": 23.2989,
|
| 8061 |
+
"step": 11480
|
| 8062 |
+
},
|
| 8063 |
+
{
|
| 8064 |
+
"epoch": 0.21322347406732128,
|
| 8065 |
+
"grad_norm": 35.59375,
|
| 8066 |
+
"learning_rate": 9.966683841837267e-06,
|
| 8067 |
+
"loss": 23.4273,
|
| 8068 |
+
"step": 11490
|
| 8069 |
+
},
|
| 8070 |
+
{
|
| 8071 |
+
"epoch": 0.2134090471518011,
|
| 8072 |
+
"grad_norm": 35.71875,
|
| 8073 |
+
"learning_rate": 9.966654846051224e-06,
|
| 8074 |
+
"loss": 23.0285,
|
| 8075 |
+
"step": 11500
|
| 8076 |
+
},
|
| 8077 |
+
{
|
| 8078 |
+
"epoch": 0.21359462023628092,
|
| 8079 |
+
"grad_norm": 34.375,
|
| 8080 |
+
"learning_rate": 9.966625850265181e-06,
|
| 8081 |
+
"loss": 23.4227,
|
| 8082 |
+
"step": 11510
|
| 8083 |
+
},
|
| 8084 |
+
{
|
| 8085 |
+
"epoch": 0.21378019332076076,
|
| 8086 |
+
"grad_norm": 33.59375,
|
| 8087 |
+
"learning_rate": 9.966596854479139e-06,
|
| 8088 |
+
"loss": 23.3722,
|
| 8089 |
+
"step": 11520
|
| 8090 |
+
},
|
| 8091 |
+
{
|
| 8092 |
+
"epoch": 0.2139657664052406,
|
| 8093 |
+
"grad_norm": 34.65625,
|
| 8094 |
+
"learning_rate": 9.966567858693096e-06,
|
| 8095 |
+
"loss": 23.19,
|
| 8096 |
+
"step": 11530
|
| 8097 |
+
},
|
| 8098 |
+
{
|
| 8099 |
+
"epoch": 0.2141513394897204,
|
| 8100 |
+
"grad_norm": 35.375,
|
| 8101 |
+
"learning_rate": 9.966538862907055e-06,
|
| 8102 |
+
"loss": 23.5036,
|
| 8103 |
+
"step": 11540
|
| 8104 |
+
},
|
| 8105 |
+
{
|
| 8106 |
+
"epoch": 0.21433691257420023,
|
| 8107 |
+
"grad_norm": 34.3125,
|
| 8108 |
+
"learning_rate": 9.966509867121013e-06,
|
| 8109 |
+
"loss": 23.3387,
|
| 8110 |
+
"step": 11550
|
| 8111 |
+
},
|
| 8112 |
+
{
|
| 8113 |
+
"epoch": 0.21452248565868007,
|
| 8114 |
+
"grad_norm": 34.0625,
|
| 8115 |
+
"learning_rate": 9.966480871334968e-06,
|
| 8116 |
+
"loss": 23.121,
|
| 8117 |
+
"step": 11560
|
| 8118 |
+
},
|
| 8119 |
+
{
|
| 8120 |
+
"epoch": 0.2147080587431599,
|
| 8121 |
+
"grad_norm": 35.71875,
|
| 8122 |
+
"learning_rate": 9.966451875548928e-06,
|
| 8123 |
+
"loss": 23.4267,
|
| 8124 |
+
"step": 11570
|
| 8125 |
+
},
|
| 8126 |
+
{
|
| 8127 |
+
"epoch": 0.2148936318276397,
|
| 8128 |
+
"grad_norm": 36.71875,
|
| 8129 |
+
"learning_rate": 9.966422879762885e-06,
|
| 8130 |
+
"loss": 23.2037,
|
| 8131 |
+
"step": 11580
|
| 8132 |
+
},
|
| 8133 |
+
{
|
| 8134 |
+
"epoch": 0.21507920491211954,
|
| 8135 |
+
"grad_norm": 34.3125,
|
| 8136 |
+
"learning_rate": 9.966393883976842e-06,
|
| 8137 |
+
"loss": 23.6055,
|
| 8138 |
+
"step": 11590
|
| 8139 |
+
},
|
| 8140 |
+
{
|
| 8141 |
+
"epoch": 0.21526477799659938,
|
| 8142 |
+
"grad_norm": 33.5,
|
| 8143 |
+
"learning_rate": 9.9663648881908e-06,
|
| 8144 |
+
"loss": 23.0454,
|
| 8145 |
+
"step": 11600
|
| 8146 |
+
},
|
| 8147 |
+
{
|
| 8148 |
+
"epoch": 0.2154503510810792,
|
| 8149 |
+
"grad_norm": 35.0625,
|
| 8150 |
+
"learning_rate": 9.966335892404759e-06,
|
| 8151 |
+
"loss": 22.7495,
|
| 8152 |
+
"step": 11610
|
| 8153 |
+
},
|
| 8154 |
+
{
|
| 8155 |
+
"epoch": 0.21563592416555902,
|
| 8156 |
+
"grad_norm": 34.9375,
|
| 8157 |
+
"learning_rate": 9.966306896618715e-06,
|
| 8158 |
+
"loss": 23.8042,
|
| 8159 |
+
"step": 11620
|
| 8160 |
+
},
|
| 8161 |
+
{
|
| 8162 |
+
"epoch": 0.21582149725003885,
|
| 8163 |
+
"grad_norm": 37.03125,
|
| 8164 |
+
"learning_rate": 9.966277900832672e-06,
|
| 8165 |
+
"loss": 23.4072,
|
| 8166 |
+
"step": 11630
|
| 8167 |
+
},
|
| 8168 |
+
{
|
| 8169 |
+
"epoch": 0.21600707033451869,
|
| 8170 |
+
"grad_norm": 37.09375,
|
| 8171 |
+
"learning_rate": 9.966248905046631e-06,
|
| 8172 |
+
"loss": 23.2861,
|
| 8173 |
+
"step": 11640
|
| 8174 |
+
},
|
| 8175 |
+
{
|
| 8176 |
+
"epoch": 0.21619264341899852,
|
| 8177 |
+
"grad_norm": 34.40625,
|
| 8178 |
+
"learning_rate": 9.966219909260589e-06,
|
| 8179 |
+
"loss": 23.3566,
|
| 8180 |
+
"step": 11650
|
| 8181 |
+
},
|
| 8182 |
+
{
|
| 8183 |
+
"epoch": 0.21637821650347833,
|
| 8184 |
+
"grad_norm": 33.875,
|
| 8185 |
+
"learning_rate": 9.966190913474546e-06,
|
| 8186 |
+
"loss": 22.9984,
|
| 8187 |
+
"step": 11660
|
| 8188 |
+
},
|
| 8189 |
+
{
|
| 8190 |
+
"epoch": 0.21656378958795816,
|
| 8191 |
+
"grad_norm": 34.09375,
|
| 8192 |
+
"learning_rate": 9.966161917688503e-06,
|
| 8193 |
+
"loss": 23.0596,
|
| 8194 |
+
"step": 11670
|
| 8195 |
+
},
|
| 8196 |
+
{
|
| 8197 |
+
"epoch": 0.216749362672438,
|
| 8198 |
+
"grad_norm": 35.46875,
|
| 8199 |
+
"learning_rate": 9.96613292190246e-06,
|
| 8200 |
+
"loss": 23.5079,
|
| 8201 |
+
"step": 11680
|
| 8202 |
+
},
|
| 8203 |
+
{
|
| 8204 |
+
"epoch": 0.21693493575691783,
|
| 8205 |
+
"grad_norm": 34.96875,
|
| 8206 |
+
"learning_rate": 9.966103926116418e-06,
|
| 8207 |
+
"loss": 22.9615,
|
| 8208 |
+
"step": 11690
|
| 8209 |
+
},
|
| 8210 |
+
{
|
| 8211 |
+
"epoch": 0.21712050884139764,
|
| 8212 |
+
"grad_norm": 35.65625,
|
| 8213 |
+
"learning_rate": 9.966074930330376e-06,
|
| 8214 |
+
"loss": 23.3229,
|
| 8215 |
+
"step": 11700
|
| 8216 |
+
},
|
| 8217 |
+
{
|
| 8218 |
+
"epoch": 0.21730608192587747,
|
| 8219 |
+
"grad_norm": 33.46875,
|
| 8220 |
+
"learning_rate": 9.966045934544335e-06,
|
| 8221 |
+
"loss": 23.2428,
|
| 8222 |
+
"step": 11710
|
| 8223 |
+
},
|
| 8224 |
+
{
|
| 8225 |
+
"epoch": 0.2174916550103573,
|
| 8226 |
+
"grad_norm": 33.625,
|
| 8227 |
+
"learning_rate": 9.96601693875829e-06,
|
| 8228 |
+
"loss": 23.3109,
|
| 8229 |
+
"step": 11720
|
| 8230 |
+
},
|
| 8231 |
+
{
|
| 8232 |
+
"epoch": 0.21767722809483714,
|
| 8233 |
+
"grad_norm": 35.34375,
|
| 8234 |
+
"learning_rate": 9.965987942972248e-06,
|
| 8235 |
+
"loss": 23.313,
|
| 8236 |
+
"step": 11730
|
| 8237 |
+
},
|
| 8238 |
+
{
|
| 8239 |
+
"epoch": 0.21786280117931694,
|
| 8240 |
+
"grad_norm": 34.8125,
|
| 8241 |
+
"learning_rate": 9.965958947186207e-06,
|
| 8242 |
+
"loss": 22.9677,
|
| 8243 |
+
"step": 11740
|
| 8244 |
+
},
|
| 8245 |
+
{
|
| 8246 |
+
"epoch": 0.21804837426379678,
|
| 8247 |
+
"grad_norm": 34.46875,
|
| 8248 |
+
"learning_rate": 9.965929951400164e-06,
|
| 8249 |
+
"loss": 23.4194,
|
| 8250 |
+
"step": 11750
|
| 8251 |
+
},
|
| 8252 |
+
{
|
| 8253 |
+
"epoch": 0.2182339473482766,
|
| 8254 |
+
"grad_norm": 35.875,
|
| 8255 |
+
"learning_rate": 9.965900955614122e-06,
|
| 8256 |
+
"loss": 23.2412,
|
| 8257 |
+
"step": 11760
|
| 8258 |
+
},
|
| 8259 |
+
{
|
| 8260 |
+
"epoch": 0.21841952043275642,
|
| 8261 |
+
"grad_norm": 34.96875,
|
| 8262 |
+
"learning_rate": 9.965871959828079e-06,
|
| 8263 |
+
"loss": 23.1838,
|
| 8264 |
+
"step": 11770
|
| 8265 |
+
},
|
| 8266 |
+
{
|
| 8267 |
+
"epoch": 0.21860509351723625,
|
| 8268 |
+
"grad_norm": 36.4375,
|
| 8269 |
+
"learning_rate": 9.965842964042036e-06,
|
| 8270 |
+
"loss": 23.4135,
|
| 8271 |
+
"step": 11780
|
| 8272 |
+
},
|
| 8273 |
+
{
|
| 8274 |
+
"epoch": 0.2187906666017161,
|
| 8275 |
+
"grad_norm": 35.21875,
|
| 8276 |
+
"learning_rate": 9.965813968255994e-06,
|
| 8277 |
+
"loss": 23.0506,
|
| 8278 |
+
"step": 11790
|
| 8279 |
+
},
|
| 8280 |
+
{
|
| 8281 |
+
"epoch": 0.21897623968619592,
|
| 8282 |
+
"grad_norm": 34.65625,
|
| 8283 |
+
"learning_rate": 9.965784972469951e-06,
|
| 8284 |
+
"loss": 23.2738,
|
| 8285 |
+
"step": 11800
|
| 8286 |
+
},
|
| 8287 |
+
{
|
| 8288 |
+
"epoch": 0.21916181277067573,
|
| 8289 |
+
"grad_norm": 36.21875,
|
| 8290 |
+
"learning_rate": 9.96575597668391e-06,
|
| 8291 |
+
"loss": 23.1185,
|
| 8292 |
+
"step": 11810
|
| 8293 |
+
},
|
| 8294 |
+
{
|
| 8295 |
+
"epoch": 0.21934738585515556,
|
| 8296 |
+
"grad_norm": 36.46875,
|
| 8297 |
+
"learning_rate": 9.965726980897868e-06,
|
| 8298 |
+
"loss": 23.2145,
|
| 8299 |
+
"step": 11820
|
| 8300 |
+
},
|
| 8301 |
+
{
|
| 8302 |
+
"epoch": 0.2195329589396354,
|
| 8303 |
+
"grad_norm": 33.96875,
|
| 8304 |
+
"learning_rate": 9.965697985111824e-06,
|
| 8305 |
+
"loss": 23.4867,
|
| 8306 |
+
"step": 11830
|
| 8307 |
+
},
|
| 8308 |
+
{
|
| 8309 |
+
"epoch": 0.21971853202411523,
|
| 8310 |
+
"grad_norm": 36.375,
|
| 8311 |
+
"learning_rate": 9.965668989325783e-06,
|
| 8312 |
+
"loss": 22.9346,
|
| 8313 |
+
"step": 11840
|
| 8314 |
+
},
|
| 8315 |
+
{
|
| 8316 |
+
"epoch": 0.21990410510859504,
|
| 8317 |
+
"grad_norm": 34.34375,
|
| 8318 |
+
"learning_rate": 9.96563999353974e-06,
|
| 8319 |
+
"loss": 23.1345,
|
| 8320 |
+
"step": 11850
|
| 8321 |
+
},
|
| 8322 |
+
{
|
| 8323 |
+
"epoch": 0.22008967819307487,
|
| 8324 |
+
"grad_norm": 36.34375,
|
| 8325 |
+
"learning_rate": 9.965610997753697e-06,
|
| 8326 |
+
"loss": 23.0301,
|
| 8327 |
+
"step": 11860
|
| 8328 |
+
},
|
| 8329 |
+
{
|
| 8330 |
+
"epoch": 0.2202752512775547,
|
| 8331 |
+
"grad_norm": 35.28125,
|
| 8332 |
+
"learning_rate": 9.965582001967655e-06,
|
| 8333 |
+
"loss": 22.9159,
|
| 8334 |
+
"step": 11870
|
| 8335 |
+
},
|
| 8336 |
+
{
|
| 8337 |
+
"epoch": 0.22046082436203454,
|
| 8338 |
+
"grad_norm": 33.9375,
|
| 8339 |
+
"learning_rate": 9.965553006181612e-06,
|
| 8340 |
+
"loss": 23.1306,
|
| 8341 |
+
"step": 11880
|
| 8342 |
+
},
|
| 8343 |
+
{
|
| 8344 |
+
"epoch": 0.22064639744651435,
|
| 8345 |
+
"grad_norm": 33.9375,
|
| 8346 |
+
"learning_rate": 9.96552401039557e-06,
|
| 8347 |
+
"loss": 23.0807,
|
| 8348 |
+
"step": 11890
|
| 8349 |
+
},
|
| 8350 |
+
{
|
| 8351 |
+
"epoch": 0.22083197053099418,
|
| 8352 |
+
"grad_norm": 37.5625,
|
| 8353 |
+
"learning_rate": 9.965495014609527e-06,
|
| 8354 |
+
"loss": 23.1038,
|
| 8355 |
+
"step": 11900
|
| 8356 |
+
},
|
| 8357 |
+
{
|
| 8358 |
+
"epoch": 0.22101754361547402,
|
| 8359 |
+
"grad_norm": 36.21875,
|
| 8360 |
+
"learning_rate": 9.965466018823484e-06,
|
| 8361 |
+
"loss": 23.5351,
|
| 8362 |
+
"step": 11910
|
| 8363 |
+
},
|
| 8364 |
+
{
|
| 8365 |
+
"epoch": 0.22120311669995385,
|
| 8366 |
+
"grad_norm": 35.59375,
|
| 8367 |
+
"learning_rate": 9.965437023037444e-06,
|
| 8368 |
+
"loss": 23.2319,
|
| 8369 |
+
"step": 11920
|
| 8370 |
+
},
|
| 8371 |
+
{
|
| 8372 |
+
"epoch": 0.22138868978443366,
|
| 8373 |
+
"grad_norm": 33.125,
|
| 8374 |
+
"learning_rate": 9.965408027251401e-06,
|
| 8375 |
+
"loss": 23.1569,
|
| 8376 |
+
"step": 11930
|
| 8377 |
+
},
|
| 8378 |
+
{
|
| 8379 |
+
"epoch": 0.2215742628689135,
|
| 8380 |
+
"grad_norm": 35.40625,
|
| 8381 |
+
"learning_rate": 9.965379031465358e-06,
|
| 8382 |
+
"loss": 23.5166,
|
| 8383 |
+
"step": 11940
|
| 8384 |
+
},
|
| 8385 |
+
{
|
| 8386 |
+
"epoch": 0.22175983595339332,
|
| 8387 |
+
"grad_norm": 35.96875,
|
| 8388 |
+
"learning_rate": 9.965350035679316e-06,
|
| 8389 |
+
"loss": 23.0607,
|
| 8390 |
+
"step": 11950
|
| 8391 |
+
},
|
| 8392 |
+
{
|
| 8393 |
+
"epoch": 0.22194540903787316,
|
| 8394 |
+
"grad_norm": 35.0,
|
| 8395 |
+
"learning_rate": 9.965321039893273e-06,
|
| 8396 |
+
"loss": 23.1148,
|
| 8397 |
+
"step": 11960
|
| 8398 |
+
},
|
| 8399 |
+
{
|
| 8400 |
+
"epoch": 0.22213098212235297,
|
| 8401 |
+
"grad_norm": 36.34375,
|
| 8402 |
+
"learning_rate": 9.96529204410723e-06,
|
| 8403 |
+
"loss": 22.8368,
|
| 8404 |
+
"step": 11970
|
| 8405 |
+
},
|
| 8406 |
+
{
|
| 8407 |
+
"epoch": 0.2223165552068328,
|
| 8408 |
+
"grad_norm": 35.40625,
|
| 8409 |
+
"learning_rate": 9.965263048321188e-06,
|
| 8410 |
+
"loss": 22.4958,
|
| 8411 |
+
"step": 11980
|
| 8412 |
+
},
|
| 8413 |
+
{
|
| 8414 |
+
"epoch": 0.22250212829131263,
|
| 8415 |
+
"grad_norm": 35.96875,
|
| 8416 |
+
"learning_rate": 9.965234052535145e-06,
|
| 8417 |
+
"loss": 22.882,
|
| 8418 |
+
"step": 11990
|
| 8419 |
+
},
|
| 8420 |
+
{
|
| 8421 |
+
"epoch": 0.22268770137579247,
|
| 8422 |
+
"grad_norm": 37.53125,
|
| 8423 |
+
"learning_rate": 9.965205056749103e-06,
|
| 8424 |
+
"loss": 23.2451,
|
| 8425 |
+
"step": 12000
|
| 8426 |
+
},
|
| 8427 |
+
{
|
| 8428 |
+
"epoch": 0.22287327446027227,
|
| 8429 |
+
"grad_norm": 36.03125,
|
| 8430 |
+
"learning_rate": 9.96517606096306e-06,
|
| 8431 |
+
"loss": 22.7075,
|
| 8432 |
+
"step": 12010
|
| 8433 |
+
},
|
| 8434 |
+
{
|
| 8435 |
+
"epoch": 0.2230588475447521,
|
| 8436 |
+
"grad_norm": 34.5,
|
| 8437 |
+
"learning_rate": 9.96514706517702e-06,
|
| 8438 |
+
"loss": 22.9387,
|
| 8439 |
+
"step": 12020
|
| 8440 |
+
},
|
| 8441 |
+
{
|
| 8442 |
+
"epoch": 0.22324442062923194,
|
| 8443 |
+
"grad_norm": 34.9375,
|
| 8444 |
+
"learning_rate": 9.965118069390977e-06,
|
| 8445 |
+
"loss": 23.5015,
|
| 8446 |
+
"step": 12030
|
| 8447 |
+
},
|
| 8448 |
+
{
|
| 8449 |
+
"epoch": 0.22342999371371175,
|
| 8450 |
+
"grad_norm": 36.375,
|
| 8451 |
+
"learning_rate": 9.965089073604932e-06,
|
| 8452 |
+
"loss": 23.2239,
|
| 8453 |
+
"step": 12040
|
| 8454 |
+
},
|
| 8455 |
+
{
|
| 8456 |
+
"epoch": 0.22361556679819158,
|
| 8457 |
+
"grad_norm": 33.78125,
|
| 8458 |
+
"learning_rate": 9.965060077818892e-06,
|
| 8459 |
+
"loss": 22.8462,
|
| 8460 |
+
"step": 12050
|
| 8461 |
+
},
|
| 8462 |
+
{
|
| 8463 |
+
"epoch": 0.22380113988267142,
|
| 8464 |
+
"grad_norm": 35.96875,
|
| 8465 |
+
"learning_rate": 9.965031082032849e-06,
|
| 8466 |
+
"loss": 23.1589,
|
| 8467 |
+
"step": 12060
|
| 8468 |
+
},
|
| 8469 |
+
{
|
| 8470 |
+
"epoch": 0.22398671296715125,
|
| 8471 |
+
"grad_norm": 33.875,
|
| 8472 |
+
"learning_rate": 9.965002086246806e-06,
|
| 8473 |
+
"loss": 23.3079,
|
| 8474 |
+
"step": 12070
|
| 8475 |
+
},
|
| 8476 |
+
{
|
| 8477 |
+
"epoch": 0.22417228605163106,
|
| 8478 |
+
"grad_norm": 35.625,
|
| 8479 |
+
"learning_rate": 9.964973090460764e-06,
|
| 8480 |
+
"loss": 22.9087,
|
| 8481 |
+
"step": 12080
|
| 8482 |
+
},
|
| 8483 |
+
{
|
| 8484 |
+
"epoch": 0.2243578591361109,
|
| 8485 |
+
"grad_norm": 33.1875,
|
| 8486 |
+
"learning_rate": 9.964944094674723e-06,
|
| 8487 |
+
"loss": 23.1929,
|
| 8488 |
+
"step": 12090
|
| 8489 |
+
},
|
| 8490 |
+
{
|
| 8491 |
+
"epoch": 0.22454343222059073,
|
| 8492 |
+
"grad_norm": 36.125,
|
| 8493 |
+
"learning_rate": 9.964915098888679e-06,
|
| 8494 |
+
"loss": 22.967,
|
| 8495 |
+
"step": 12100
|
| 8496 |
+
},
|
| 8497 |
+
{
|
| 8498 |
+
"epoch": 0.22472900530507056,
|
| 8499 |
+
"grad_norm": 35.03125,
|
| 8500 |
+
"learning_rate": 9.964886103102636e-06,
|
| 8501 |
+
"loss": 22.9228,
|
| 8502 |
+
"step": 12110
|
| 8503 |
+
},
|
| 8504 |
+
{
|
| 8505 |
+
"epoch": 0.22491457838955037,
|
| 8506 |
+
"grad_norm": 35.03125,
|
| 8507 |
+
"learning_rate": 9.964857107316595e-06,
|
| 8508 |
+
"loss": 22.9758,
|
| 8509 |
+
"step": 12120
|
| 8510 |
+
},
|
| 8511 |
+
{
|
| 8512 |
+
"epoch": 0.2251001514740302,
|
| 8513 |
+
"grad_norm": 34.0625,
|
| 8514 |
+
"learning_rate": 9.964828111530553e-06,
|
| 8515 |
+
"loss": 22.2338,
|
| 8516 |
+
"step": 12130
|
| 8517 |
+
},
|
| 8518 |
+
{
|
| 8519 |
+
"epoch": 0.22528572455851004,
|
| 8520 |
+
"grad_norm": 36.09375,
|
| 8521 |
+
"learning_rate": 9.96479911574451e-06,
|
| 8522 |
+
"loss": 23.1649,
|
| 8523 |
+
"step": 12140
|
| 8524 |
+
},
|
| 8525 |
+
{
|
| 8526 |
+
"epoch": 0.22547129764298987,
|
| 8527 |
+
"grad_norm": 35.5625,
|
| 8528 |
+
"learning_rate": 9.964770119958467e-06,
|
| 8529 |
+
"loss": 23.1733,
|
| 8530 |
+
"step": 12150
|
| 8531 |
+
},
|
| 8532 |
+
{
|
| 8533 |
+
"epoch": 0.22565687072746968,
|
| 8534 |
+
"grad_norm": 36.53125,
|
| 8535 |
+
"learning_rate": 9.964741124172425e-06,
|
| 8536 |
+
"loss": 22.8313,
|
| 8537 |
+
"step": 12160
|
| 8538 |
+
},
|
| 8539 |
+
{
|
| 8540 |
+
"epoch": 0.2258424438119495,
|
| 8541 |
+
"grad_norm": 36.40625,
|
| 8542 |
+
"learning_rate": 9.964712128386382e-06,
|
| 8543 |
+
"loss": 22.7984,
|
| 8544 |
+
"step": 12170
|
| 8545 |
+
},
|
| 8546 |
+
{
|
| 8547 |
+
"epoch": 0.22602801689642935,
|
| 8548 |
+
"grad_norm": 34.4375,
|
| 8549 |
+
"learning_rate": 9.96468313260034e-06,
|
| 8550 |
+
"loss": 23.1878,
|
| 8551 |
+
"step": 12180
|
| 8552 |
+
},
|
| 8553 |
+
{
|
| 8554 |
+
"epoch": 0.22621358998090918,
|
| 8555 |
+
"grad_norm": 36.25,
|
| 8556 |
+
"learning_rate": 9.964654136814299e-06,
|
| 8557 |
+
"loss": 23.0327,
|
| 8558 |
+
"step": 12190
|
| 8559 |
+
},
|
| 8560 |
+
{
|
| 8561 |
+
"epoch": 0.226399163065389,
|
| 8562 |
+
"grad_norm": 36.09375,
|
| 8563 |
+
"learning_rate": 9.964625141028256e-06,
|
| 8564 |
+
"loss": 22.7311,
|
| 8565 |
+
"step": 12200
|
| 8566 |
+
},
|
| 8567 |
+
{
|
| 8568 |
+
"epoch": 0.22658473614986882,
|
| 8569 |
+
"grad_norm": 34.15625,
|
| 8570 |
+
"learning_rate": 9.964596145242212e-06,
|
| 8571 |
+
"loss": 22.8199,
|
| 8572 |
+
"step": 12210
|
| 8573 |
+
},
|
| 8574 |
+
{
|
| 8575 |
+
"epoch": 0.22677030923434865,
|
| 8576 |
+
"grad_norm": 33.9375,
|
| 8577 |
+
"learning_rate": 9.964567149456171e-06,
|
| 8578 |
+
"loss": 22.6879,
|
| 8579 |
+
"step": 12220
|
| 8580 |
+
},
|
| 8581 |
+
{
|
| 8582 |
+
"epoch": 0.2269558823188285,
|
| 8583 |
+
"grad_norm": 35.3125,
|
| 8584 |
+
"learning_rate": 9.964538153670128e-06,
|
| 8585 |
+
"loss": 23.2458,
|
| 8586 |
+
"step": 12230
|
| 8587 |
+
},
|
| 8588 |
+
{
|
| 8589 |
+
"epoch": 0.2271414554033083,
|
| 8590 |
+
"grad_norm": 32.4375,
|
| 8591 |
+
"learning_rate": 9.964509157884086e-06,
|
| 8592 |
+
"loss": 22.7513,
|
| 8593 |
+
"step": 12240
|
| 8594 |
+
},
|
| 8595 |
+
{
|
| 8596 |
+
"epoch": 0.22732702848778813,
|
| 8597 |
+
"grad_norm": 34.5,
|
| 8598 |
+
"learning_rate": 9.964480162098043e-06,
|
| 8599 |
+
"loss": 23.2389,
|
| 8600 |
+
"step": 12250
|
| 8601 |
+
},
|
| 8602 |
+
{
|
| 8603 |
+
"epoch": 0.22751260157226796,
|
| 8604 |
+
"grad_norm": 37.0625,
|
| 8605 |
+
"learning_rate": 9.964451166312e-06,
|
| 8606 |
+
"loss": 23.358,
|
| 8607 |
+
"step": 12260
|
| 8608 |
+
},
|
| 8609 |
+
{
|
| 8610 |
+
"epoch": 0.2276981746567478,
|
| 8611 |
+
"grad_norm": 36.75,
|
| 8612 |
+
"learning_rate": 9.964422170525958e-06,
|
| 8613 |
+
"loss": 23.0215,
|
| 8614 |
+
"step": 12270
|
| 8615 |
+
},
|
| 8616 |
+
{
|
| 8617 |
+
"epoch": 0.2278837477412276,
|
| 8618 |
+
"grad_norm": 34.4375,
|
| 8619 |
+
"learning_rate": 9.964393174739915e-06,
|
| 8620 |
+
"loss": 22.7708,
|
| 8621 |
+
"step": 12280
|
| 8622 |
+
},
|
| 8623 |
+
{
|
| 8624 |
+
"epoch": 0.22806932082570744,
|
| 8625 |
+
"grad_norm": 35.46875,
|
| 8626 |
+
"learning_rate": 9.964364178953874e-06,
|
| 8627 |
+
"loss": 23.462,
|
| 8628 |
+
"step": 12290
|
| 8629 |
+
},
|
| 8630 |
+
{
|
| 8631 |
+
"epoch": 0.22825489391018727,
|
| 8632 |
+
"grad_norm": 35.28125,
|
| 8633 |
+
"learning_rate": 9.964335183167832e-06,
|
| 8634 |
+
"loss": 23.129,
|
| 8635 |
+
"step": 12300
|
| 8636 |
+
},
|
| 8637 |
+
{
|
| 8638 |
+
"epoch": 0.22844046699466708,
|
| 8639 |
+
"grad_norm": 34.65625,
|
| 8640 |
+
"learning_rate": 9.964306187381788e-06,
|
| 8641 |
+
"loss": 23.0191,
|
| 8642 |
+
"step": 12310
|
| 8643 |
+
},
|
| 8644 |
+
{
|
| 8645 |
+
"epoch": 0.2286260400791469,
|
| 8646 |
+
"grad_norm": 36.09375,
|
| 8647 |
+
"learning_rate": 9.964277191595747e-06,
|
| 8648 |
+
"loss": 23.0469,
|
| 8649 |
+
"step": 12320
|
| 8650 |
+
},
|
| 8651 |
+
{
|
| 8652 |
+
"epoch": 0.22881161316362675,
|
| 8653 |
+
"grad_norm": 33.53125,
|
| 8654 |
+
"learning_rate": 9.964248195809704e-06,
|
| 8655 |
+
"loss": 22.6963,
|
| 8656 |
+
"step": 12330
|
| 8657 |
+
},
|
| 8658 |
+
{
|
| 8659 |
+
"epoch": 0.22899718624810658,
|
| 8660 |
+
"grad_norm": 35.15625,
|
| 8661 |
+
"learning_rate": 9.964219200023661e-06,
|
| 8662 |
+
"loss": 22.5734,
|
| 8663 |
+
"step": 12340
|
| 8664 |
+
},
|
| 8665 |
+
{
|
| 8666 |
+
"epoch": 0.2291827593325864,
|
| 8667 |
+
"grad_norm": 35.0625,
|
| 8668 |
+
"learning_rate": 9.964190204237619e-06,
|
| 8669 |
+
"loss": 22.6649,
|
| 8670 |
+
"step": 12350
|
| 8671 |
+
},
|
| 8672 |
+
{
|
| 8673 |
+
"epoch": 0.22936833241706622,
|
| 8674 |
+
"grad_norm": 33.59375,
|
| 8675 |
+
"learning_rate": 9.964161208451576e-06,
|
| 8676 |
+
"loss": 23.0639,
|
| 8677 |
+
"step": 12360
|
| 8678 |
+
},
|
| 8679 |
+
{
|
| 8680 |
+
"epoch": 0.22955390550154606,
|
| 8681 |
+
"grad_norm": 34.8125,
|
| 8682 |
+
"learning_rate": 9.964132212665534e-06,
|
| 8683 |
+
"loss": 23.0662,
|
| 8684 |
+
"step": 12370
|
| 8685 |
+
},
|
| 8686 |
+
{
|
| 8687 |
+
"epoch": 0.2297394785860259,
|
| 8688 |
+
"grad_norm": 34.71875,
|
| 8689 |
+
"learning_rate": 9.964103216879491e-06,
|
| 8690 |
+
"loss": 22.8609,
|
| 8691 |
+
"step": 12380
|
| 8692 |
+
},
|
| 8693 |
+
{
|
| 8694 |
+
"epoch": 0.2299250516705057,
|
| 8695 |
+
"grad_norm": 36.8125,
|
| 8696 |
+
"learning_rate": 9.96407422109345e-06,
|
| 8697 |
+
"loss": 23.0976,
|
| 8698 |
+
"step": 12390
|
| 8699 |
+
},
|
| 8700 |
+
{
|
| 8701 |
+
"epoch": 0.23011062475498553,
|
| 8702 |
+
"grad_norm": 34.75,
|
| 8703 |
+
"learning_rate": 9.964045225307408e-06,
|
| 8704 |
+
"loss": 22.8197,
|
| 8705 |
+
"step": 12400
|
| 8706 |
+
},
|
| 8707 |
+
{
|
| 8708 |
+
"epoch": 0.23029619783946537,
|
| 8709 |
+
"grad_norm": 35.5625,
|
| 8710 |
+
"learning_rate": 9.964016229521365e-06,
|
| 8711 |
+
"loss": 23.1144,
|
| 8712 |
+
"step": 12410
|
| 8713 |
+
},
|
| 8714 |
+
{
|
| 8715 |
+
"epoch": 0.2304817709239452,
|
| 8716 |
+
"grad_norm": 33.1875,
|
| 8717 |
+
"learning_rate": 9.963987233735322e-06,
|
| 8718 |
+
"loss": 22.8939,
|
| 8719 |
+
"step": 12420
|
| 8720 |
+
},
|
| 8721 |
+
{
|
| 8722 |
+
"epoch": 0.230667344008425,
|
| 8723 |
+
"grad_norm": 37.0,
|
| 8724 |
+
"learning_rate": 9.96395823794928e-06,
|
| 8725 |
+
"loss": 23.3261,
|
| 8726 |
+
"step": 12430
|
| 8727 |
+
},
|
| 8728 |
+
{
|
| 8729 |
+
"epoch": 0.23085291709290484,
|
| 8730 |
+
"grad_norm": 35.0,
|
| 8731 |
+
"learning_rate": 9.963929242163237e-06,
|
| 8732 |
+
"loss": 22.6032,
|
| 8733 |
+
"step": 12440
|
| 8734 |
+
},
|
| 8735 |
+
{
|
| 8736 |
+
"epoch": 0.23103849017738468,
|
| 8737 |
+
"grad_norm": 36.28125,
|
| 8738 |
+
"learning_rate": 9.963900246377195e-06,
|
| 8739 |
+
"loss": 22.7235,
|
| 8740 |
+
"step": 12450
|
| 8741 |
+
},
|
| 8742 |
+
{
|
| 8743 |
+
"epoch": 0.2312240632618645,
|
| 8744 |
+
"grad_norm": 34.75,
|
| 8745 |
+
"learning_rate": 9.963871250591152e-06,
|
| 8746 |
+
"loss": 22.9642,
|
| 8747 |
+
"step": 12460
|
| 8748 |
+
},
|
| 8749 |
+
{
|
| 8750 |
+
"epoch": 0.23140963634634432,
|
| 8751 |
+
"grad_norm": 34.78125,
|
| 8752 |
+
"learning_rate": 9.96384225480511e-06,
|
| 8753 |
+
"loss": 22.6095,
|
| 8754 |
+
"step": 12470
|
| 8755 |
+
},
|
| 8756 |
+
{
|
| 8757 |
+
"epoch": 0.23159520943082415,
|
| 8758 |
+
"grad_norm": 35.3125,
|
| 8759 |
+
"learning_rate": 9.963813259019067e-06,
|
| 8760 |
+
"loss": 22.7798,
|
| 8761 |
+
"step": 12480
|
| 8762 |
+
},
|
| 8763 |
+
{
|
| 8764 |
+
"epoch": 0.23178078251530398,
|
| 8765 |
+
"grad_norm": 37.90625,
|
| 8766 |
+
"learning_rate": 9.963784263233024e-06,
|
| 8767 |
+
"loss": 22.798,
|
| 8768 |
+
"step": 12490
|
| 8769 |
+
},
|
| 8770 |
+
{
|
| 8771 |
+
"epoch": 0.23196635559978382,
|
| 8772 |
+
"grad_norm": 36.6875,
|
| 8773 |
+
"learning_rate": 9.963755267446983e-06,
|
| 8774 |
+
"loss": 23.0762,
|
| 8775 |
+
"step": 12500
|
| 8776 |
+
},
|
| 8777 |
+
{
|
| 8778 |
+
"epoch": 0.23215192868426363,
|
| 8779 |
+
"grad_norm": 32.65625,
|
| 8780 |
+
"learning_rate": 9.96372627166094e-06,
|
| 8781 |
+
"loss": 22.8446,
|
| 8782 |
+
"step": 12510
|
| 8783 |
+
},
|
| 8784 |
+
{
|
| 8785 |
+
"epoch": 0.23233750176874346,
|
| 8786 |
+
"grad_norm": 35.21875,
|
| 8787 |
+
"learning_rate": 9.963697275874898e-06,
|
| 8788 |
+
"loss": 22.7945,
|
| 8789 |
+
"step": 12520
|
| 8790 |
+
},
|
| 8791 |
+
{
|
| 8792 |
+
"epoch": 0.2325230748532233,
|
| 8793 |
+
"grad_norm": 34.40625,
|
| 8794 |
+
"learning_rate": 9.963668280088856e-06,
|
| 8795 |
+
"loss": 22.8364,
|
| 8796 |
+
"step": 12530
|
| 8797 |
+
},
|
| 8798 |
+
{
|
| 8799 |
+
"epoch": 0.23270864793770313,
|
| 8800 |
+
"grad_norm": 34.96875,
|
| 8801 |
+
"learning_rate": 9.963639284302813e-06,
|
| 8802 |
+
"loss": 23.0125,
|
| 8803 |
+
"step": 12540
|
| 8804 |
+
},
|
| 8805 |
+
{
|
| 8806 |
+
"epoch": 0.23289422102218293,
|
| 8807 |
+
"grad_norm": 33.53125,
|
| 8808 |
+
"learning_rate": 9.96361028851677e-06,
|
| 8809 |
+
"loss": 22.2002,
|
| 8810 |
+
"step": 12550
|
| 8811 |
+
},
|
| 8812 |
+
{
|
| 8813 |
+
"epoch": 0.23307979410666277,
|
| 8814 |
+
"grad_norm": 33.90625,
|
| 8815 |
+
"learning_rate": 9.963581292730728e-06,
|
| 8816 |
+
"loss": 23.0257,
|
| 8817 |
+
"step": 12560
|
| 8818 |
+
},
|
| 8819 |
+
{
|
| 8820 |
+
"epoch": 0.2332653671911426,
|
| 8821 |
+
"grad_norm": 34.6875,
|
| 8822 |
+
"learning_rate": 9.963552296944687e-06,
|
| 8823 |
+
"loss": 22.9528,
|
| 8824 |
+
"step": 12570
|
| 8825 |
+
},
|
| 8826 |
+
{
|
| 8827 |
+
"epoch": 0.23345094027562244,
|
| 8828 |
+
"grad_norm": 34.0,
|
| 8829 |
+
"learning_rate": 9.963523301158643e-06,
|
| 8830 |
+
"loss": 22.9254,
|
| 8831 |
+
"step": 12580
|
| 8832 |
+
},
|
| 8833 |
+
{
|
| 8834 |
+
"epoch": 0.23363651336010224,
|
| 8835 |
+
"grad_norm": 38.78125,
|
| 8836 |
+
"learning_rate": 9.9634943053726e-06,
|
| 8837 |
+
"loss": 23.0386,
|
| 8838 |
+
"step": 12590
|
| 8839 |
+
},
|
| 8840 |
+
{
|
| 8841 |
+
"epoch": 0.23382208644458208,
|
| 8842 |
+
"grad_norm": 35.5,
|
| 8843 |
+
"learning_rate": 9.963465309586559e-06,
|
| 8844 |
+
"loss": 22.6576,
|
| 8845 |
+
"step": 12600
|
| 8846 |
+
},
|
| 8847 |
+
{
|
| 8848 |
+
"epoch": 0.2340076595290619,
|
| 8849 |
+
"grad_norm": 33.375,
|
| 8850 |
+
"learning_rate": 9.963436313800517e-06,
|
| 8851 |
+
"loss": 22.8264,
|
| 8852 |
+
"step": 12610
|
| 8853 |
+
},
|
| 8854 |
+
{
|
| 8855 |
+
"epoch": 0.23419323261354172,
|
| 8856 |
+
"grad_norm": 34.40625,
|
| 8857 |
+
"learning_rate": 9.963407318014474e-06,
|
| 8858 |
+
"loss": 22.7533,
|
| 8859 |
+
"step": 12620
|
| 8860 |
+
},
|
| 8861 |
+
{
|
| 8862 |
+
"epoch": 0.23437880569802155,
|
| 8863 |
+
"grad_norm": 36.84375,
|
| 8864 |
+
"learning_rate": 9.963378322228431e-06,
|
| 8865 |
+
"loss": 23.3032,
|
| 8866 |
+
"step": 12630
|
| 8867 |
+
},
|
| 8868 |
+
{
|
| 8869 |
+
"epoch": 0.2345643787825014,
|
| 8870 |
+
"grad_norm": 32.5,
|
| 8871 |
+
"learning_rate": 9.963349326442389e-06,
|
| 8872 |
+
"loss": 23.052,
|
| 8873 |
+
"step": 12640
|
| 8874 |
+
},
|
| 8875 |
+
{
|
| 8876 |
+
"epoch": 0.23474995186698122,
|
| 8877 |
+
"grad_norm": 33.28125,
|
| 8878 |
+
"learning_rate": 9.963320330656346e-06,
|
| 8879 |
+
"loss": 22.4562,
|
| 8880 |
+
"step": 12650
|
| 8881 |
+
},
|
| 8882 |
+
{
|
| 8883 |
+
"epoch": 0.23493552495146103,
|
| 8884 |
+
"grad_norm": 35.65625,
|
| 8885 |
+
"learning_rate": 9.963291334870304e-06,
|
| 8886 |
+
"loss": 22.9898,
|
| 8887 |
+
"step": 12660
|
| 8888 |
+
},
|
| 8889 |
+
{
|
| 8890 |
+
"epoch": 0.23512109803594086,
|
| 8891 |
+
"grad_norm": 33.625,
|
| 8892 |
+
"learning_rate": 9.963262339084263e-06,
|
| 8893 |
+
"loss": 22.7312,
|
| 8894 |
+
"step": 12670
|
| 8895 |
+
},
|
| 8896 |
+
{
|
| 8897 |
+
"epoch": 0.2353066711204207,
|
| 8898 |
+
"grad_norm": 33.09375,
|
| 8899 |
+
"learning_rate": 9.96323334329822e-06,
|
| 8900 |
+
"loss": 22.2514,
|
| 8901 |
+
"step": 12680
|
| 8902 |
+
},
|
| 8903 |
+
{
|
| 8904 |
+
"epoch": 0.23549224420490053,
|
| 8905 |
+
"grad_norm": 35.40625,
|
| 8906 |
+
"learning_rate": 9.963204347512176e-06,
|
| 8907 |
+
"loss": 22.7423,
|
| 8908 |
+
"step": 12690
|
| 8909 |
+
},
|
| 8910 |
+
{
|
| 8911 |
+
"epoch": 0.23567781728938034,
|
| 8912 |
+
"grad_norm": 33.3125,
|
| 8913 |
+
"learning_rate": 9.963175351726135e-06,
|
| 8914 |
+
"loss": 22.7989,
|
| 8915 |
+
"step": 12700
|
| 8916 |
+
},
|
| 8917 |
+
{
|
| 8918 |
+
"epoch": 0.23586339037386017,
|
| 8919 |
+
"grad_norm": 34.15625,
|
| 8920 |
+
"learning_rate": 9.963146355940092e-06,
|
| 8921 |
+
"loss": 23.4889,
|
| 8922 |
+
"step": 12710
|
| 8923 |
+
},
|
| 8924 |
+
{
|
| 8925 |
+
"epoch": 0.23604896345834,
|
| 8926 |
+
"grad_norm": 33.84375,
|
| 8927 |
+
"learning_rate": 9.96311736015405e-06,
|
| 8928 |
+
"loss": 22.6084,
|
| 8929 |
+
"step": 12720
|
| 8930 |
+
},
|
| 8931 |
+
{
|
| 8932 |
+
"epoch": 0.23623453654281984,
|
| 8933 |
+
"grad_norm": 35.65625,
|
| 8934 |
+
"learning_rate": 9.963088364368007e-06,
|
| 8935 |
+
"loss": 22.9018,
|
| 8936 |
+
"step": 12730
|
| 8937 |
+
},
|
| 8938 |
+
{
|
| 8939 |
+
"epoch": 0.23642010962729965,
|
| 8940 |
+
"grad_norm": 35.53125,
|
| 8941 |
+
"learning_rate": 9.963059368581965e-06,
|
| 8942 |
+
"loss": 22.4362,
|
| 8943 |
+
"step": 12740
|
| 8944 |
+
},
|
| 8945 |
+
{
|
| 8946 |
+
"epoch": 0.23660568271177948,
|
| 8947 |
+
"grad_norm": 36.25,
|
| 8948 |
+
"learning_rate": 9.963030372795922e-06,
|
| 8949 |
+
"loss": 22.874,
|
| 8950 |
+
"step": 12750
|
| 8951 |
+
},
|
| 8952 |
+
{
|
| 8953 |
+
"epoch": 0.23679125579625931,
|
| 8954 |
+
"grad_norm": 33.5,
|
| 8955 |
+
"learning_rate": 9.96300137700988e-06,
|
| 8956 |
+
"loss": 22.7573,
|
| 8957 |
+
"step": 12760
|
| 8958 |
+
},
|
| 8959 |
+
{
|
| 8960 |
+
"epoch": 0.23697682888073915,
|
| 8961 |
+
"grad_norm": 36.46875,
|
| 8962 |
+
"learning_rate": 9.962972381223838e-06,
|
| 8963 |
+
"loss": 22.9307,
|
| 8964 |
+
"step": 12770
|
| 8965 |
+
},
|
| 8966 |
+
{
|
| 8967 |
+
"epoch": 0.23716240196521896,
|
| 8968 |
+
"grad_norm": 35.65625,
|
| 8969 |
+
"learning_rate": 9.962943385437796e-06,
|
| 8970 |
+
"loss": 22.9256,
|
| 8971 |
+
"step": 12780
|
| 8972 |
+
},
|
| 8973 |
+
{
|
| 8974 |
+
"epoch": 0.2373479750496988,
|
| 8975 |
+
"grad_norm": 33.46875,
|
| 8976 |
+
"learning_rate": 9.962914389651752e-06,
|
| 8977 |
+
"loss": 22.8426,
|
| 8978 |
+
"step": 12790
|
| 8979 |
+
},
|
| 8980 |
+
{
|
| 8981 |
+
"epoch": 0.23753354813417862,
|
| 8982 |
+
"grad_norm": 36.28125,
|
| 8983 |
+
"learning_rate": 9.96288539386571e-06,
|
| 8984 |
+
"loss": 23.1097,
|
| 8985 |
+
"step": 12800
|
| 8986 |
+
},
|
| 8987 |
+
{
|
| 8988 |
+
"epoch": 0.23771912121865846,
|
| 8989 |
+
"grad_norm": 33.25,
|
| 8990 |
+
"learning_rate": 9.962856398079668e-06,
|
| 8991 |
+
"loss": 23.2849,
|
| 8992 |
+
"step": 12810
|
| 8993 |
+
},
|
| 8994 |
+
{
|
| 8995 |
+
"epoch": 0.23790469430313826,
|
| 8996 |
+
"grad_norm": 33.65625,
|
| 8997 |
+
"learning_rate": 9.962827402293625e-06,
|
| 8998 |
+
"loss": 22.7846,
|
| 8999 |
+
"step": 12820
|
| 9000 |
+
},
|
| 9001 |
+
{
|
| 9002 |
+
"epoch": 0.2380902673876181,
|
| 9003 |
+
"grad_norm": 34.6875,
|
| 9004 |
+
"learning_rate": 9.962798406507583e-06,
|
| 9005 |
+
"loss": 22.387,
|
| 9006 |
+
"step": 12830
|
| 9007 |
+
},
|
| 9008 |
+
{
|
| 9009 |
+
"epoch": 0.23827584047209793,
|
| 9010 |
+
"grad_norm": 37.28125,
|
| 9011 |
+
"learning_rate": 9.962769410721542e-06,
|
| 9012 |
+
"loss": 22.7212,
|
| 9013 |
+
"step": 12840
|
| 9014 |
+
},
|
| 9015 |
+
{
|
| 9016 |
+
"epoch": 0.23846141355657777,
|
| 9017 |
+
"grad_norm": 37.75,
|
| 9018 |
+
"learning_rate": 9.962740414935498e-06,
|
| 9019 |
+
"loss": 22.644,
|
| 9020 |
+
"step": 12850
|
| 9021 |
+
},
|
| 9022 |
+
{
|
| 9023 |
+
"epoch": 0.23864698664105757,
|
| 9024 |
+
"grad_norm": 37.0625,
|
| 9025 |
+
"learning_rate": 9.962711419149455e-06,
|
| 9026 |
+
"loss": 22.7497,
|
| 9027 |
+
"step": 12860
|
| 9028 |
+
},
|
| 9029 |
+
{
|
| 9030 |
+
"epoch": 0.2388325597255374,
|
| 9031 |
+
"grad_norm": 36.5625,
|
| 9032 |
+
"learning_rate": 9.962682423363414e-06,
|
| 9033 |
+
"loss": 22.9721,
|
| 9034 |
+
"step": 12870
|
| 9035 |
+
},
|
| 9036 |
+
{
|
| 9037 |
+
"epoch": 0.23901813281001724,
|
| 9038 |
+
"grad_norm": 33.46875,
|
| 9039 |
+
"learning_rate": 9.962653427577372e-06,
|
| 9040 |
+
"loss": 22.7019,
|
| 9041 |
+
"step": 12880
|
| 9042 |
+
},
|
| 9043 |
+
{
|
| 9044 |
+
"epoch": 0.23920370589449705,
|
| 9045 |
+
"grad_norm": 34.75,
|
| 9046 |
+
"learning_rate": 9.962624431791329e-06,
|
| 9047 |
+
"loss": 22.5389,
|
| 9048 |
+
"step": 12890
|
| 9049 |
+
},
|
| 9050 |
+
{
|
| 9051 |
+
"epoch": 0.23938927897897688,
|
| 9052 |
+
"grad_norm": 35.09375,
|
| 9053 |
+
"learning_rate": 9.962595436005286e-06,
|
| 9054 |
+
"loss": 22.7755,
|
| 9055 |
+
"step": 12900
|
| 9056 |
+
},
|
| 9057 |
+
{
|
| 9058 |
+
"epoch": 0.23957485206345672,
|
| 9059 |
+
"grad_norm": 35.6875,
|
| 9060 |
+
"learning_rate": 9.962566440219244e-06,
|
| 9061 |
+
"loss": 22.3278,
|
| 9062 |
+
"step": 12910
|
| 9063 |
+
},
|
| 9064 |
+
{
|
| 9065 |
+
"epoch": 0.23976042514793655,
|
| 9066 |
+
"grad_norm": 34.75,
|
| 9067 |
+
"learning_rate": 9.962537444433201e-06,
|
| 9068 |
+
"loss": 22.8705,
|
| 9069 |
+
"step": 12920
|
| 9070 |
+
},
|
| 9071 |
+
{
|
| 9072 |
+
"epoch": 0.23994599823241636,
|
| 9073 |
+
"grad_norm": 35.09375,
|
| 9074 |
+
"learning_rate": 9.962508448647159e-06,
|
| 9075 |
+
"loss": 22.7376,
|
| 9076 |
+
"step": 12930
|
| 9077 |
+
},
|
| 9078 |
+
{
|
| 9079 |
+
"epoch": 0.2401315713168962,
|
| 9080 |
+
"grad_norm": 35.5,
|
| 9081 |
+
"learning_rate": 9.962479452861116e-06,
|
| 9082 |
+
"loss": 22.7123,
|
| 9083 |
+
"step": 12940
|
| 9084 |
+
},
|
| 9085 |
+
{
|
| 9086 |
+
"epoch": 0.24031714440137603,
|
| 9087 |
+
"grad_norm": 32.1875,
|
| 9088 |
+
"learning_rate": 9.962450457075075e-06,
|
| 9089 |
+
"loss": 22.9982,
|
| 9090 |
+
"step": 12950
|
| 9091 |
+
},
|
| 9092 |
+
{
|
| 9093 |
+
"epoch": 0.24050271748585586,
|
| 9094 |
+
"grad_norm": 33.9375,
|
| 9095 |
+
"learning_rate": 9.962421461289031e-06,
|
| 9096 |
+
"loss": 22.7566,
|
| 9097 |
+
"step": 12960
|
| 9098 |
+
},
|
| 9099 |
+
{
|
| 9100 |
+
"epoch": 0.24068829057033567,
|
| 9101 |
+
"grad_norm": 34.15625,
|
| 9102 |
+
"learning_rate": 9.962392465502988e-06,
|
| 9103 |
+
"loss": 22.563,
|
| 9104 |
+
"step": 12970
|
| 9105 |
+
},
|
| 9106 |
+
{
|
| 9107 |
+
"epoch": 0.2408738636548155,
|
| 9108 |
+
"grad_norm": 33.65625,
|
| 9109 |
+
"learning_rate": 9.962363469716947e-06,
|
| 9110 |
+
"loss": 22.8709,
|
| 9111 |
+
"step": 12980
|
| 9112 |
+
},
|
| 9113 |
+
{
|
| 9114 |
+
"epoch": 0.24105943673929534,
|
| 9115 |
+
"grad_norm": 34.5625,
|
| 9116 |
+
"learning_rate": 9.962334473930905e-06,
|
| 9117 |
+
"loss": 22.8837,
|
| 9118 |
+
"step": 12990
|
| 9119 |
+
},
|
| 9120 |
+
{
|
| 9121 |
+
"epoch": 0.24124500982377517,
|
| 9122 |
+
"grad_norm": 36.375,
|
| 9123 |
+
"learning_rate": 9.962305478144862e-06,
|
| 9124 |
+
"loss": 22.7508,
|
| 9125 |
+
"step": 13000
|
| 9126 |
+
},
|
| 9127 |
+
{
|
| 9128 |
+
"epoch": 0.24143058290825498,
|
| 9129 |
+
"grad_norm": 34.0,
|
| 9130 |
+
"learning_rate": 9.96227648235882e-06,
|
| 9131 |
+
"loss": 23.1134,
|
| 9132 |
+
"step": 13010
|
| 9133 |
+
},
|
| 9134 |
+
{
|
| 9135 |
+
"epoch": 0.2416161559927348,
|
| 9136 |
+
"grad_norm": 35.25,
|
| 9137 |
+
"learning_rate": 9.962247486572777e-06,
|
| 9138 |
+
"loss": 22.9102,
|
| 9139 |
+
"step": 13020
|
| 9140 |
+
},
|
| 9141 |
+
{
|
| 9142 |
+
"epoch": 0.24180172907721464,
|
| 9143 |
+
"grad_norm": 34.875,
|
| 9144 |
+
"learning_rate": 9.962218490786734e-06,
|
| 9145 |
+
"loss": 22.9471,
|
| 9146 |
+
"step": 13030
|
| 9147 |
+
},
|
| 9148 |
+
{
|
| 9149 |
+
"epoch": 0.24198730216169448,
|
| 9150 |
+
"grad_norm": 34.53125,
|
| 9151 |
+
"learning_rate": 9.962189495000692e-06,
|
| 9152 |
+
"loss": 22.7062,
|
| 9153 |
+
"step": 13040
|
| 9154 |
+
},
|
| 9155 |
+
{
|
| 9156 |
+
"epoch": 0.24217287524617429,
|
| 9157 |
+
"grad_norm": 37.625,
|
| 9158 |
+
"learning_rate": 9.962160499214651e-06,
|
| 9159 |
+
"loss": 22.6628,
|
| 9160 |
+
"step": 13050
|
| 9161 |
+
},
|
| 9162 |
+
{
|
| 9163 |
+
"epoch": 0.24235844833065412,
|
| 9164 |
+
"grad_norm": 36.15625,
|
| 9165 |
+
"learning_rate": 9.962131503428607e-06,
|
| 9166 |
+
"loss": 22.793,
|
| 9167 |
+
"step": 13060
|
| 9168 |
+
},
|
| 9169 |
+
{
|
| 9170 |
+
"epoch": 0.24254402141513395,
|
| 9171 |
+
"grad_norm": 34.8125,
|
| 9172 |
+
"learning_rate": 9.962102507642564e-06,
|
| 9173 |
+
"loss": 22.6144,
|
| 9174 |
+
"step": 13070
|
| 9175 |
+
},
|
| 9176 |
+
{
|
| 9177 |
+
"epoch": 0.2427295944996138,
|
| 9178 |
+
"grad_norm": 35.21875,
|
| 9179 |
+
"learning_rate": 9.962073511856523e-06,
|
| 9180 |
+
"loss": 22.596,
|
| 9181 |
+
"step": 13080
|
| 9182 |
+
},
|
| 9183 |
+
{
|
| 9184 |
+
"epoch": 0.2429151675840936,
|
| 9185 |
+
"grad_norm": 34.9375,
|
| 9186 |
+
"learning_rate": 9.96204451607048e-06,
|
| 9187 |
+
"loss": 22.6484,
|
| 9188 |
+
"step": 13090
|
| 9189 |
+
},
|
| 9190 |
+
{
|
| 9191 |
+
"epoch": 0.24310074066857343,
|
| 9192 |
+
"grad_norm": 36.0,
|
| 9193 |
+
"learning_rate": 9.962015520284438e-06,
|
| 9194 |
+
"loss": 22.9452,
|
| 9195 |
+
"step": 13100
|
| 9196 |
+
},
|
| 9197 |
+
{
|
| 9198 |
+
"epoch": 0.24328631375305326,
|
| 9199 |
+
"grad_norm": 35.53125,
|
| 9200 |
+
"learning_rate": 9.961986524498395e-06,
|
| 9201 |
+
"loss": 23.1317,
|
| 9202 |
+
"step": 13110
|
| 9203 |
+
},
|
| 9204 |
+
{
|
| 9205 |
+
"epoch": 0.2434718868375331,
|
| 9206 |
+
"grad_norm": 36.03125,
|
| 9207 |
+
"learning_rate": 9.961957528712353e-06,
|
| 9208 |
+
"loss": 22.8217,
|
| 9209 |
+
"step": 13120
|
| 9210 |
+
},
|
| 9211 |
+
{
|
| 9212 |
+
"epoch": 0.2436574599220129,
|
| 9213 |
+
"grad_norm": 34.9375,
|
| 9214 |
+
"learning_rate": 9.96192853292631e-06,
|
| 9215 |
+
"loss": 22.6672,
|
| 9216 |
+
"step": 13130
|
| 9217 |
+
},
|
| 9218 |
+
{
|
| 9219 |
+
"epoch": 0.24384303300649274,
|
| 9220 |
+
"grad_norm": 36.65625,
|
| 9221 |
+
"learning_rate": 9.961899537140268e-06,
|
| 9222 |
+
"loss": 22.8079,
|
| 9223 |
+
"step": 13140
|
| 9224 |
+
},
|
| 9225 |
+
{
|
| 9226 |
+
"epoch": 0.24402860609097257,
|
| 9227 |
+
"grad_norm": 35.65625,
|
| 9228 |
+
"learning_rate": 9.961870541354227e-06,
|
| 9229 |
+
"loss": 22.5951,
|
| 9230 |
+
"step": 13150
|
| 9231 |
+
},
|
| 9232 |
+
{
|
| 9233 |
+
"epoch": 0.2442141791754524,
|
| 9234 |
+
"grad_norm": 34.875,
|
| 9235 |
+
"learning_rate": 9.961841545568184e-06,
|
| 9236 |
+
"loss": 22.7453,
|
| 9237 |
+
"step": 13160
|
| 9238 |
+
},
|
| 9239 |
+
{
|
| 9240 |
+
"epoch": 0.2443997522599322,
|
| 9241 |
+
"grad_norm": 35.03125,
|
| 9242 |
+
"learning_rate": 9.96181254978214e-06,
|
| 9243 |
+
"loss": 22.7423,
|
| 9244 |
+
"step": 13170
|
| 9245 |
+
},
|
| 9246 |
+
{
|
| 9247 |
+
"epoch": 0.24458532534441205,
|
| 9248 |
+
"grad_norm": 35.71875,
|
| 9249 |
+
"learning_rate": 9.961783553996099e-06,
|
| 9250 |
+
"loss": 22.5886,
|
| 9251 |
+
"step": 13180
|
| 9252 |
+
},
|
| 9253 |
+
{
|
| 9254 |
+
"epoch": 0.24477089842889188,
|
| 9255 |
+
"grad_norm": 37.09375,
|
| 9256 |
+
"learning_rate": 9.961754558210056e-06,
|
| 9257 |
+
"loss": 22.8475,
|
| 9258 |
+
"step": 13190
|
| 9259 |
+
},
|
| 9260 |
+
{
|
| 9261 |
+
"epoch": 0.2449564715133717,
|
| 9262 |
+
"grad_norm": 34.03125,
|
| 9263 |
+
"learning_rate": 9.961725562424014e-06,
|
| 9264 |
+
"loss": 22.7532,
|
| 9265 |
+
"step": 13200
|
| 9266 |
+
},
|
| 9267 |
+
{
|
| 9268 |
+
"epoch": 0.24514204459785152,
|
| 9269 |
+
"grad_norm": 35.84375,
|
| 9270 |
+
"learning_rate": 9.961696566637971e-06,
|
| 9271 |
+
"loss": 23.1457,
|
| 9272 |
+
"step": 13210
|
| 9273 |
+
},
|
| 9274 |
+
{
|
| 9275 |
+
"epoch": 0.24532761768233136,
|
| 9276 |
+
"grad_norm": 35.78125,
|
| 9277 |
+
"learning_rate": 9.961667570851929e-06,
|
| 9278 |
+
"loss": 22.8149,
|
| 9279 |
+
"step": 13220
|
| 9280 |
+
},
|
| 9281 |
+
{
|
| 9282 |
+
"epoch": 0.2455131907668112,
|
| 9283 |
+
"grad_norm": 35.8125,
|
| 9284 |
+
"learning_rate": 9.961638575065886e-06,
|
| 9285 |
+
"loss": 22.6625,
|
| 9286 |
+
"step": 13230
|
| 9287 |
+
},
|
| 9288 |
+
{
|
| 9289 |
+
"epoch": 0.245698763851291,
|
| 9290 |
+
"grad_norm": 34.5625,
|
| 9291 |
+
"learning_rate": 9.961609579279843e-06,
|
| 9292 |
+
"loss": 22.3682,
|
| 9293 |
+
"step": 13240
|
| 9294 |
+
},
|
| 9295 |
+
{
|
| 9296 |
+
"epoch": 0.24588433693577083,
|
| 9297 |
+
"grad_norm": 35.03125,
|
| 9298 |
+
"learning_rate": 9.961580583493802e-06,
|
| 9299 |
+
"loss": 22.7562,
|
| 9300 |
+
"step": 13250
|
| 9301 |
+
},
|
| 9302 |
+
{
|
| 9303 |
+
"epoch": 0.24606991002025067,
|
| 9304 |
+
"grad_norm": 36.03125,
|
| 9305 |
+
"learning_rate": 9.96155158770776e-06,
|
| 9306 |
+
"loss": 23.0357,
|
| 9307 |
+
"step": 13260
|
| 9308 |
+
},
|
| 9309 |
+
{
|
| 9310 |
+
"epoch": 0.2462554831047305,
|
| 9311 |
+
"grad_norm": 35.5625,
|
| 9312 |
+
"learning_rate": 9.961522591921717e-06,
|
| 9313 |
+
"loss": 22.7249,
|
| 9314 |
+
"step": 13270
|
| 9315 |
+
},
|
| 9316 |
+
{
|
| 9317 |
+
"epoch": 0.2464410561892103,
|
| 9318 |
+
"grad_norm": 34.25,
|
| 9319 |
+
"learning_rate": 9.961493596135675e-06,
|
| 9320 |
+
"loss": 22.897,
|
| 9321 |
+
"step": 13280
|
| 9322 |
+
},
|
| 9323 |
+
{
|
| 9324 |
+
"epoch": 0.24662662927369014,
|
| 9325 |
+
"grad_norm": 34.0,
|
| 9326 |
+
"learning_rate": 9.961464600349632e-06,
|
| 9327 |
+
"loss": 22.0965,
|
| 9328 |
+
"step": 13290
|
| 9329 |
+
},
|
| 9330 |
+
{
|
| 9331 |
+
"epoch": 0.24681220235816997,
|
| 9332 |
+
"grad_norm": 35.21875,
|
| 9333 |
+
"learning_rate": 9.96143560456359e-06,
|
| 9334 |
+
"loss": 22.5676,
|
| 9335 |
+
"step": 13300
|
| 9336 |
+
},
|
| 9337 |
+
{
|
| 9338 |
+
"epoch": 0.2469977754426498,
|
| 9339 |
+
"grad_norm": 34.28125,
|
| 9340 |
+
"learning_rate": 9.961406608777547e-06,
|
| 9341 |
+
"loss": 23.0798,
|
| 9342 |
+
"step": 13310
|
| 9343 |
+
},
|
| 9344 |
+
{
|
| 9345 |
+
"epoch": 0.24718334852712961,
|
| 9346 |
+
"grad_norm": 33.34375,
|
| 9347 |
+
"learning_rate": 9.961377612991506e-06,
|
| 9348 |
+
"loss": 22.7305,
|
| 9349 |
+
"step": 13320
|
| 9350 |
+
},
|
| 9351 |
+
{
|
| 9352 |
+
"epoch": 0.24736892161160945,
|
| 9353 |
+
"grad_norm": 34.09375,
|
| 9354 |
+
"learning_rate": 9.961348617205462e-06,
|
| 9355 |
+
"loss": 22.8526,
|
| 9356 |
+
"step": 13330
|
| 9357 |
+
},
|
| 9358 |
+
{
|
| 9359 |
+
"epoch": 0.24755449469608928,
|
| 9360 |
+
"grad_norm": 35.9375,
|
| 9361 |
+
"learning_rate": 9.961319621419419e-06,
|
| 9362 |
+
"loss": 22.5833,
|
| 9363 |
+
"step": 13340
|
| 9364 |
+
},
|
| 9365 |
+
{
|
| 9366 |
+
"epoch": 0.24774006778056912,
|
| 9367 |
+
"grad_norm": 34.46875,
|
| 9368 |
+
"learning_rate": 9.961290625633378e-06,
|
| 9369 |
+
"loss": 22.9855,
|
| 9370 |
+
"step": 13350
|
| 9371 |
+
},
|
| 9372 |
+
{
|
| 9373 |
+
"epoch": 0.24792564086504892,
|
| 9374 |
+
"grad_norm": 35.375,
|
| 9375 |
+
"learning_rate": 9.961261629847336e-06,
|
| 9376 |
+
"loss": 22.8834,
|
| 9377 |
+
"step": 13360
|
| 9378 |
+
},
|
| 9379 |
+
{
|
| 9380 |
+
"epoch": 0.24811121394952876,
|
| 9381 |
+
"grad_norm": 34.59375,
|
| 9382 |
+
"learning_rate": 9.961232634061293e-06,
|
| 9383 |
+
"loss": 22.9293,
|
| 9384 |
+
"step": 13370
|
| 9385 |
+
},
|
| 9386 |
+
{
|
| 9387 |
+
"epoch": 0.2482967870340086,
|
| 9388 |
+
"grad_norm": 33.9375,
|
| 9389 |
+
"learning_rate": 9.96120363827525e-06,
|
| 9390 |
+
"loss": 22.2887,
|
| 9391 |
+
"step": 13380
|
| 9392 |
+
},
|
| 9393 |
+
{
|
| 9394 |
+
"epoch": 0.24848236011848843,
|
| 9395 |
+
"grad_norm": 33.5625,
|
| 9396 |
+
"learning_rate": 9.961174642489208e-06,
|
| 9397 |
+
"loss": 23.0661,
|
| 9398 |
+
"step": 13390
|
| 9399 |
+
},
|
| 9400 |
+
{
|
| 9401 |
+
"epoch": 0.24866793320296823,
|
| 9402 |
+
"grad_norm": 36.125,
|
| 9403 |
+
"learning_rate": 9.961145646703165e-06,
|
| 9404 |
+
"loss": 23.0035,
|
| 9405 |
+
"step": 13400
|
| 9406 |
+
},
|
| 9407 |
+
{
|
| 9408 |
+
"epoch": 0.24885350628744807,
|
| 9409 |
+
"grad_norm": 36.875,
|
| 9410 |
+
"learning_rate": 9.961116650917123e-06,
|
| 9411 |
+
"loss": 23.2064,
|
| 9412 |
+
"step": 13410
|
| 9413 |
+
},
|
| 9414 |
+
{
|
| 9415 |
+
"epoch": 0.2490390793719279,
|
| 9416 |
+
"grad_norm": 34.625,
|
| 9417 |
+
"learning_rate": 9.96108765513108e-06,
|
| 9418 |
+
"loss": 22.6221,
|
| 9419 |
+
"step": 13420
|
| 9420 |
+
},
|
| 9421 |
+
{
|
| 9422 |
+
"epoch": 0.24922465245640774,
|
| 9423 |
+
"grad_norm": 35.21875,
|
| 9424 |
+
"learning_rate": 9.96105865934504e-06,
|
| 9425 |
+
"loss": 22.5216,
|
| 9426 |
+
"step": 13430
|
| 9427 |
+
},
|
| 9428 |
+
{
|
| 9429 |
+
"epoch": 0.24941022554088754,
|
| 9430 |
+
"grad_norm": 35.34375,
|
| 9431 |
+
"learning_rate": 9.961029663558995e-06,
|
| 9432 |
+
"loss": 22.5853,
|
| 9433 |
+
"step": 13440
|
| 9434 |
+
},
|
| 9435 |
+
{
|
| 9436 |
+
"epoch": 0.24959579862536738,
|
| 9437 |
+
"grad_norm": 36.25,
|
| 9438 |
+
"learning_rate": 9.961000667772954e-06,
|
| 9439 |
+
"loss": 22.6705,
|
| 9440 |
+
"step": 13450
|
| 9441 |
+
},
|
| 9442 |
+
{
|
| 9443 |
+
"epoch": 0.2497813717098472,
|
| 9444 |
+
"grad_norm": 33.9375,
|
| 9445 |
+
"learning_rate": 9.960971671986911e-06,
|
| 9446 |
+
"loss": 22.7321,
|
| 9447 |
+
"step": 13460
|
| 9448 |
+
},
|
| 9449 |
+
{
|
| 9450 |
+
"epoch": 0.24996694479432702,
|
| 9451 |
+
"grad_norm": 35.4375,
|
| 9452 |
+
"learning_rate": 9.960942676200869e-06,
|
| 9453 |
+
"loss": 22.6743,
|
| 9454 |
+
"step": 13470
|
| 9455 |
+
},
|
| 9456 |
+
{
|
| 9457 |
+
"epoch": 0.2501525178788069,
|
| 9458 |
+
"grad_norm": 33.53125,
|
| 9459 |
+
"learning_rate": 9.960913680414826e-06,
|
| 9460 |
+
"loss": 22.8717,
|
| 9461 |
+
"step": 13480
|
| 9462 |
+
},
|
| 9463 |
+
{
|
| 9464 |
+
"epoch": 0.2503380909632867,
|
| 9465 |
+
"grad_norm": 35.59375,
|
| 9466 |
+
"learning_rate": 9.960884684628784e-06,
|
| 9467 |
+
"loss": 22.5434,
|
| 9468 |
+
"step": 13490
|
| 9469 |
+
},
|
| 9470 |
+
{
|
| 9471 |
+
"epoch": 0.2505236640477665,
|
| 9472 |
+
"grad_norm": 35.40625,
|
| 9473 |
+
"learning_rate": 9.960855688842741e-06,
|
| 9474 |
+
"loss": 22.4418,
|
| 9475 |
+
"step": 13500
|
| 9476 |
+
},
|
| 9477 |
+
{
|
| 9478 |
+
"epoch": 0.25070923713224635,
|
| 9479 |
+
"grad_norm": 34.375,
|
| 9480 |
+
"learning_rate": 9.960826693056698e-06,
|
| 9481 |
+
"loss": 22.7085,
|
| 9482 |
+
"step": 13510
|
| 9483 |
+
},
|
| 9484 |
+
{
|
| 9485 |
+
"epoch": 0.25089481021672616,
|
| 9486 |
+
"grad_norm": 34.71875,
|
| 9487 |
+
"learning_rate": 9.960797697270656e-06,
|
| 9488 |
+
"loss": 22.6424,
|
| 9489 |
+
"step": 13520
|
| 9490 |
+
},
|
| 9491 |
+
{
|
| 9492 |
+
"epoch": 0.25108038330120597,
|
| 9493 |
+
"grad_norm": 34.46875,
|
| 9494 |
+
"learning_rate": 9.960768701484615e-06,
|
| 9495 |
+
"loss": 22.6786,
|
| 9496 |
+
"step": 13530
|
| 9497 |
+
},
|
| 9498 |
+
{
|
| 9499 |
+
"epoch": 0.25126595638568583,
|
| 9500 |
+
"grad_norm": 35.21875,
|
| 9501 |
+
"learning_rate": 9.960739705698572e-06,
|
| 9502 |
+
"loss": 22.7157,
|
| 9503 |
+
"step": 13540
|
| 9504 |
+
},
|
| 9505 |
+
{
|
| 9506 |
+
"epoch": 0.25145152947016564,
|
| 9507 |
+
"grad_norm": 34.84375,
|
| 9508 |
+
"learning_rate": 9.960710709912528e-06,
|
| 9509 |
+
"loss": 22.7138,
|
| 9510 |
+
"step": 13550
|
| 9511 |
+
},
|
| 9512 |
+
{
|
| 9513 |
+
"epoch": 0.2516371025546455,
|
| 9514 |
+
"grad_norm": 34.625,
|
| 9515 |
+
"learning_rate": 9.960681714126487e-06,
|
| 9516 |
+
"loss": 22.8282,
|
| 9517 |
+
"step": 13560
|
| 9518 |
+
},
|
| 9519 |
+
{
|
| 9520 |
+
"epoch": 0.2518226756391253,
|
| 9521 |
+
"grad_norm": 33.28125,
|
| 9522 |
+
"learning_rate": 9.960652718340445e-06,
|
| 9523 |
+
"loss": 22.7318,
|
| 9524 |
+
"step": 13570
|
| 9525 |
+
},
|
| 9526 |
+
{
|
| 9527 |
+
"epoch": 0.2520082487236051,
|
| 9528 |
+
"grad_norm": 35.59375,
|
| 9529 |
+
"learning_rate": 9.960623722554402e-06,
|
| 9530 |
+
"loss": 22.5847,
|
| 9531 |
+
"step": 13580
|
| 9532 |
+
},
|
| 9533 |
+
{
|
| 9534 |
+
"epoch": 0.252193821808085,
|
| 9535 |
+
"grad_norm": 35.0,
|
| 9536 |
+
"learning_rate": 9.96059472676836e-06,
|
| 9537 |
+
"loss": 22.6681,
|
| 9538 |
+
"step": 13590
|
| 9539 |
+
},
|
| 9540 |
+
{
|
| 9541 |
+
"epoch": 0.2523793948925648,
|
| 9542 |
+
"grad_norm": 35.8125,
|
| 9543 |
+
"learning_rate": 9.960565730982317e-06,
|
| 9544 |
+
"loss": 22.6368,
|
| 9545 |
+
"step": 13600
|
| 9546 |
+
},
|
| 9547 |
+
{
|
| 9548 |
+
"epoch": 0.2525649679770446,
|
| 9549 |
+
"grad_norm": 36.3125,
|
| 9550 |
+
"learning_rate": 9.960536735196274e-06,
|
| 9551 |
+
"loss": 22.5425,
|
| 9552 |
+
"step": 13610
|
| 9553 |
+
},
|
| 9554 |
+
{
|
| 9555 |
+
"epoch": 0.25275054106152445,
|
| 9556 |
+
"grad_norm": 34.78125,
|
| 9557 |
+
"learning_rate": 9.960507739410232e-06,
|
| 9558 |
+
"loss": 23.2371,
|
| 9559 |
+
"step": 13620
|
| 9560 |
+
},
|
| 9561 |
+
{
|
| 9562 |
+
"epoch": 0.25293611414600425,
|
| 9563 |
+
"grad_norm": 33.96875,
|
| 9564 |
+
"learning_rate": 9.96047874362419e-06,
|
| 9565 |
+
"loss": 22.8813,
|
| 9566 |
+
"step": 13630
|
| 9567 |
+
},
|
| 9568 |
+
{
|
| 9569 |
+
"epoch": 0.2531216872304841,
|
| 9570 |
+
"grad_norm": 34.65625,
|
| 9571 |
+
"learning_rate": 9.960449747838148e-06,
|
| 9572 |
+
"loss": 22.4802,
|
| 9573 |
+
"step": 13640
|
| 9574 |
+
},
|
| 9575 |
+
{
|
| 9576 |
+
"epoch": 0.2533072603149639,
|
| 9577 |
+
"grad_norm": 36.0625,
|
| 9578 |
+
"learning_rate": 9.960420752052104e-06,
|
| 9579 |
+
"loss": 22.7154,
|
| 9580 |
+
"step": 13650
|
| 9581 |
+
},
|
| 9582 |
+
{
|
| 9583 |
+
"epoch": 0.25349283339944373,
|
| 9584 |
+
"grad_norm": 35.21875,
|
| 9585 |
+
"learning_rate": 9.960391756266063e-06,
|
| 9586 |
+
"loss": 22.3837,
|
| 9587 |
+
"step": 13660
|
| 9588 |
+
},
|
| 9589 |
+
{
|
| 9590 |
+
"epoch": 0.2536784064839236,
|
| 9591 |
+
"grad_norm": 35.625,
|
| 9592 |
+
"learning_rate": 9.96036276048002e-06,
|
| 9593 |
+
"loss": 22.4617,
|
| 9594 |
+
"step": 13670
|
| 9595 |
+
},
|
| 9596 |
+
{
|
| 9597 |
+
"epoch": 0.2538639795684034,
|
| 9598 |
+
"grad_norm": 34.0,
|
| 9599 |
+
"learning_rate": 9.960333764693978e-06,
|
| 9600 |
+
"loss": 22.7565,
|
| 9601 |
+
"step": 13680
|
| 9602 |
+
},
|
| 9603 |
+
{
|
| 9604 |
+
"epoch": 0.2540495526528832,
|
| 9605 |
+
"grad_norm": 34.90625,
|
| 9606 |
+
"learning_rate": 9.960304768907935e-06,
|
| 9607 |
+
"loss": 22.7332,
|
| 9608 |
+
"step": 13690
|
| 9609 |
+
},
|
| 9610 |
+
{
|
| 9611 |
+
"epoch": 0.25423512573736307,
|
| 9612 |
+
"grad_norm": 34.28125,
|
| 9613 |
+
"learning_rate": 9.960275773121894e-06,
|
| 9614 |
+
"loss": 22.6901,
|
| 9615 |
+
"step": 13700
|
| 9616 |
+
},
|
| 9617 |
+
{
|
| 9618 |
+
"epoch": 0.25442069882184287,
|
| 9619 |
+
"grad_norm": 33.8125,
|
| 9620 |
+
"learning_rate": 9.96024677733585e-06,
|
| 9621 |
+
"loss": 22.8362,
|
| 9622 |
+
"step": 13710
|
| 9623 |
+
},
|
| 9624 |
+
{
|
| 9625 |
+
"epoch": 0.25460627190632273,
|
| 9626 |
+
"grad_norm": 33.78125,
|
| 9627 |
+
"learning_rate": 9.960217781549807e-06,
|
| 9628 |
+
"loss": 22.5693,
|
| 9629 |
+
"step": 13720
|
| 9630 |
+
},
|
| 9631 |
+
{
|
| 9632 |
+
"epoch": 0.25479184499080254,
|
| 9633 |
+
"grad_norm": 34.3125,
|
| 9634 |
+
"learning_rate": 9.960188785763766e-06,
|
| 9635 |
+
"loss": 22.2942,
|
| 9636 |
+
"step": 13730
|
| 9637 |
+
},
|
| 9638 |
+
{
|
| 9639 |
+
"epoch": 0.25497741807528235,
|
| 9640 |
+
"grad_norm": 35.375,
|
| 9641 |
+
"learning_rate": 9.960159789977724e-06,
|
| 9642 |
+
"loss": 23.0086,
|
| 9643 |
+
"step": 13740
|
| 9644 |
+
},
|
| 9645 |
+
{
|
| 9646 |
+
"epoch": 0.2551629911597622,
|
| 9647 |
+
"grad_norm": 35.90625,
|
| 9648 |
+
"learning_rate": 9.960130794191681e-06,
|
| 9649 |
+
"loss": 22.6488,
|
| 9650 |
+
"step": 13750
|
| 9651 |
+
},
|
| 9652 |
+
{
|
| 9653 |
+
"epoch": 0.255348564244242,
|
| 9654 |
+
"grad_norm": 35.59375,
|
| 9655 |
+
"learning_rate": 9.960101798405639e-06,
|
| 9656 |
+
"loss": 23.0556,
|
| 9657 |
+
"step": 13760
|
| 9658 |
+
},
|
| 9659 |
+
{
|
| 9660 |
+
"epoch": 0.2555341373287218,
|
| 9661 |
+
"grad_norm": 34.8125,
|
| 9662 |
+
"learning_rate": 9.960072802619596e-06,
|
| 9663 |
+
"loss": 22.7591,
|
| 9664 |
+
"step": 13770
|
| 9665 |
+
},
|
| 9666 |
+
{
|
| 9667 |
+
"epoch": 0.2557197104132017,
|
| 9668 |
+
"grad_norm": 35.59375,
|
| 9669 |
+
"learning_rate": 9.960043806833553e-06,
|
| 9670 |
+
"loss": 22.4961,
|
| 9671 |
+
"step": 13780
|
| 9672 |
+
},
|
| 9673 |
+
{
|
| 9674 |
+
"epoch": 0.2559052834976815,
|
| 9675 |
+
"grad_norm": 35.0,
|
| 9676 |
+
"learning_rate": 9.960014811047511e-06,
|
| 9677 |
+
"loss": 22.5756,
|
| 9678 |
+
"step": 13790
|
| 9679 |
+
},
|
| 9680 |
+
{
|
| 9681 |
+
"epoch": 0.2560908565821613,
|
| 9682 |
+
"grad_norm": 35.3125,
|
| 9683 |
+
"learning_rate": 9.95998581526147e-06,
|
| 9684 |
+
"loss": 22.5104,
|
| 9685 |
+
"step": 13800
|
| 9686 |
+
},
|
| 9687 |
+
{
|
| 9688 |
+
"epoch": 0.25627642966664116,
|
| 9689 |
+
"grad_norm": 36.6875,
|
| 9690 |
+
"learning_rate": 9.959956819475426e-06,
|
| 9691 |
+
"loss": 23.0103,
|
| 9692 |
+
"step": 13810
|
| 9693 |
+
},
|
| 9694 |
+
{
|
| 9695 |
+
"epoch": 0.25646200275112097,
|
| 9696 |
+
"grad_norm": 36.125,
|
| 9697 |
+
"learning_rate": 9.959927823689383e-06,
|
| 9698 |
+
"loss": 22.4589,
|
| 9699 |
+
"step": 13820
|
| 9700 |
+
},
|
| 9701 |
+
{
|
| 9702 |
+
"epoch": 0.2566475758356008,
|
| 9703 |
+
"grad_norm": 38.625,
|
| 9704 |
+
"learning_rate": 9.959898827903342e-06,
|
| 9705 |
+
"loss": 22.8177,
|
| 9706 |
+
"step": 13830
|
| 9707 |
+
},
|
| 9708 |
+
{
|
| 9709 |
+
"epoch": 0.25683314892008063,
|
| 9710 |
+
"grad_norm": 35.15625,
|
| 9711 |
+
"learning_rate": 9.9598698321173e-06,
|
| 9712 |
+
"loss": 22.5105,
|
| 9713 |
+
"step": 13840
|
| 9714 |
+
},
|
| 9715 |
+
{
|
| 9716 |
+
"epoch": 0.25701872200456044,
|
| 9717 |
+
"grad_norm": 35.875,
|
| 9718 |
+
"learning_rate": 9.959840836331257e-06,
|
| 9719 |
+
"loss": 22.8462,
|
| 9720 |
+
"step": 13850
|
| 9721 |
+
},
|
| 9722 |
+
{
|
| 9723 |
+
"epoch": 0.2572042950890403,
|
| 9724 |
+
"grad_norm": 36.1875,
|
| 9725 |
+
"learning_rate": 9.959811840545214e-06,
|
| 9726 |
+
"loss": 22.693,
|
| 9727 |
+
"step": 13860
|
| 9728 |
+
},
|
| 9729 |
+
{
|
| 9730 |
+
"epoch": 0.2573898681735201,
|
| 9731 |
+
"grad_norm": 35.53125,
|
| 9732 |
+
"learning_rate": 9.959782844759172e-06,
|
| 9733 |
+
"loss": 22.4419,
|
| 9734 |
+
"step": 13870
|
| 9735 |
+
},
|
| 9736 |
+
{
|
| 9737 |
+
"epoch": 0.2575754412579999,
|
| 9738 |
+
"grad_norm": 35.84375,
|
| 9739 |
+
"learning_rate": 9.95975384897313e-06,
|
| 9740 |
+
"loss": 22.8257,
|
| 9741 |
+
"step": 13880
|
| 9742 |
+
},
|
| 9743 |
+
{
|
| 9744 |
+
"epoch": 0.2577610143424798,
|
| 9745 |
+
"grad_norm": 33.9375,
|
| 9746 |
+
"learning_rate": 9.959724853187087e-06,
|
| 9747 |
+
"loss": 22.5976,
|
| 9748 |
+
"step": 13890
|
| 9749 |
+
},
|
| 9750 |
+
{
|
| 9751 |
+
"epoch": 0.2579465874269596,
|
| 9752 |
+
"grad_norm": 32.59375,
|
| 9753 |
+
"learning_rate": 9.959695857401046e-06,
|
| 9754 |
+
"loss": 22.5116,
|
| 9755 |
+
"step": 13900
|
| 9756 |
+
},
|
| 9757 |
+
{
|
| 9758 |
+
"epoch": 0.25813216051143945,
|
| 9759 |
+
"grad_norm": 34.625,
|
| 9760 |
+
"learning_rate": 9.959666861615003e-06,
|
| 9761 |
+
"loss": 22.3501,
|
| 9762 |
+
"step": 13910
|
| 9763 |
+
},
|
| 9764 |
+
{
|
| 9765 |
+
"epoch": 0.25831773359591925,
|
| 9766 |
+
"grad_norm": 34.09375,
|
| 9767 |
+
"learning_rate": 9.959637865828959e-06,
|
| 9768 |
+
"loss": 22.4398,
|
| 9769 |
+
"step": 13920
|
| 9770 |
+
},
|
| 9771 |
+
{
|
| 9772 |
+
"epoch": 0.25850330668039906,
|
| 9773 |
+
"grad_norm": 36.21875,
|
| 9774 |
+
"learning_rate": 9.959608870042918e-06,
|
| 9775 |
+
"loss": 22.5873,
|
| 9776 |
+
"step": 13930
|
| 9777 |
+
},
|
| 9778 |
+
{
|
| 9779 |
+
"epoch": 0.2586888797648789,
|
| 9780 |
+
"grad_norm": 36.21875,
|
| 9781 |
+
"learning_rate": 9.959579874256875e-06,
|
| 9782 |
+
"loss": 22.4183,
|
| 9783 |
+
"step": 13940
|
| 9784 |
+
},
|
| 9785 |
+
{
|
| 9786 |
+
"epoch": 0.2588744528493587,
|
| 9787 |
+
"grad_norm": 35.40625,
|
| 9788 |
+
"learning_rate": 9.959550878470833e-06,
|
| 9789 |
+
"loss": 22.896,
|
| 9790 |
+
"step": 13950
|
| 9791 |
+
},
|
| 9792 |
+
{
|
| 9793 |
+
"epoch": 0.25906002593383853,
|
| 9794 |
+
"grad_norm": 34.5,
|
| 9795 |
+
"learning_rate": 9.95952188268479e-06,
|
| 9796 |
+
"loss": 22.2744,
|
| 9797 |
+
"step": 13960
|
| 9798 |
+
},
|
| 9799 |
+
{
|
| 9800 |
+
"epoch": 0.2592455990183184,
|
| 9801 |
+
"grad_norm": 33.25,
|
| 9802 |
+
"learning_rate": 9.959492886898748e-06,
|
| 9803 |
+
"loss": 22.7941,
|
| 9804 |
+
"step": 13970
|
| 9805 |
+
},
|
| 9806 |
+
{
|
| 9807 |
+
"epoch": 0.2594311721027982,
|
| 9808 |
+
"grad_norm": 34.65625,
|
| 9809 |
+
"learning_rate": 9.959463891112705e-06,
|
| 9810 |
+
"loss": 22.2191,
|
| 9811 |
+
"step": 13980
|
| 9812 |
+
},
|
| 9813 |
+
{
|
| 9814 |
+
"epoch": 0.25961674518727806,
|
| 9815 |
+
"grad_norm": 35.15625,
|
| 9816 |
+
"learning_rate": 9.959434895326662e-06,
|
| 9817 |
+
"loss": 22.6185,
|
| 9818 |
+
"step": 13990
|
| 9819 |
+
},
|
| 9820 |
+
{
|
| 9821 |
+
"epoch": 0.25980231827175787,
|
| 9822 |
+
"grad_norm": 34.4375,
|
| 9823 |
+
"learning_rate": 9.95940589954062e-06,
|
| 9824 |
+
"loss": 22.1749,
|
| 9825 |
+
"step": 14000
|
| 9826 |
+
},
|
| 9827 |
+
{
|
| 9828 |
+
"epoch": 0.2599878913562377,
|
| 9829 |
+
"grad_norm": 35.25,
|
| 9830 |
+
"learning_rate": 9.959376903754579e-06,
|
| 9831 |
+
"loss": 22.8769,
|
| 9832 |
+
"step": 14010
|
| 9833 |
+
},
|
| 9834 |
+
{
|
| 9835 |
+
"epoch": 0.26017346444071754,
|
| 9836 |
+
"grad_norm": 33.25,
|
| 9837 |
+
"learning_rate": 9.959347907968536e-06,
|
| 9838 |
+
"loss": 22.8237,
|
| 9839 |
+
"step": 14020
|
| 9840 |
+
},
|
| 9841 |
+
{
|
| 9842 |
+
"epoch": 0.26035903752519735,
|
| 9843 |
+
"grad_norm": 33.4375,
|
| 9844 |
+
"learning_rate": 9.959318912182494e-06,
|
| 9845 |
+
"loss": 22.6494,
|
| 9846 |
+
"step": 14030
|
| 9847 |
+
},
|
| 9848 |
+
{
|
| 9849 |
+
"epoch": 0.26054461060967715,
|
| 9850 |
+
"grad_norm": 35.9375,
|
| 9851 |
+
"learning_rate": 9.959289916396451e-06,
|
| 9852 |
+
"loss": 22.4676,
|
| 9853 |
+
"step": 14040
|
| 9854 |
+
},
|
| 9855 |
+
{
|
| 9856 |
+
"epoch": 0.260730183694157,
|
| 9857 |
+
"grad_norm": 35.84375,
|
| 9858 |
+
"learning_rate": 9.959260920610409e-06,
|
| 9859 |
+
"loss": 22.5966,
|
| 9860 |
+
"step": 14050
|
| 9861 |
+
},
|
| 9862 |
+
{
|
| 9863 |
+
"epoch": 0.2609157567786368,
|
| 9864 |
+
"grad_norm": 35.46875,
|
| 9865 |
+
"learning_rate": 9.959231924824366e-06,
|
| 9866 |
+
"loss": 22.6443,
|
| 9867 |
+
"step": 14060
|
| 9868 |
+
},
|
| 9869 |
+
{
|
| 9870 |
+
"epoch": 0.2611013298631166,
|
| 9871 |
+
"grad_norm": 35.28125,
|
| 9872 |
+
"learning_rate": 9.959202929038323e-06,
|
| 9873 |
+
"loss": 22.7265,
|
| 9874 |
+
"step": 14070
|
| 9875 |
+
},
|
| 9876 |
+
{
|
| 9877 |
+
"epoch": 0.2612869029475965,
|
| 9878 |
+
"grad_norm": 33.375,
|
| 9879 |
+
"learning_rate": 9.95917393325228e-06,
|
| 9880 |
+
"loss": 22.7487,
|
| 9881 |
+
"step": 14080
|
| 9882 |
+
},
|
| 9883 |
+
{
|
| 9884 |
+
"epoch": 0.2614724760320763,
|
| 9885 |
+
"grad_norm": 35.6875,
|
| 9886 |
+
"learning_rate": 9.959144937466238e-06,
|
| 9887 |
+
"loss": 22.6146,
|
| 9888 |
+
"step": 14090
|
| 9889 |
+
},
|
| 9890 |
+
{
|
| 9891 |
+
"epoch": 0.26165804911655616,
|
| 9892 |
+
"grad_norm": 32.78125,
|
| 9893 |
+
"learning_rate": 9.959115941680196e-06,
|
| 9894 |
+
"loss": 22.4781,
|
| 9895 |
+
"step": 14100
|
| 9896 |
+
},
|
| 9897 |
+
{
|
| 9898 |
+
"epoch": 0.26184362220103596,
|
| 9899 |
+
"grad_norm": 33.625,
|
| 9900 |
+
"learning_rate": 9.959086945894155e-06,
|
| 9901 |
+
"loss": 22.5717,
|
| 9902 |
+
"step": 14110
|
| 9903 |
+
},
|
| 9904 |
+
{
|
| 9905 |
+
"epoch": 0.26202919528551577,
|
| 9906 |
+
"grad_norm": 35.625,
|
| 9907 |
+
"learning_rate": 9.959057950108112e-06,
|
| 9908 |
+
"loss": 22.8922,
|
| 9909 |
+
"step": 14120
|
| 9910 |
+
},
|
| 9911 |
+
{
|
| 9912 |
+
"epoch": 0.26221476836999563,
|
| 9913 |
+
"grad_norm": 34.65625,
|
| 9914 |
+
"learning_rate": 9.95902895432207e-06,
|
| 9915 |
+
"loss": 22.4869,
|
| 9916 |
+
"step": 14130
|
| 9917 |
+
},
|
| 9918 |
+
{
|
| 9919 |
+
"epoch": 0.26240034145447544,
|
| 9920 |
+
"grad_norm": 34.90625,
|
| 9921 |
+
"learning_rate": 9.958999958536027e-06,
|
| 9922 |
+
"loss": 22.5448,
|
| 9923 |
+
"step": 14140
|
| 9924 |
+
},
|
| 9925 |
+
{
|
| 9926 |
+
"epoch": 0.26258591453895525,
|
| 9927 |
+
"grad_norm": 37.15625,
|
| 9928 |
+
"learning_rate": 9.958970962749984e-06,
|
| 9929 |
+
"loss": 22.8666,
|
| 9930 |
+
"step": 14150
|
| 9931 |
+
},
|
| 9932 |
+
{
|
| 9933 |
+
"epoch": 0.2627714876234351,
|
| 9934 |
+
"grad_norm": 34.15625,
|
| 9935 |
+
"learning_rate": 9.958941966963942e-06,
|
| 9936 |
+
"loss": 22.4539,
|
| 9937 |
+
"step": 14160
|
| 9938 |
+
},
|
| 9939 |
+
{
|
| 9940 |
+
"epoch": 0.2629570607079149,
|
| 9941 |
+
"grad_norm": 34.03125,
|
| 9942 |
+
"learning_rate": 9.958912971177899e-06,
|
| 9943 |
+
"loss": 22.6467,
|
| 9944 |
+
"step": 14170
|
| 9945 |
+
},
|
| 9946 |
+
{
|
| 9947 |
+
"epoch": 0.2631426337923948,
|
| 9948 |
+
"grad_norm": 35.5625,
|
| 9949 |
+
"learning_rate": 9.958883975391858e-06,
|
| 9950 |
+
"loss": 22.4136,
|
| 9951 |
+
"step": 14180
|
| 9952 |
+
},
|
| 9953 |
+
{
|
| 9954 |
+
"epoch": 0.2633282068768746,
|
| 9955 |
+
"grad_norm": 36.6875,
|
| 9956 |
+
"learning_rate": 9.958854979605814e-06,
|
| 9957 |
+
"loss": 22.759,
|
| 9958 |
+
"step": 14190
|
| 9959 |
+
},
|
| 9960 |
+
{
|
| 9961 |
+
"epoch": 0.2635137799613544,
|
| 9962 |
+
"grad_norm": 34.5625,
|
| 9963 |
+
"learning_rate": 9.958825983819771e-06,
|
| 9964 |
+
"loss": 22.6078,
|
| 9965 |
+
"step": 14200
|
| 9966 |
+
},
|
| 9967 |
+
{
|
| 9968 |
+
"epoch": 0.26369935304583425,
|
| 9969 |
+
"grad_norm": 34.96875,
|
| 9970 |
+
"learning_rate": 9.95879698803373e-06,
|
| 9971 |
+
"loss": 22.6325,
|
| 9972 |
+
"step": 14210
|
| 9973 |
+
},
|
| 9974 |
+
{
|
| 9975 |
+
"epoch": 0.26388492613031406,
|
| 9976 |
+
"grad_norm": 36.59375,
|
| 9977 |
+
"learning_rate": 9.958767992247688e-06,
|
| 9978 |
+
"loss": 22.7356,
|
| 9979 |
+
"step": 14220
|
| 9980 |
+
},
|
| 9981 |
+
{
|
| 9982 |
+
"epoch": 0.26407049921479386,
|
| 9983 |
+
"grad_norm": 36.1875,
|
| 9984 |
+
"learning_rate": 9.958738996461645e-06,
|
| 9985 |
+
"loss": 22.3037,
|
| 9986 |
+
"step": 14230
|
| 9987 |
+
},
|
| 9988 |
+
{
|
| 9989 |
+
"epoch": 0.2642560722992737,
|
| 9990 |
+
"grad_norm": 35.75,
|
| 9991 |
+
"learning_rate": 9.958710000675603e-06,
|
| 9992 |
+
"loss": 22.3392,
|
| 9993 |
+
"step": 14240
|
| 9994 |
+
},
|
| 9995 |
+
{
|
| 9996 |
+
"epoch": 0.26444164538375353,
|
| 9997 |
+
"grad_norm": 34.0,
|
| 9998 |
+
"learning_rate": 9.95868100488956e-06,
|
| 9999 |
+
"loss": 21.9539,
|
| 10000 |
+
"step": 14250
|
| 10001 |
+
},
|
| 10002 |
+
{
|
| 10003 |
+
"epoch": 0.2646272184682334,
|
| 10004 |
+
"grad_norm": 34.0,
|
| 10005 |
+
"learning_rate": 9.958652009103518e-06,
|
| 10006 |
+
"loss": 22.5607,
|
| 10007 |
+
"step": 14260
|
| 10008 |
+
},
|
| 10009 |
+
{
|
| 10010 |
+
"epoch": 0.2648127915527132,
|
| 10011 |
+
"grad_norm": 37.40625,
|
| 10012 |
+
"learning_rate": 9.958623013317475e-06,
|
| 10013 |
+
"loss": 22.2516,
|
| 10014 |
+
"step": 14270
|
| 10015 |
+
},
|
| 10016 |
+
{
|
| 10017 |
+
"epoch": 0.264998364637193,
|
| 10018 |
+
"grad_norm": 35.875,
|
| 10019 |
+
"learning_rate": 9.958594017531434e-06,
|
| 10020 |
+
"loss": 21.6971,
|
| 10021 |
+
"step": 14280
|
| 10022 |
+
},
|
| 10023 |
+
{
|
| 10024 |
+
"epoch": 0.26518393772167287,
|
| 10025 |
+
"grad_norm": 35.46875,
|
| 10026 |
+
"learning_rate": 9.958565021745391e-06,
|
| 10027 |
+
"loss": 22.7012,
|
| 10028 |
+
"step": 14290
|
| 10029 |
+
},
|
| 10030 |
+
{
|
| 10031 |
+
"epoch": 0.2653695108061527,
|
| 10032 |
+
"grad_norm": 36.3125,
|
| 10033 |
+
"learning_rate": 9.958536025959347e-06,
|
| 10034 |
+
"loss": 22.4616,
|
| 10035 |
+
"step": 14300
|
| 10036 |
+
},
|
| 10037 |
+
{
|
| 10038 |
+
"epoch": 0.2655550838906325,
|
| 10039 |
+
"grad_norm": 37.3125,
|
| 10040 |
+
"learning_rate": 9.958507030173306e-06,
|
| 10041 |
+
"loss": 22.4242,
|
| 10042 |
+
"step": 14310
|
| 10043 |
+
},
|
| 10044 |
+
{
|
| 10045 |
+
"epoch": 0.26574065697511234,
|
| 10046 |
+
"grad_norm": 34.5625,
|
| 10047 |
+
"learning_rate": 9.958478034387264e-06,
|
| 10048 |
+
"loss": 22.3209,
|
| 10049 |
+
"step": 14320
|
| 10050 |
+
},
|
| 10051 |
+
{
|
| 10052 |
+
"epoch": 0.26592623005959215,
|
| 10053 |
+
"grad_norm": 33.09375,
|
| 10054 |
+
"learning_rate": 9.958449038601221e-06,
|
| 10055 |
+
"loss": 22.8311,
|
| 10056 |
+
"step": 14330
|
| 10057 |
+
},
|
| 10058 |
+
{
|
| 10059 |
+
"epoch": 0.26611180314407196,
|
| 10060 |
+
"grad_norm": 34.75,
|
| 10061 |
+
"learning_rate": 9.958420042815178e-06,
|
| 10062 |
+
"loss": 22.636,
|
| 10063 |
+
"step": 14340
|
| 10064 |
+
},
|
| 10065 |
+
{
|
| 10066 |
+
"epoch": 0.2662973762285518,
|
| 10067 |
+
"grad_norm": 35.78125,
|
| 10068 |
+
"learning_rate": 9.958391047029136e-06,
|
| 10069 |
+
"loss": 22.0774,
|
| 10070 |
+
"step": 14350
|
| 10071 |
+
},
|
| 10072 |
+
{
|
| 10073 |
+
"epoch": 0.2664829493130316,
|
| 10074 |
+
"grad_norm": 35.71875,
|
| 10075 |
+
"learning_rate": 9.958362051243093e-06,
|
| 10076 |
+
"loss": 22.4367,
|
| 10077 |
+
"step": 14360
|
| 10078 |
+
},
|
| 10079 |
+
{
|
| 10080 |
+
"epoch": 0.2666685223975115,
|
| 10081 |
+
"grad_norm": 35.90625,
|
| 10082 |
+
"learning_rate": 9.95833305545705e-06,
|
| 10083 |
+
"loss": 22.3071,
|
| 10084 |
+
"step": 14370
|
| 10085 |
+
},
|
| 10086 |
+
{
|
| 10087 |
+
"epoch": 0.2668540954819913,
|
| 10088 |
+
"grad_norm": 33.96875,
|
| 10089 |
+
"learning_rate": 9.95830405967101e-06,
|
| 10090 |
+
"loss": 22.3167,
|
| 10091 |
+
"step": 14380
|
| 10092 |
+
},
|
| 10093 |
+
{
|
| 10094 |
+
"epoch": 0.2670396685664711,
|
| 10095 |
+
"grad_norm": 35.5,
|
| 10096 |
+
"learning_rate": 9.958275063884967e-06,
|
| 10097 |
+
"loss": 22.671,
|
| 10098 |
+
"step": 14390
|
| 10099 |
+
},
|
| 10100 |
+
{
|
| 10101 |
+
"epoch": 0.26722524165095096,
|
| 10102 |
+
"grad_norm": 35.1875,
|
| 10103 |
+
"learning_rate": 9.958246068098923e-06,
|
| 10104 |
+
"loss": 22.7599,
|
| 10105 |
+
"step": 14400
|
| 10106 |
+
},
|
| 10107 |
+
{
|
| 10108 |
+
"epoch": 0.26741081473543077,
|
| 10109 |
+
"grad_norm": 36.8125,
|
| 10110 |
+
"learning_rate": 9.958217072312882e-06,
|
| 10111 |
+
"loss": 22.1926,
|
| 10112 |
+
"step": 14410
|
| 10113 |
+
},
|
| 10114 |
+
{
|
| 10115 |
+
"epoch": 0.2675963878199106,
|
| 10116 |
+
"grad_norm": 36.1875,
|
| 10117 |
+
"learning_rate": 9.95818807652684e-06,
|
| 10118 |
+
"loss": 22.4728,
|
| 10119 |
+
"step": 14420
|
| 10120 |
+
},
|
| 10121 |
+
{
|
| 10122 |
+
"epoch": 0.26778196090439044,
|
| 10123 |
+
"grad_norm": 35.875,
|
| 10124 |
+
"learning_rate": 9.958159080740797e-06,
|
| 10125 |
+
"loss": 22.0861,
|
| 10126 |
+
"step": 14430
|
| 10127 |
+
},
|
| 10128 |
+
{
|
| 10129 |
+
"epoch": 0.26796753398887024,
|
| 10130 |
+
"grad_norm": 38.03125,
|
| 10131 |
+
"learning_rate": 9.958130084954754e-06,
|
| 10132 |
+
"loss": 22.6352,
|
| 10133 |
+
"step": 14440
|
| 10134 |
+
},
|
| 10135 |
+
{
|
| 10136 |
+
"epoch": 0.2681531070733501,
|
| 10137 |
+
"grad_norm": 34.21875,
|
| 10138 |
+
"learning_rate": 9.958101089168712e-06,
|
| 10139 |
+
"loss": 22.3403,
|
| 10140 |
+
"step": 14450
|
| 10141 |
+
},
|
| 10142 |
+
{
|
| 10143 |
+
"epoch": 0.2683386801578299,
|
| 10144 |
+
"grad_norm": 34.0,
|
| 10145 |
+
"learning_rate": 9.958072093382669e-06,
|
| 10146 |
+
"loss": 22.5642,
|
| 10147 |
+
"step": 14460
|
| 10148 |
+
},
|
| 10149 |
+
{
|
| 10150 |
+
"epoch": 0.2685242532423097,
|
| 10151 |
+
"grad_norm": 34.71875,
|
| 10152 |
+
"learning_rate": 9.958043097596626e-06,
|
| 10153 |
+
"loss": 22.4688,
|
| 10154 |
+
"step": 14470
|
| 10155 |
+
},
|
| 10156 |
+
{
|
| 10157 |
+
"epoch": 0.2687098263267896,
|
| 10158 |
+
"grad_norm": 33.0625,
|
| 10159 |
+
"learning_rate": 9.958014101810584e-06,
|
| 10160 |
+
"loss": 22.3117,
|
| 10161 |
+
"step": 14480
|
| 10162 |
+
},
|
| 10163 |
+
{
|
| 10164 |
+
"epoch": 0.2688953994112694,
|
| 10165 |
+
"grad_norm": 34.0625,
|
| 10166 |
+
"learning_rate": 9.957985106024543e-06,
|
| 10167 |
+
"loss": 22.4656,
|
| 10168 |
+
"step": 14490
|
| 10169 |
+
},
|
| 10170 |
+
{
|
| 10171 |
+
"epoch": 0.2690809724957492,
|
| 10172 |
+
"grad_norm": 33.90625,
|
| 10173 |
+
"learning_rate": 9.9579561102385e-06,
|
| 10174 |
+
"loss": 22.5187,
|
| 10175 |
+
"step": 14500
|
| 10176 |
+
},
|
| 10177 |
+
{
|
| 10178 |
+
"epoch": 0.26926654558022906,
|
| 10179 |
+
"grad_norm": 35.84375,
|
| 10180 |
+
"learning_rate": 9.957927114452458e-06,
|
| 10181 |
+
"loss": 22.3564,
|
| 10182 |
+
"step": 14510
|
| 10183 |
+
},
|
| 10184 |
+
{
|
| 10185 |
+
"epoch": 0.26945211866470886,
|
| 10186 |
+
"grad_norm": 34.34375,
|
| 10187 |
+
"learning_rate": 9.957898118666415e-06,
|
| 10188 |
+
"loss": 22.2059,
|
| 10189 |
+
"step": 14520
|
| 10190 |
+
},
|
| 10191 |
+
{
|
| 10192 |
+
"epoch": 0.2696376917491887,
|
| 10193 |
+
"grad_norm": 36.65625,
|
| 10194 |
+
"learning_rate": 9.957869122880373e-06,
|
| 10195 |
+
"loss": 22.3464,
|
| 10196 |
+
"step": 14530
|
| 10197 |
+
},
|
| 10198 |
+
{
|
| 10199 |
+
"epoch": 0.26982326483366853,
|
| 10200 |
+
"grad_norm": 34.71875,
|
| 10201 |
+
"learning_rate": 9.95784012709433e-06,
|
| 10202 |
+
"loss": 22.0195,
|
| 10203 |
+
"step": 14540
|
| 10204 |
+
},
|
| 10205 |
+
{
|
| 10206 |
+
"epoch": 0.27000883791814834,
|
| 10207 |
+
"grad_norm": 33.46875,
|
| 10208 |
+
"learning_rate": 9.957811131308287e-06,
|
| 10209 |
+
"loss": 22.4502,
|
| 10210 |
+
"step": 14550
|
| 10211 |
+
},
|
| 10212 |
+
{
|
| 10213 |
+
"epoch": 0.2701944110026282,
|
| 10214 |
+
"grad_norm": 35.25,
|
| 10215 |
+
"learning_rate": 9.957782135522245e-06,
|
| 10216 |
+
"loss": 22.4365,
|
| 10217 |
+
"step": 14560
|
| 10218 |
+
},
|
| 10219 |
+
{
|
| 10220 |
+
"epoch": 0.270379984087108,
|
| 10221 |
+
"grad_norm": 35.5,
|
| 10222 |
+
"learning_rate": 9.957753139736202e-06,
|
| 10223 |
+
"loss": 22.2782,
|
| 10224 |
+
"step": 14570
|
| 10225 |
+
},
|
| 10226 |
+
{
|
| 10227 |
+
"epoch": 0.2705655571715878,
|
| 10228 |
+
"grad_norm": 35.03125,
|
| 10229 |
+
"learning_rate": 9.95772414395016e-06,
|
| 10230 |
+
"loss": 22.4778,
|
| 10231 |
+
"step": 14580
|
| 10232 |
+
},
|
| 10233 |
+
{
|
| 10234 |
+
"epoch": 0.2707511302560677,
|
| 10235 |
+
"grad_norm": 35.25,
|
| 10236 |
+
"learning_rate": 9.957695148164119e-06,
|
| 10237 |
+
"loss": 22.5201,
|
| 10238 |
+
"step": 14590
|
| 10239 |
+
},
|
| 10240 |
+
{
|
| 10241 |
+
"epoch": 0.2709367033405475,
|
| 10242 |
+
"grad_norm": 37.84375,
|
| 10243 |
+
"learning_rate": 9.957666152378076e-06,
|
| 10244 |
+
"loss": 22.5597,
|
| 10245 |
+
"step": 14600
|
| 10246 |
+
},
|
| 10247 |
+
{
|
| 10248 |
+
"epoch": 0.2711222764250273,
|
| 10249 |
+
"grad_norm": 35.03125,
|
| 10250 |
+
"learning_rate": 9.957637156592034e-06,
|
| 10251 |
+
"loss": 22.3051,
|
| 10252 |
+
"step": 14610
|
| 10253 |
+
},
|
| 10254 |
+
{
|
| 10255 |
+
"epoch": 0.27130784950950715,
|
| 10256 |
+
"grad_norm": 35.0,
|
| 10257 |
+
"learning_rate": 9.957608160805991e-06,
|
| 10258 |
+
"loss": 22.3427,
|
| 10259 |
+
"step": 14620
|
| 10260 |
+
},
|
| 10261 |
+
{
|
| 10262 |
+
"epoch": 0.27149342259398695,
|
| 10263 |
+
"grad_norm": 36.65625,
|
| 10264 |
+
"learning_rate": 9.957579165019948e-06,
|
| 10265 |
+
"loss": 22.1068,
|
| 10266 |
+
"step": 14630
|
| 10267 |
+
},
|
| 10268 |
+
{
|
| 10269 |
+
"epoch": 0.2716789956784668,
|
| 10270 |
+
"grad_norm": 35.71875,
|
| 10271 |
+
"learning_rate": 9.957550169233906e-06,
|
| 10272 |
+
"loss": 22.4494,
|
| 10273 |
+
"step": 14640
|
| 10274 |
+
},
|
| 10275 |
+
{
|
| 10276 |
+
"epoch": 0.2718645687629466,
|
| 10277 |
+
"grad_norm": 32.84375,
|
| 10278 |
+
"learning_rate": 9.957521173447863e-06,
|
| 10279 |
+
"loss": 22.6318,
|
| 10280 |
+
"step": 14650
|
| 10281 |
+
},
|
| 10282 |
+
{
|
| 10283 |
+
"epoch": 0.27205014184742643,
|
| 10284 |
+
"grad_norm": 33.78125,
|
| 10285 |
+
"learning_rate": 9.957492177661822e-06,
|
| 10286 |
+
"loss": 22.7068,
|
| 10287 |
+
"step": 14660
|
| 10288 |
+
},
|
| 10289 |
+
{
|
| 10290 |
+
"epoch": 0.2722357149319063,
|
| 10291 |
+
"grad_norm": 35.09375,
|
| 10292 |
+
"learning_rate": 9.957463181875778e-06,
|
| 10293 |
+
"loss": 21.9763,
|
| 10294 |
+
"step": 14670
|
| 10295 |
+
},
|
| 10296 |
+
{
|
| 10297 |
+
"epoch": 0.2724212880163861,
|
| 10298 |
+
"grad_norm": 35.625,
|
| 10299 |
+
"learning_rate": 9.957434186089735e-06,
|
| 10300 |
+
"loss": 22.4956,
|
| 10301 |
+
"step": 14680
|
| 10302 |
+
},
|
| 10303 |
+
{
|
| 10304 |
+
"epoch": 0.2726068611008659,
|
| 10305 |
+
"grad_norm": 33.40625,
|
| 10306 |
+
"learning_rate": 9.957405190303694e-06,
|
| 10307 |
+
"loss": 22.6001,
|
| 10308 |
+
"step": 14690
|
| 10309 |
+
},
|
| 10310 |
+
{
|
| 10311 |
+
"epoch": 0.27279243418534577,
|
| 10312 |
+
"grad_norm": 35.375,
|
| 10313 |
+
"learning_rate": 9.957376194517652e-06,
|
| 10314 |
+
"loss": 22.5121,
|
| 10315 |
+
"step": 14700
|
| 10316 |
+
},
|
| 10317 |
+
{
|
| 10318 |
+
"epoch": 0.2729780072698256,
|
| 10319 |
+
"grad_norm": 34.34375,
|
| 10320 |
+
"learning_rate": 9.95734719873161e-06,
|
| 10321 |
+
"loss": 22.2686,
|
| 10322 |
+
"step": 14710
|
| 10323 |
+
},
|
| 10324 |
+
{
|
| 10325 |
+
"epoch": 0.27316358035430544,
|
| 10326 |
+
"grad_norm": 34.90625,
|
| 10327 |
+
"learning_rate": 9.957318202945567e-06,
|
| 10328 |
+
"loss": 22.4447,
|
| 10329 |
+
"step": 14720
|
| 10330 |
+
},
|
| 10331 |
+
{
|
| 10332 |
+
"epoch": 0.27334915343878524,
|
| 10333 |
+
"grad_norm": 37.34375,
|
| 10334 |
+
"learning_rate": 9.957289207159524e-06,
|
| 10335 |
+
"loss": 22.6572,
|
| 10336 |
+
"step": 14730
|
| 10337 |
+
},
|
| 10338 |
+
{
|
| 10339 |
+
"epoch": 0.27353472652326505,
|
| 10340 |
+
"grad_norm": 35.5,
|
| 10341 |
+
"learning_rate": 9.957260211373482e-06,
|
| 10342 |
+
"loss": 22.6281,
|
| 10343 |
+
"step": 14740
|
| 10344 |
+
},
|
| 10345 |
+
{
|
| 10346 |
+
"epoch": 0.2737202996077449,
|
| 10347 |
+
"grad_norm": 36.125,
|
| 10348 |
+
"learning_rate": 9.957231215587439e-06,
|
| 10349 |
+
"loss": 22.1592,
|
| 10350 |
+
"step": 14750
|
| 10351 |
+
},
|
| 10352 |
+
{
|
| 10353 |
+
"epoch": 0.2739058726922247,
|
| 10354 |
+
"grad_norm": 35.46875,
|
| 10355 |
+
"learning_rate": 9.957202219801398e-06,
|
| 10356 |
+
"loss": 22.5348,
|
| 10357 |
+
"step": 14760
|
| 10358 |
+
},
|
| 10359 |
+
{
|
| 10360 |
+
"epoch": 0.2740914457767045,
|
| 10361 |
+
"grad_norm": 36.3125,
|
| 10362 |
+
"learning_rate": 9.957173224015355e-06,
|
| 10363 |
+
"loss": 22.4853,
|
| 10364 |
+
"step": 14770
|
| 10365 |
+
},
|
| 10366 |
+
{
|
| 10367 |
+
"epoch": 0.2742770188611844,
|
| 10368 |
+
"grad_norm": 35.84375,
|
| 10369 |
+
"learning_rate": 9.957144228229311e-06,
|
| 10370 |
+
"loss": 22.3393,
|
| 10371 |
+
"step": 14780
|
| 10372 |
+
},
|
| 10373 |
+
{
|
| 10374 |
+
"epoch": 0.2744625919456642,
|
| 10375 |
+
"grad_norm": 34.65625,
|
| 10376 |
+
"learning_rate": 9.95711523244327e-06,
|
| 10377 |
+
"loss": 22.3089,
|
| 10378 |
+
"step": 14790
|
| 10379 |
+
},
|
| 10380 |
+
{
|
| 10381 |
+
"epoch": 0.27464816503014405,
|
| 10382 |
+
"grad_norm": 34.5625,
|
| 10383 |
+
"learning_rate": 9.957086236657228e-06,
|
| 10384 |
+
"loss": 22.4888,
|
| 10385 |
+
"step": 14800
|
| 10386 |
+
},
|
| 10387 |
+
{
|
| 10388 |
+
"epoch": 0.27483373811462386,
|
| 10389 |
+
"grad_norm": 36.34375,
|
| 10390 |
+
"learning_rate": 9.957057240871185e-06,
|
| 10391 |
+
"loss": 21.9885,
|
| 10392 |
+
"step": 14810
|
| 10393 |
+
},
|
| 10394 |
+
{
|
| 10395 |
+
"epoch": 0.27501931119910367,
|
| 10396 |
+
"grad_norm": 33.75,
|
| 10397 |
+
"learning_rate": 9.957028245085142e-06,
|
| 10398 |
+
"loss": 22.3091,
|
| 10399 |
+
"step": 14820
|
| 10400 |
+
},
|
| 10401 |
+
{
|
| 10402 |
+
"epoch": 0.27520488428358353,
|
| 10403 |
+
"grad_norm": 35.59375,
|
| 10404 |
+
"learning_rate": 9.9569992492991e-06,
|
| 10405 |
+
"loss": 22.2194,
|
| 10406 |
+
"step": 14830
|
| 10407 |
+
},
|
| 10408 |
+
{
|
| 10409 |
+
"epoch": 0.27539045736806333,
|
| 10410 |
+
"grad_norm": 36.96875,
|
| 10411 |
+
"learning_rate": 9.956970253513057e-06,
|
| 10412 |
+
"loss": 22.424,
|
| 10413 |
+
"step": 14840
|
| 10414 |
+
},
|
| 10415 |
+
{
|
| 10416 |
+
"epoch": 0.27557603045254314,
|
| 10417 |
+
"grad_norm": 35.875,
|
| 10418 |
+
"learning_rate": 9.956941257727015e-06,
|
| 10419 |
+
"loss": 22.548,
|
| 10420 |
+
"step": 14850
|
| 10421 |
+
},
|
| 10422 |
+
{
|
| 10423 |
+
"epoch": 0.275761603537023,
|
| 10424 |
+
"grad_norm": 33.6875,
|
| 10425 |
+
"learning_rate": 9.956912261940974e-06,
|
| 10426 |
+
"loss": 22.2094,
|
| 10427 |
+
"step": 14860
|
| 10428 |
+
},
|
| 10429 |
+
{
|
| 10430 |
+
"epoch": 0.2759471766215028,
|
| 10431 |
+
"grad_norm": 36.21875,
|
| 10432 |
+
"learning_rate": 9.956883266154931e-06,
|
| 10433 |
+
"loss": 22.644,
|
| 10434 |
+
"step": 14870
|
| 10435 |
+
},
|
| 10436 |
+
{
|
| 10437 |
+
"epoch": 0.27613274970598267,
|
| 10438 |
+
"grad_norm": 36.1875,
|
| 10439 |
+
"learning_rate": 9.956854270368889e-06,
|
| 10440 |
+
"loss": 22.7149,
|
| 10441 |
+
"step": 14880
|
| 10442 |
+
},
|
| 10443 |
+
{
|
| 10444 |
+
"epoch": 0.2763183227904625,
|
| 10445 |
+
"grad_norm": 34.9375,
|
| 10446 |
+
"learning_rate": 9.956825274582846e-06,
|
| 10447 |
+
"loss": 22.0443,
|
| 10448 |
+
"step": 14890
|
| 10449 |
+
},
|
| 10450 |
+
{
|
| 10451 |
+
"epoch": 0.2765038958749423,
|
| 10452 |
+
"grad_norm": 34.59375,
|
| 10453 |
+
"learning_rate": 9.956796278796803e-06,
|
| 10454 |
+
"loss": 22.2534,
|
| 10455 |
+
"step": 14900
|
| 10456 |
+
},
|
| 10457 |
+
{
|
| 10458 |
+
"epoch": 0.27668946895942215,
|
| 10459 |
+
"grad_norm": 34.875,
|
| 10460 |
+
"learning_rate": 9.95676728301076e-06,
|
| 10461 |
+
"loss": 22.5268,
|
| 10462 |
+
"step": 14910
|
| 10463 |
+
},
|
| 10464 |
+
{
|
| 10465 |
+
"epoch": 0.27687504204390195,
|
| 10466 |
+
"grad_norm": 36.3125,
|
| 10467 |
+
"learning_rate": 9.956738287224718e-06,
|
| 10468 |
+
"loss": 22.2169,
|
| 10469 |
+
"step": 14920
|
| 10470 |
+
},
|
| 10471 |
+
{
|
| 10472 |
+
"epoch": 0.27706061512838176,
|
| 10473 |
+
"grad_norm": 34.46875,
|
| 10474 |
+
"learning_rate": 9.956709291438676e-06,
|
| 10475 |
+
"loss": 22.2188,
|
| 10476 |
+
"step": 14930
|
| 10477 |
+
},
|
| 10478 |
+
{
|
| 10479 |
+
"epoch": 0.2772461882128616,
|
| 10480 |
+
"grad_norm": 33.625,
|
| 10481 |
+
"learning_rate": 9.956680295652633e-06,
|
| 10482 |
+
"loss": 22.4894,
|
| 10483 |
+
"step": 14940
|
| 10484 |
+
},
|
| 10485 |
+
{
|
| 10486 |
+
"epoch": 0.27743176129734143,
|
| 10487 |
+
"grad_norm": 34.53125,
|
| 10488 |
+
"learning_rate": 9.95665129986659e-06,
|
| 10489 |
+
"loss": 21.8868,
|
| 10490 |
+
"step": 14950
|
| 10491 |
+
},
|
| 10492 |
+
{
|
| 10493 |
+
"epoch": 0.27761733438182123,
|
| 10494 |
+
"grad_norm": 35.6875,
|
| 10495 |
+
"learning_rate": 9.95662230408055e-06,
|
| 10496 |
+
"loss": 22.3338,
|
| 10497 |
+
"step": 14960
|
| 10498 |
+
},
|
| 10499 |
+
{
|
| 10500 |
+
"epoch": 0.2778029074663011,
|
| 10501 |
+
"grad_norm": 34.4375,
|
| 10502 |
+
"learning_rate": 9.956593308294507e-06,
|
| 10503 |
+
"loss": 22.4738,
|
| 10504 |
+
"step": 14970
|
| 10505 |
+
},
|
| 10506 |
+
{
|
| 10507 |
+
"epoch": 0.2779884805507809,
|
| 10508 |
+
"grad_norm": 36.6875,
|
| 10509 |
+
"learning_rate": 9.956564312508464e-06,
|
| 10510 |
+
"loss": 22.3862,
|
| 10511 |
+
"step": 14980
|
| 10512 |
+
},
|
| 10513 |
+
{
|
| 10514 |
+
"epoch": 0.27817405363526077,
|
| 10515 |
+
"grad_norm": 33.96875,
|
| 10516 |
+
"learning_rate": 9.956535316722422e-06,
|
| 10517 |
+
"loss": 21.9725,
|
| 10518 |
+
"step": 14990
|
| 10519 |
+
},
|
| 10520 |
+
{
|
| 10521 |
+
"epoch": 0.27835962671974057,
|
| 10522 |
+
"grad_norm": 34.5,
|
| 10523 |
+
"learning_rate": 9.95650632093638e-06,
|
| 10524 |
+
"loss": 22.3186,
|
| 10525 |
+
"step": 15000
|
| 10526 |
+
},
|
| 10527 |
+
{
|
| 10528 |
+
"epoch": 0.27835962671974057,
|
| 10529 |
+
"eval_loss": 2.7910494804382324,
|
| 10530 |
+
"eval_runtime": 454.4569,
|
| 10531 |
+
"eval_samples_per_second": 3195.28,
|
| 10532 |
+
"eval_steps_per_second": 49.928,
|
| 10533 |
+
"step": 15000
|
| 10534 |
}
|
| 10535 |
],
|
| 10536 |
"logging_steps": 10,
|
|
|
|
| 10550 |
"attributes": {}
|
| 10551 |
}
|
| 10552 |
},
|
| 10553 |
+
"total_flos": 2.618284138561536e+18,
|
| 10554 |
"train_batch_size": 8,
|
| 10555 |
"trial_name": null,
|
| 10556 |
"trial_params": null
|