Training in progress, step 7500, checkpoint
Browse files- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state_0.pth +2 -2
- 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 +1761 -3
last-checkpoint/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 368988278
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:67e015b0f596fee442abd71af93158ddf702d0b03ddbafe627cd1e3991ffb717
|
3 |
size 368988278
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1107079290
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:500d83d312df8ef8772133628901aa3ec8b169d37b27727f6771ed38095522d3
|
3 |
size 1107079290
|
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
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:69bcb5a0833345d8383f742fc9fcb902b5c0cae43ce63aee4b910dcd39dcda65
|
3 |
+
size 15920
|
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 sha256:13f8e14c2c7c91bc7deff16a401fb72a8859494fc7925b25ee4ffb4e04a50c64
|
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": 2500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -3523,6 +3523,1764 @@
|
|
3523 |
"eval_samples_per_second": 1567.295,
|
3524 |
"eval_steps_per_second": 48.978,
|
3525 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3526 |
}
|
3527 |
],
|
3528 |
"logging_steps": 10,
|
@@ -3542,7 +5300,7 @@
|
|
3542 |
"attributes": {}
|
3543 |
}
|
3544 |
},
|
3545 |
-
"total_flos":
|
3546 |
"train_batch_size": 4,
|
3547 |
"trial_name": null,
|
3548 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.6002318395480254,
|
5 |
"eval_steps": 2500,
|
6 |
+
"global_step": 7500,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3523 |
"eval_samples_per_second": 1567.295,
|
3524 |
"eval_steps_per_second": 48.978,
|
3525 |
"step": 5000
|
3526 |
+
},
|
3527 |
+
{
|
3528 |
+
"epoch": 0.40095486881808096,
|
3529 |
+
"grad_norm": 72.4375,
|
3530 |
+
"learning_rate": 9.992168853890664e-07,
|
3531 |
+
"loss": 133.3489,
|
3532 |
+
"step": 5010
|
3533 |
+
},
|
3534 |
+
{
|
3535 |
+
"epoch": 0.40175517793747834,
|
3536 |
+
"grad_norm": 74.5625,
|
3537 |
+
"learning_rate": 9.992153222860506e-07,
|
3538 |
+
"loss": 132.38,
|
3539 |
+
"step": 5020
|
3540 |
+
},
|
3541 |
+
{
|
3542 |
+
"epoch": 0.4025554870568757,
|
3543 |
+
"grad_norm": 81.25,
|
3544 |
+
"learning_rate": 9.992137591830347e-07,
|
3545 |
+
"loss": 133.7667,
|
3546 |
+
"step": 5030
|
3547 |
+
},
|
3548 |
+
{
|
3549 |
+
"epoch": 0.4033557961762731,
|
3550 |
+
"grad_norm": 74.5625,
|
3551 |
+
"learning_rate": 9.992121960800189e-07,
|
3552 |
+
"loss": 133.8293,
|
3553 |
+
"step": 5040
|
3554 |
+
},
|
3555 |
+
{
|
3556 |
+
"epoch": 0.4041561052956705,
|
3557 |
+
"grad_norm": 71.25,
|
3558 |
+
"learning_rate": 9.99210632977003e-07,
|
3559 |
+
"loss": 133.3581,
|
3560 |
+
"step": 5050
|
3561 |
+
},
|
3562 |
+
{
|
3563 |
+
"epoch": 0.4049564144150678,
|
3564 |
+
"grad_norm": 79.5625,
|
3565 |
+
"learning_rate": 9.992090698739873e-07,
|
3566 |
+
"loss": 132.9888,
|
3567 |
+
"step": 5060
|
3568 |
+
},
|
3569 |
+
{
|
3570 |
+
"epoch": 0.4057567235344652,
|
3571 |
+
"grad_norm": 73.9375,
|
3572 |
+
"learning_rate": 9.992075067709715e-07,
|
3573 |
+
"loss": 132.4104,
|
3574 |
+
"step": 5070
|
3575 |
+
},
|
3576 |
+
{
|
3577 |
+
"epoch": 0.40655703265386256,
|
3578 |
+
"grad_norm": 72.5,
|
3579 |
+
"learning_rate": 9.992059436679555e-07,
|
3580 |
+
"loss": 132.3801,
|
3581 |
+
"step": 5080
|
3582 |
+
},
|
3583 |
+
{
|
3584 |
+
"epoch": 0.40735734177325994,
|
3585 |
+
"grad_norm": 76.3125,
|
3586 |
+
"learning_rate": 9.992043805649397e-07,
|
3587 |
+
"loss": 132.4045,
|
3588 |
+
"step": 5090
|
3589 |
+
},
|
3590 |
+
{
|
3591 |
+
"epoch": 0.40815765089265726,
|
3592 |
+
"grad_norm": 76.5625,
|
3593 |
+
"learning_rate": 9.99202817461924e-07,
|
3594 |
+
"loss": 132.4824,
|
3595 |
+
"step": 5100
|
3596 |
+
},
|
3597 |
+
{
|
3598 |
+
"epoch": 0.40895796001205464,
|
3599 |
+
"grad_norm": 74.125,
|
3600 |
+
"learning_rate": 9.992012543589082e-07,
|
3601 |
+
"loss": 132.0051,
|
3602 |
+
"step": 5110
|
3603 |
+
},
|
3604 |
+
{
|
3605 |
+
"epoch": 0.409758269131452,
|
3606 |
+
"grad_norm": 72.9375,
|
3607 |
+
"learning_rate": 9.991996912558924e-07,
|
3608 |
+
"loss": 133.8627,
|
3609 |
+
"step": 5120
|
3610 |
+
},
|
3611 |
+
{
|
3612 |
+
"epoch": 0.4105585782508494,
|
3613 |
+
"grad_norm": 78.9375,
|
3614 |
+
"learning_rate": 9.991981281528764e-07,
|
3615 |
+
"loss": 132.2818,
|
3616 |
+
"step": 5130
|
3617 |
+
},
|
3618 |
+
{
|
3619 |
+
"epoch": 0.4113588873702468,
|
3620 |
+
"grad_norm": 75.375,
|
3621 |
+
"learning_rate": 9.991965650498606e-07,
|
3622 |
+
"loss": 131.2578,
|
3623 |
+
"step": 5140
|
3624 |
+
},
|
3625 |
+
{
|
3626 |
+
"epoch": 0.4121591964896441,
|
3627 |
+
"grad_norm": 77.125,
|
3628 |
+
"learning_rate": 9.991950019468448e-07,
|
3629 |
+
"loss": 133.7901,
|
3630 |
+
"step": 5150
|
3631 |
+
},
|
3632 |
+
{
|
3633 |
+
"epoch": 0.4129595056090415,
|
3634 |
+
"grad_norm": 80.9375,
|
3635 |
+
"learning_rate": 9.991934388438289e-07,
|
3636 |
+
"loss": 132.6137,
|
3637 |
+
"step": 5160
|
3638 |
+
},
|
3639 |
+
{
|
3640 |
+
"epoch": 0.41375981472843887,
|
3641 |
+
"grad_norm": 81.875,
|
3642 |
+
"learning_rate": 9.99191875740813e-07,
|
3643 |
+
"loss": 132.9288,
|
3644 |
+
"step": 5170
|
3645 |
+
},
|
3646 |
+
{
|
3647 |
+
"epoch": 0.41456012384783625,
|
3648 |
+
"grad_norm": 78.6875,
|
3649 |
+
"learning_rate": 9.991903126377973e-07,
|
3650 |
+
"loss": 132.9595,
|
3651 |
+
"step": 5180
|
3652 |
+
},
|
3653 |
+
{
|
3654 |
+
"epoch": 0.4153604329672336,
|
3655 |
+
"grad_norm": 76.9375,
|
3656 |
+
"learning_rate": 9.991887495347813e-07,
|
3657 |
+
"loss": 132.3299,
|
3658 |
+
"step": 5190
|
3659 |
+
},
|
3660 |
+
{
|
3661 |
+
"epoch": 0.41616074208663095,
|
3662 |
+
"grad_norm": 79.0,
|
3663 |
+
"learning_rate": 9.991871864317655e-07,
|
3664 |
+
"loss": 132.1807,
|
3665 |
+
"step": 5200
|
3666 |
+
},
|
3667 |
+
{
|
3668 |
+
"epoch": 0.41696105120602833,
|
3669 |
+
"grad_norm": 75.4375,
|
3670 |
+
"learning_rate": 9.991856233287497e-07,
|
3671 |
+
"loss": 133.9722,
|
3672 |
+
"step": 5210
|
3673 |
+
},
|
3674 |
+
{
|
3675 |
+
"epoch": 0.4177613603254257,
|
3676 |
+
"grad_norm": 73.4375,
|
3677 |
+
"learning_rate": 9.99184060225734e-07,
|
3678 |
+
"loss": 132.654,
|
3679 |
+
"step": 5220
|
3680 |
+
},
|
3681 |
+
{
|
3682 |
+
"epoch": 0.41856166944482304,
|
3683 |
+
"grad_norm": 74.1875,
|
3684 |
+
"learning_rate": 9.991824971227182e-07,
|
3685 |
+
"loss": 133.0139,
|
3686 |
+
"step": 5230
|
3687 |
+
},
|
3688 |
+
{
|
3689 |
+
"epoch": 0.4193619785642204,
|
3690 |
+
"grad_norm": 79.25,
|
3691 |
+
"learning_rate": 9.991809340197022e-07,
|
3692 |
+
"loss": 131.9049,
|
3693 |
+
"step": 5240
|
3694 |
+
},
|
3695 |
+
{
|
3696 |
+
"epoch": 0.4201622876836178,
|
3697 |
+
"grad_norm": 75.8125,
|
3698 |
+
"learning_rate": 9.991793709166864e-07,
|
3699 |
+
"loss": 131.9633,
|
3700 |
+
"step": 5250
|
3701 |
+
},
|
3702 |
+
{
|
3703 |
+
"epoch": 0.4209625968030152,
|
3704 |
+
"grad_norm": 74.25,
|
3705 |
+
"learning_rate": 9.991778078136706e-07,
|
3706 |
+
"loss": 132.2361,
|
3707 |
+
"step": 5260
|
3708 |
+
},
|
3709 |
+
{
|
3710 |
+
"epoch": 0.42176290592241256,
|
3711 |
+
"grad_norm": 72.8125,
|
3712 |
+
"learning_rate": 9.991762447106548e-07,
|
3713 |
+
"loss": 131.267,
|
3714 |
+
"step": 5270
|
3715 |
+
},
|
3716 |
+
{
|
3717 |
+
"epoch": 0.4225632150418099,
|
3718 |
+
"grad_norm": 77.5625,
|
3719 |
+
"learning_rate": 9.991746816076388e-07,
|
3720 |
+
"loss": 132.4257,
|
3721 |
+
"step": 5280
|
3722 |
+
},
|
3723 |
+
{
|
3724 |
+
"epoch": 0.42336352416120726,
|
3725 |
+
"grad_norm": 79.3125,
|
3726 |
+
"learning_rate": 9.99173118504623e-07,
|
3727 |
+
"loss": 132.785,
|
3728 |
+
"step": 5290
|
3729 |
+
},
|
3730 |
+
{
|
3731 |
+
"epoch": 0.42416383328060464,
|
3732 |
+
"grad_norm": 76.9375,
|
3733 |
+
"learning_rate": 9.991715554016073e-07,
|
3734 |
+
"loss": 131.3379,
|
3735 |
+
"step": 5300
|
3736 |
+
},
|
3737 |
+
{
|
3738 |
+
"epoch": 0.424964142400002,
|
3739 |
+
"grad_norm": 75.25,
|
3740 |
+
"learning_rate": 9.991699922985915e-07,
|
3741 |
+
"loss": 133.1422,
|
3742 |
+
"step": 5310
|
3743 |
+
},
|
3744 |
+
{
|
3745 |
+
"epoch": 0.42576445151939935,
|
3746 |
+
"grad_norm": 72.5625,
|
3747 |
+
"learning_rate": 9.991684291955755e-07,
|
3748 |
+
"loss": 132.1037,
|
3749 |
+
"step": 5320
|
3750 |
+
},
|
3751 |
+
{
|
3752 |
+
"epoch": 0.4265647606387967,
|
3753 |
+
"grad_norm": 79.0,
|
3754 |
+
"learning_rate": 9.991668660925597e-07,
|
3755 |
+
"loss": 132.1573,
|
3756 |
+
"step": 5330
|
3757 |
+
},
|
3758 |
+
{
|
3759 |
+
"epoch": 0.4273650697581941,
|
3760 |
+
"grad_norm": 75.0625,
|
3761 |
+
"learning_rate": 9.99165302989544e-07,
|
3762 |
+
"loss": 132.0645,
|
3763 |
+
"step": 5340
|
3764 |
+
},
|
3765 |
+
{
|
3766 |
+
"epoch": 0.4281653788775915,
|
3767 |
+
"grad_norm": 72.3125,
|
3768 |
+
"learning_rate": 9.99163739886528e-07,
|
3769 |
+
"loss": 132.8727,
|
3770 |
+
"step": 5350
|
3771 |
+
},
|
3772 |
+
{
|
3773 |
+
"epoch": 0.4289656879969888,
|
3774 |
+
"grad_norm": 77.4375,
|
3775 |
+
"learning_rate": 9.991621767835122e-07,
|
3776 |
+
"loss": 133.5671,
|
3777 |
+
"step": 5360
|
3778 |
+
},
|
3779 |
+
{
|
3780 |
+
"epoch": 0.4297659971163862,
|
3781 |
+
"grad_norm": 74.3125,
|
3782 |
+
"learning_rate": 9.991606136804964e-07,
|
3783 |
+
"loss": 132.3078,
|
3784 |
+
"step": 5370
|
3785 |
+
},
|
3786 |
+
{
|
3787 |
+
"epoch": 0.43056630623578357,
|
3788 |
+
"grad_norm": 81.375,
|
3789 |
+
"learning_rate": 9.991590505774806e-07,
|
3790 |
+
"loss": 133.0639,
|
3791 |
+
"step": 5380
|
3792 |
+
},
|
3793 |
+
{
|
3794 |
+
"epoch": 0.43136661535518095,
|
3795 |
+
"grad_norm": 74.0625,
|
3796 |
+
"learning_rate": 9.991574874744648e-07,
|
3797 |
+
"loss": 131.2669,
|
3798 |
+
"step": 5390
|
3799 |
+
},
|
3800 |
+
{
|
3801 |
+
"epoch": 0.43216692447457833,
|
3802 |
+
"grad_norm": 79.375,
|
3803 |
+
"learning_rate": 9.991559243714488e-07,
|
3804 |
+
"loss": 131.9368,
|
3805 |
+
"step": 5400
|
3806 |
+
},
|
3807 |
+
{
|
3808 |
+
"epoch": 0.43296723359397565,
|
3809 |
+
"grad_norm": 77.9375,
|
3810 |
+
"learning_rate": 9.99154361268433e-07,
|
3811 |
+
"loss": 132.4038,
|
3812 |
+
"step": 5410
|
3813 |
+
},
|
3814 |
+
{
|
3815 |
+
"epoch": 0.43376754271337303,
|
3816 |
+
"grad_norm": 83.375,
|
3817 |
+
"learning_rate": 9.991527981654173e-07,
|
3818 |
+
"loss": 132.8988,
|
3819 |
+
"step": 5420
|
3820 |
+
},
|
3821 |
+
{
|
3822 |
+
"epoch": 0.4345678518327704,
|
3823 |
+
"grad_norm": 73.6875,
|
3824 |
+
"learning_rate": 9.991512350624015e-07,
|
3825 |
+
"loss": 133.3637,
|
3826 |
+
"step": 5430
|
3827 |
+
},
|
3828 |
+
{
|
3829 |
+
"epoch": 0.4353681609521678,
|
3830 |
+
"grad_norm": 74.5625,
|
3831 |
+
"learning_rate": 9.991496719593855e-07,
|
3832 |
+
"loss": 133.3608,
|
3833 |
+
"step": 5440
|
3834 |
+
},
|
3835 |
+
{
|
3836 |
+
"epoch": 0.4361684700715651,
|
3837 |
+
"grad_norm": 71.0,
|
3838 |
+
"learning_rate": 9.991481088563697e-07,
|
3839 |
+
"loss": 131.265,
|
3840 |
+
"step": 5450
|
3841 |
+
},
|
3842 |
+
{
|
3843 |
+
"epoch": 0.4369687791909625,
|
3844 |
+
"grad_norm": 73.4375,
|
3845 |
+
"learning_rate": 9.99146545753354e-07,
|
3846 |
+
"loss": 132.0777,
|
3847 |
+
"step": 5460
|
3848 |
+
},
|
3849 |
+
{
|
3850 |
+
"epoch": 0.4377690883103599,
|
3851 |
+
"grad_norm": 75.125,
|
3852 |
+
"learning_rate": 9.99144982650338e-07,
|
3853 |
+
"loss": 132.4698,
|
3854 |
+
"step": 5470
|
3855 |
+
},
|
3856 |
+
{
|
3857 |
+
"epoch": 0.43856939742975726,
|
3858 |
+
"grad_norm": 73.875,
|
3859 |
+
"learning_rate": 9.991434195473221e-07,
|
3860 |
+
"loss": 131.6167,
|
3861 |
+
"step": 5480
|
3862 |
+
},
|
3863 |
+
{
|
3864 |
+
"epoch": 0.43936970654915464,
|
3865 |
+
"grad_norm": 76.9375,
|
3866 |
+
"learning_rate": 9.991418564443064e-07,
|
3867 |
+
"loss": 131.9013,
|
3868 |
+
"step": 5490
|
3869 |
+
},
|
3870 |
+
{
|
3871 |
+
"epoch": 0.44017001566855196,
|
3872 |
+
"grad_norm": 73.4375,
|
3873 |
+
"learning_rate": 9.991402933412906e-07,
|
3874 |
+
"loss": 131.2487,
|
3875 |
+
"step": 5500
|
3876 |
+
},
|
3877 |
+
{
|
3878 |
+
"epoch": 0.44097032478794934,
|
3879 |
+
"grad_norm": 83.4375,
|
3880 |
+
"learning_rate": 9.991387302382746e-07,
|
3881 |
+
"loss": 133.703,
|
3882 |
+
"step": 5510
|
3883 |
+
},
|
3884 |
+
{
|
3885 |
+
"epoch": 0.4417706339073467,
|
3886 |
+
"grad_norm": 78.375,
|
3887 |
+
"learning_rate": 9.991371671352588e-07,
|
3888 |
+
"loss": 133.2442,
|
3889 |
+
"step": 5520
|
3890 |
+
},
|
3891 |
+
{
|
3892 |
+
"epoch": 0.4425709430267441,
|
3893 |
+
"grad_norm": 71.375,
|
3894 |
+
"learning_rate": 9.99135604032243e-07,
|
3895 |
+
"loss": 131.6987,
|
3896 |
+
"step": 5530
|
3897 |
+
},
|
3898 |
+
{
|
3899 |
+
"epoch": 0.4433712521461414,
|
3900 |
+
"grad_norm": 76.1875,
|
3901 |
+
"learning_rate": 9.991340409292272e-07,
|
3902 |
+
"loss": 131.8815,
|
3903 |
+
"step": 5540
|
3904 |
+
},
|
3905 |
+
{
|
3906 |
+
"epoch": 0.4441715612655388,
|
3907 |
+
"grad_norm": 80.8125,
|
3908 |
+
"learning_rate": 9.991324778262112e-07,
|
3909 |
+
"loss": 132.4904,
|
3910 |
+
"step": 5550
|
3911 |
+
},
|
3912 |
+
{
|
3913 |
+
"epoch": 0.4449718703849362,
|
3914 |
+
"grad_norm": 75.1875,
|
3915 |
+
"learning_rate": 9.991309147231955e-07,
|
3916 |
+
"loss": 130.7195,
|
3917 |
+
"step": 5560
|
3918 |
+
},
|
3919 |
+
{
|
3920 |
+
"epoch": 0.44577217950433357,
|
3921 |
+
"grad_norm": 74.5625,
|
3922 |
+
"learning_rate": 9.991293516201797e-07,
|
3923 |
+
"loss": 132.3294,
|
3924 |
+
"step": 5570
|
3925 |
+
},
|
3926 |
+
{
|
3927 |
+
"epoch": 0.4465724886237309,
|
3928 |
+
"grad_norm": 77.5,
|
3929 |
+
"learning_rate": 9.99127788517164e-07,
|
3930 |
+
"loss": 132.4166,
|
3931 |
+
"step": 5580
|
3932 |
+
},
|
3933 |
+
{
|
3934 |
+
"epoch": 0.44737279774312827,
|
3935 |
+
"grad_norm": 74.375,
|
3936 |
+
"learning_rate": 9.991262254141481e-07,
|
3937 |
+
"loss": 132.624,
|
3938 |
+
"step": 5590
|
3939 |
+
},
|
3940 |
+
{
|
3941 |
+
"epoch": 0.44817310686252565,
|
3942 |
+
"grad_norm": 82.3125,
|
3943 |
+
"learning_rate": 9.991246623111321e-07,
|
3944 |
+
"loss": 131.5993,
|
3945 |
+
"step": 5600
|
3946 |
+
},
|
3947 |
+
{
|
3948 |
+
"epoch": 0.44897341598192303,
|
3949 |
+
"grad_norm": 70.5,
|
3950 |
+
"learning_rate": 9.991230992081163e-07,
|
3951 |
+
"loss": 130.5851,
|
3952 |
+
"step": 5610
|
3953 |
+
},
|
3954 |
+
{
|
3955 |
+
"epoch": 0.4497737251013204,
|
3956 |
+
"grad_norm": 78.9375,
|
3957 |
+
"learning_rate": 9.991215361051006e-07,
|
3958 |
+
"loss": 131.7671,
|
3959 |
+
"step": 5620
|
3960 |
+
},
|
3961 |
+
{
|
3962 |
+
"epoch": 0.45057403422071773,
|
3963 |
+
"grad_norm": 78.9375,
|
3964 |
+
"learning_rate": 9.991199730020846e-07,
|
3965 |
+
"loss": 132.6647,
|
3966 |
+
"step": 5630
|
3967 |
+
},
|
3968 |
+
{
|
3969 |
+
"epoch": 0.4513743433401151,
|
3970 |
+
"grad_norm": 77.5,
|
3971 |
+
"learning_rate": 9.991184098990688e-07,
|
3972 |
+
"loss": 131.5212,
|
3973 |
+
"step": 5640
|
3974 |
+
},
|
3975 |
+
{
|
3976 |
+
"epoch": 0.4521746524595125,
|
3977 |
+
"grad_norm": 80.25,
|
3978 |
+
"learning_rate": 9.99116846796053e-07,
|
3979 |
+
"loss": 131.0642,
|
3980 |
+
"step": 5650
|
3981 |
+
},
|
3982 |
+
{
|
3983 |
+
"epoch": 0.4529749615789099,
|
3984 |
+
"grad_norm": 80.875,
|
3985 |
+
"learning_rate": 9.991152836930372e-07,
|
3986 |
+
"loss": 131.6869,
|
3987 |
+
"step": 5660
|
3988 |
+
},
|
3989 |
+
{
|
3990 |
+
"epoch": 0.4537752706983072,
|
3991 |
+
"grad_norm": 73.125,
|
3992 |
+
"learning_rate": 9.991137205900212e-07,
|
3993 |
+
"loss": 132.0529,
|
3994 |
+
"step": 5670
|
3995 |
+
},
|
3996 |
+
{
|
3997 |
+
"epoch": 0.4545755798177046,
|
3998 |
+
"grad_norm": 77.8125,
|
3999 |
+
"learning_rate": 9.991121574870055e-07,
|
4000 |
+
"loss": 133.4353,
|
4001 |
+
"step": 5680
|
4002 |
+
},
|
4003 |
+
{
|
4004 |
+
"epoch": 0.45537588893710196,
|
4005 |
+
"grad_norm": 78.9375,
|
4006 |
+
"learning_rate": 9.991105943839897e-07,
|
4007 |
+
"loss": 131.1083,
|
4008 |
+
"step": 5690
|
4009 |
+
},
|
4010 |
+
{
|
4011 |
+
"epoch": 0.45617619805649934,
|
4012 |
+
"grad_norm": 81.6875,
|
4013 |
+
"learning_rate": 9.991090312809739e-07,
|
4014 |
+
"loss": 132.5374,
|
4015 |
+
"step": 5700
|
4016 |
+
},
|
4017 |
+
{
|
4018 |
+
"epoch": 0.45697650717589666,
|
4019 |
+
"grad_norm": 75.8125,
|
4020 |
+
"learning_rate": 9.99107468177958e-07,
|
4021 |
+
"loss": 131.5541,
|
4022 |
+
"step": 5710
|
4023 |
+
},
|
4024 |
+
{
|
4025 |
+
"epoch": 0.45777681629529404,
|
4026 |
+
"grad_norm": 77.875,
|
4027 |
+
"learning_rate": 9.991059050749421e-07,
|
4028 |
+
"loss": 133.4234,
|
4029 |
+
"step": 5720
|
4030 |
+
},
|
4031 |
+
{
|
4032 |
+
"epoch": 0.4585771254146914,
|
4033 |
+
"grad_norm": 82.6875,
|
4034 |
+
"learning_rate": 9.991043419719263e-07,
|
4035 |
+
"loss": 132.8354,
|
4036 |
+
"step": 5730
|
4037 |
+
},
|
4038 |
+
{
|
4039 |
+
"epoch": 0.4593774345340888,
|
4040 |
+
"grad_norm": 69.4375,
|
4041 |
+
"learning_rate": 9.991027788689105e-07,
|
4042 |
+
"loss": 130.9775,
|
4043 |
+
"step": 5740
|
4044 |
+
},
|
4045 |
+
{
|
4046 |
+
"epoch": 0.4601777436534862,
|
4047 |
+
"grad_norm": 73.875,
|
4048 |
+
"learning_rate": 9.991012157658948e-07,
|
4049 |
+
"loss": 133.6315,
|
4050 |
+
"step": 5750
|
4051 |
+
},
|
4052 |
+
{
|
4053 |
+
"epoch": 0.4609780527728835,
|
4054 |
+
"grad_norm": 77.125,
|
4055 |
+
"learning_rate": 9.990996526628788e-07,
|
4056 |
+
"loss": 133.1302,
|
4057 |
+
"step": 5760
|
4058 |
+
},
|
4059 |
+
{
|
4060 |
+
"epoch": 0.4617783618922809,
|
4061 |
+
"grad_norm": 75.875,
|
4062 |
+
"learning_rate": 9.99098089559863e-07,
|
4063 |
+
"loss": 132.4343,
|
4064 |
+
"step": 5770
|
4065 |
+
},
|
4066 |
+
{
|
4067 |
+
"epoch": 0.46257867101167827,
|
4068 |
+
"grad_norm": 74.8125,
|
4069 |
+
"learning_rate": 9.990965264568472e-07,
|
4070 |
+
"loss": 132.1773,
|
4071 |
+
"step": 5780
|
4072 |
+
},
|
4073 |
+
{
|
4074 |
+
"epoch": 0.46337898013107565,
|
4075 |
+
"grad_norm": 74.5625,
|
4076 |
+
"learning_rate": 9.990949633538312e-07,
|
4077 |
+
"loss": 131.587,
|
4078 |
+
"step": 5790
|
4079 |
+
},
|
4080 |
+
{
|
4081 |
+
"epoch": 0.46417928925047297,
|
4082 |
+
"grad_norm": 73.6875,
|
4083 |
+
"learning_rate": 9.990934002508154e-07,
|
4084 |
+
"loss": 132.3335,
|
4085 |
+
"step": 5800
|
4086 |
+
},
|
4087 |
+
{
|
4088 |
+
"epoch": 0.46497959836987035,
|
4089 |
+
"grad_norm": 78.375,
|
4090 |
+
"learning_rate": 9.990918371477997e-07,
|
4091 |
+
"loss": 131.2027,
|
4092 |
+
"step": 5810
|
4093 |
+
},
|
4094 |
+
{
|
4095 |
+
"epoch": 0.46577990748926773,
|
4096 |
+
"grad_norm": 74.875,
|
4097 |
+
"learning_rate": 9.990902740447837e-07,
|
4098 |
+
"loss": 132.1983,
|
4099 |
+
"step": 5820
|
4100 |
+
},
|
4101 |
+
{
|
4102 |
+
"epoch": 0.4665802166086651,
|
4103 |
+
"grad_norm": 76.8125,
|
4104 |
+
"learning_rate": 9.990887109417679e-07,
|
4105 |
+
"loss": 130.0307,
|
4106 |
+
"step": 5830
|
4107 |
+
},
|
4108 |
+
{
|
4109 |
+
"epoch": 0.4673805257280625,
|
4110 |
+
"grad_norm": 74.375,
|
4111 |
+
"learning_rate": 9.99087147838752e-07,
|
4112 |
+
"loss": 132.1818,
|
4113 |
+
"step": 5840
|
4114 |
+
},
|
4115 |
+
{
|
4116 |
+
"epoch": 0.4681808348474598,
|
4117 |
+
"grad_norm": 80.25,
|
4118 |
+
"learning_rate": 9.990855847357363e-07,
|
4119 |
+
"loss": 131.9706,
|
4120 |
+
"step": 5850
|
4121 |
+
},
|
4122 |
+
{
|
4123 |
+
"epoch": 0.4689811439668572,
|
4124 |
+
"grad_norm": 78.3125,
|
4125 |
+
"learning_rate": 9.990840216327205e-07,
|
4126 |
+
"loss": 132.3331,
|
4127 |
+
"step": 5860
|
4128 |
+
},
|
4129 |
+
{
|
4130 |
+
"epoch": 0.4697814530862546,
|
4131 |
+
"grad_norm": 75.25,
|
4132 |
+
"learning_rate": 9.990824585297045e-07,
|
4133 |
+
"loss": 132.2483,
|
4134 |
+
"step": 5870
|
4135 |
+
},
|
4136 |
+
{
|
4137 |
+
"epoch": 0.47058176220565195,
|
4138 |
+
"grad_norm": 73.125,
|
4139 |
+
"learning_rate": 9.990808954266888e-07,
|
4140 |
+
"loss": 132.6989,
|
4141 |
+
"step": 5880
|
4142 |
+
},
|
4143 |
+
{
|
4144 |
+
"epoch": 0.4713820713250493,
|
4145 |
+
"grad_norm": 78.625,
|
4146 |
+
"learning_rate": 9.99079332323673e-07,
|
4147 |
+
"loss": 133.315,
|
4148 |
+
"step": 5890
|
4149 |
+
},
|
4150 |
+
{
|
4151 |
+
"epoch": 0.47218238044444666,
|
4152 |
+
"grad_norm": 73.8125,
|
4153 |
+
"learning_rate": 9.990777692206572e-07,
|
4154 |
+
"loss": 131.2918,
|
4155 |
+
"step": 5900
|
4156 |
+
},
|
4157 |
+
{
|
4158 |
+
"epoch": 0.47298268956384404,
|
4159 |
+
"grad_norm": 82.0625,
|
4160 |
+
"learning_rate": 9.990762061176414e-07,
|
4161 |
+
"loss": 132.8047,
|
4162 |
+
"step": 5910
|
4163 |
+
},
|
4164 |
+
{
|
4165 |
+
"epoch": 0.4737829986832414,
|
4166 |
+
"grad_norm": 83.4375,
|
4167 |
+
"learning_rate": 9.990746430146254e-07,
|
4168 |
+
"loss": 133.7549,
|
4169 |
+
"step": 5920
|
4170 |
+
},
|
4171 |
+
{
|
4172 |
+
"epoch": 0.47458330780263874,
|
4173 |
+
"grad_norm": 71.9375,
|
4174 |
+
"learning_rate": 9.990730799116096e-07,
|
4175 |
+
"loss": 132.3743,
|
4176 |
+
"step": 5930
|
4177 |
+
},
|
4178 |
+
{
|
4179 |
+
"epoch": 0.4753836169220361,
|
4180 |
+
"grad_norm": 82.25,
|
4181 |
+
"learning_rate": 9.990715168085939e-07,
|
4182 |
+
"loss": 130.8257,
|
4183 |
+
"step": 5940
|
4184 |
+
},
|
4185 |
+
{
|
4186 |
+
"epoch": 0.4761839260414335,
|
4187 |
+
"grad_norm": 75.8125,
|
4188 |
+
"learning_rate": 9.990699537055779e-07,
|
4189 |
+
"loss": 130.2371,
|
4190 |
+
"step": 5950
|
4191 |
+
},
|
4192 |
+
{
|
4193 |
+
"epoch": 0.4769842351608309,
|
4194 |
+
"grad_norm": 78.1875,
|
4195 |
+
"learning_rate": 9.99068390602562e-07,
|
4196 |
+
"loss": 131.3828,
|
4197 |
+
"step": 5960
|
4198 |
+
},
|
4199 |
+
{
|
4200 |
+
"epoch": 0.47778454428022826,
|
4201 |
+
"grad_norm": 75.9375,
|
4202 |
+
"learning_rate": 9.990668274995463e-07,
|
4203 |
+
"loss": 132.8423,
|
4204 |
+
"step": 5970
|
4205 |
+
},
|
4206 |
+
{
|
4207 |
+
"epoch": 0.4785848533996256,
|
4208 |
+
"grad_norm": 77.3125,
|
4209 |
+
"learning_rate": 9.990652643965303e-07,
|
4210 |
+
"loss": 133.5517,
|
4211 |
+
"step": 5980
|
4212 |
+
},
|
4213 |
+
{
|
4214 |
+
"epoch": 0.47938516251902297,
|
4215 |
+
"grad_norm": 72.5,
|
4216 |
+
"learning_rate": 9.990637012935145e-07,
|
4217 |
+
"loss": 132.8932,
|
4218 |
+
"step": 5990
|
4219 |
+
},
|
4220 |
+
{
|
4221 |
+
"epoch": 0.48018547163842035,
|
4222 |
+
"grad_norm": 74.25,
|
4223 |
+
"learning_rate": 9.990621381904987e-07,
|
4224 |
+
"loss": 133.124,
|
4225 |
+
"step": 6000
|
4226 |
+
},
|
4227 |
+
{
|
4228 |
+
"epoch": 0.4809857807578177,
|
4229 |
+
"grad_norm": 77.0625,
|
4230 |
+
"learning_rate": 9.99060575087483e-07,
|
4231 |
+
"loss": 132.1078,
|
4232 |
+
"step": 6010
|
4233 |
+
},
|
4234 |
+
{
|
4235 |
+
"epoch": 0.48178608987721505,
|
4236 |
+
"grad_norm": 76.0625,
|
4237 |
+
"learning_rate": 9.99059011984467e-07,
|
4238 |
+
"loss": 132.8352,
|
4239 |
+
"step": 6020
|
4240 |
+
},
|
4241 |
+
{
|
4242 |
+
"epoch": 0.48258639899661243,
|
4243 |
+
"grad_norm": 79.4375,
|
4244 |
+
"learning_rate": 9.990574488814512e-07,
|
4245 |
+
"loss": 132.3257,
|
4246 |
+
"step": 6030
|
4247 |
+
},
|
4248 |
+
{
|
4249 |
+
"epoch": 0.4833867081160098,
|
4250 |
+
"grad_norm": 74.375,
|
4251 |
+
"learning_rate": 9.990558857784354e-07,
|
4252 |
+
"loss": 132.8362,
|
4253 |
+
"step": 6040
|
4254 |
+
},
|
4255 |
+
{
|
4256 |
+
"epoch": 0.4841870172354072,
|
4257 |
+
"grad_norm": 74.375,
|
4258 |
+
"learning_rate": 9.990543226754196e-07,
|
4259 |
+
"loss": 131.3189,
|
4260 |
+
"step": 6050
|
4261 |
+
},
|
4262 |
+
{
|
4263 |
+
"epoch": 0.48498732635480457,
|
4264 |
+
"grad_norm": 72.0625,
|
4265 |
+
"learning_rate": 9.990527595724038e-07,
|
4266 |
+
"loss": 131.8034,
|
4267 |
+
"step": 6060
|
4268 |
+
},
|
4269 |
+
{
|
4270 |
+
"epoch": 0.4857876354742019,
|
4271 |
+
"grad_norm": 80.5625,
|
4272 |
+
"learning_rate": 9.99051196469388e-07,
|
4273 |
+
"loss": 131.7185,
|
4274 |
+
"step": 6070
|
4275 |
+
},
|
4276 |
+
{
|
4277 |
+
"epoch": 0.4865879445935993,
|
4278 |
+
"grad_norm": 77.5625,
|
4279 |
+
"learning_rate": 9.99049633366372e-07,
|
4280 |
+
"loss": 133.0989,
|
4281 |
+
"step": 6080
|
4282 |
+
},
|
4283 |
+
{
|
4284 |
+
"epoch": 0.48738825371299666,
|
4285 |
+
"grad_norm": 74.4375,
|
4286 |
+
"learning_rate": 9.990480702633563e-07,
|
4287 |
+
"loss": 131.1641,
|
4288 |
+
"step": 6090
|
4289 |
+
},
|
4290 |
+
{
|
4291 |
+
"epoch": 0.48818856283239404,
|
4292 |
+
"grad_norm": 77.625,
|
4293 |
+
"learning_rate": 9.990465071603405e-07,
|
4294 |
+
"loss": 131.6567,
|
4295 |
+
"step": 6100
|
4296 |
+
},
|
4297 |
+
{
|
4298 |
+
"epoch": 0.48898887195179136,
|
4299 |
+
"grad_norm": 80.25,
|
4300 |
+
"learning_rate": 9.990449440573245e-07,
|
4301 |
+
"loss": 132.3284,
|
4302 |
+
"step": 6110
|
4303 |
+
},
|
4304 |
+
{
|
4305 |
+
"epoch": 0.48978918107118874,
|
4306 |
+
"grad_norm": 72.9375,
|
4307 |
+
"learning_rate": 9.990433809543087e-07,
|
4308 |
+
"loss": 131.856,
|
4309 |
+
"step": 6120
|
4310 |
+
},
|
4311 |
+
{
|
4312 |
+
"epoch": 0.4905894901905861,
|
4313 |
+
"grad_norm": 72.6875,
|
4314 |
+
"learning_rate": 9.99041817851293e-07,
|
4315 |
+
"loss": 132.8506,
|
4316 |
+
"step": 6130
|
4317 |
+
},
|
4318 |
+
{
|
4319 |
+
"epoch": 0.4913897993099835,
|
4320 |
+
"grad_norm": 76.0625,
|
4321 |
+
"learning_rate": 9.99040254748277e-07,
|
4322 |
+
"loss": 132.9671,
|
4323 |
+
"step": 6140
|
4324 |
+
},
|
4325 |
+
{
|
4326 |
+
"epoch": 0.4921901084293808,
|
4327 |
+
"grad_norm": 81.875,
|
4328 |
+
"learning_rate": 9.990386916452612e-07,
|
4329 |
+
"loss": 133.4054,
|
4330 |
+
"step": 6150
|
4331 |
+
},
|
4332 |
+
{
|
4333 |
+
"epoch": 0.4929904175487782,
|
4334 |
+
"grad_norm": 73.1875,
|
4335 |
+
"learning_rate": 9.990371285422454e-07,
|
4336 |
+
"loss": 132.2609,
|
4337 |
+
"step": 6160
|
4338 |
+
},
|
4339 |
+
{
|
4340 |
+
"epoch": 0.4937907266681756,
|
4341 |
+
"grad_norm": 73.9375,
|
4342 |
+
"learning_rate": 9.990355654392296e-07,
|
4343 |
+
"loss": 133.1529,
|
4344 |
+
"step": 6170
|
4345 |
+
},
|
4346 |
+
{
|
4347 |
+
"epoch": 0.49459103578757296,
|
4348 |
+
"grad_norm": 74.125,
|
4349 |
+
"learning_rate": 9.990340023362136e-07,
|
4350 |
+
"loss": 133.8595,
|
4351 |
+
"step": 6180
|
4352 |
+
},
|
4353 |
+
{
|
4354 |
+
"epoch": 0.49539134490697034,
|
4355 |
+
"grad_norm": 72.0625,
|
4356 |
+
"learning_rate": 9.990324392331978e-07,
|
4357 |
+
"loss": 132.0925,
|
4358 |
+
"step": 6190
|
4359 |
+
},
|
4360 |
+
{
|
4361 |
+
"epoch": 0.49619165402636767,
|
4362 |
+
"grad_norm": 72.5,
|
4363 |
+
"learning_rate": 9.99030876130182e-07,
|
4364 |
+
"loss": 133.2302,
|
4365 |
+
"step": 6200
|
4366 |
+
},
|
4367 |
+
{
|
4368 |
+
"epoch": 0.49699196314576505,
|
4369 |
+
"grad_norm": 71.9375,
|
4370 |
+
"learning_rate": 9.990293130271663e-07,
|
4371 |
+
"loss": 132.7785,
|
4372 |
+
"step": 6210
|
4373 |
+
},
|
4374 |
+
{
|
4375 |
+
"epoch": 0.4977922722651624,
|
4376 |
+
"grad_norm": 78.0625,
|
4377 |
+
"learning_rate": 9.990277499241505e-07,
|
4378 |
+
"loss": 133.0601,
|
4379 |
+
"step": 6220
|
4380 |
+
},
|
4381 |
+
{
|
4382 |
+
"epoch": 0.4985925813845598,
|
4383 |
+
"grad_norm": 74.6875,
|
4384 |
+
"learning_rate": 9.990261868211345e-07,
|
4385 |
+
"loss": 132.1573,
|
4386 |
+
"step": 6230
|
4387 |
+
},
|
4388 |
+
{
|
4389 |
+
"epoch": 0.49939289050395713,
|
4390 |
+
"grad_norm": 80.3125,
|
4391 |
+
"learning_rate": 9.990246237181187e-07,
|
4392 |
+
"loss": 132.7871,
|
4393 |
+
"step": 6240
|
4394 |
+
},
|
4395 |
+
{
|
4396 |
+
"epoch": 0.5001931996233545,
|
4397 |
+
"grad_norm": 74.875,
|
4398 |
+
"learning_rate": 9.99023060615103e-07,
|
4399 |
+
"loss": 132.2812,
|
4400 |
+
"step": 6250
|
4401 |
+
},
|
4402 |
+
{
|
4403 |
+
"epoch": 0.5009935087427518,
|
4404 |
+
"grad_norm": 80.875,
|
4405 |
+
"learning_rate": 9.990214975120871e-07,
|
4406 |
+
"loss": 130.9305,
|
4407 |
+
"step": 6260
|
4408 |
+
},
|
4409 |
+
{
|
4410 |
+
"epoch": 0.5017938178621493,
|
4411 |
+
"grad_norm": 72.4375,
|
4412 |
+
"learning_rate": 9.990199344090712e-07,
|
4413 |
+
"loss": 133.3334,
|
4414 |
+
"step": 6270
|
4415 |
+
},
|
4416 |
+
{
|
4417 |
+
"epoch": 0.5025941269815466,
|
4418 |
+
"grad_norm": 71.875,
|
4419 |
+
"learning_rate": 9.990183713060554e-07,
|
4420 |
+
"loss": 132.1767,
|
4421 |
+
"step": 6280
|
4422 |
+
},
|
4423 |
+
{
|
4424 |
+
"epoch": 0.503394436100944,
|
4425 |
+
"grad_norm": 79.375,
|
4426 |
+
"learning_rate": 9.990168082030396e-07,
|
4427 |
+
"loss": 133.0485,
|
4428 |
+
"step": 6290
|
4429 |
+
},
|
4430 |
+
{
|
4431 |
+
"epoch": 0.5041947452203414,
|
4432 |
+
"grad_norm": 72.8125,
|
4433 |
+
"learning_rate": 9.990152451000236e-07,
|
4434 |
+
"loss": 132.0657,
|
4435 |
+
"step": 6300
|
4436 |
+
},
|
4437 |
+
{
|
4438 |
+
"epoch": 0.5049950543397387,
|
4439 |
+
"grad_norm": 82.5,
|
4440 |
+
"learning_rate": 9.990136819970078e-07,
|
4441 |
+
"loss": 130.4747,
|
4442 |
+
"step": 6310
|
4443 |
+
},
|
4444 |
+
{
|
4445 |
+
"epoch": 0.5057953634591361,
|
4446 |
+
"grad_norm": 77.5,
|
4447 |
+
"learning_rate": 9.99012118893992e-07,
|
4448 |
+
"loss": 133.164,
|
4449 |
+
"step": 6320
|
4450 |
+
},
|
4451 |
+
{
|
4452 |
+
"epoch": 0.5065956725785334,
|
4453 |
+
"grad_norm": 72.5625,
|
4454 |
+
"learning_rate": 9.990105557909763e-07,
|
4455 |
+
"loss": 131.2053,
|
4456 |
+
"step": 6330
|
4457 |
+
},
|
4458 |
+
{
|
4459 |
+
"epoch": 0.5073959816979309,
|
4460 |
+
"grad_norm": 74.5,
|
4461 |
+
"learning_rate": 9.990089926879603e-07,
|
4462 |
+
"loss": 132.6195,
|
4463 |
+
"step": 6340
|
4464 |
+
},
|
4465 |
+
{
|
4466 |
+
"epoch": 0.5081962908173282,
|
4467 |
+
"grad_norm": 78.5625,
|
4468 |
+
"learning_rate": 9.990074295849445e-07,
|
4469 |
+
"loss": 131.7617,
|
4470 |
+
"step": 6350
|
4471 |
+
},
|
4472 |
+
{
|
4473 |
+
"epoch": 0.5089965999367255,
|
4474 |
+
"grad_norm": 73.5,
|
4475 |
+
"learning_rate": 9.990058664819287e-07,
|
4476 |
+
"loss": 131.7562,
|
4477 |
+
"step": 6360
|
4478 |
+
},
|
4479 |
+
{
|
4480 |
+
"epoch": 0.509796909056123,
|
4481 |
+
"grad_norm": 79.4375,
|
4482 |
+
"learning_rate": 9.99004303378913e-07,
|
4483 |
+
"loss": 133.1088,
|
4484 |
+
"step": 6370
|
4485 |
+
},
|
4486 |
+
{
|
4487 |
+
"epoch": 0.5105972181755203,
|
4488 |
+
"grad_norm": 79.3125,
|
4489 |
+
"learning_rate": 9.990027402758971e-07,
|
4490 |
+
"loss": 133.757,
|
4491 |
+
"step": 6380
|
4492 |
+
},
|
4493 |
+
{
|
4494 |
+
"epoch": 0.5113975272949176,
|
4495 |
+
"grad_norm": 74.0625,
|
4496 |
+
"learning_rate": 9.990011771728811e-07,
|
4497 |
+
"loss": 133.9952,
|
4498 |
+
"step": 6390
|
4499 |
+
},
|
4500 |
+
{
|
4501 |
+
"epoch": 0.512197836414315,
|
4502 |
+
"grad_norm": 72.5625,
|
4503 |
+
"learning_rate": 9.989996140698654e-07,
|
4504 |
+
"loss": 131.275,
|
4505 |
+
"step": 6400
|
4506 |
+
},
|
4507 |
+
{
|
4508 |
+
"epoch": 0.5129981455337124,
|
4509 |
+
"grad_norm": 78.4375,
|
4510 |
+
"learning_rate": 9.989980509668496e-07,
|
4511 |
+
"loss": 131.3286,
|
4512 |
+
"step": 6410
|
4513 |
+
},
|
4514 |
+
{
|
4515 |
+
"epoch": 0.5137984546531098,
|
4516 |
+
"grad_norm": 73.6875,
|
4517 |
+
"learning_rate": 9.989964878638338e-07,
|
4518 |
+
"loss": 131.1604,
|
4519 |
+
"step": 6420
|
4520 |
+
},
|
4521 |
+
{
|
4522 |
+
"epoch": 0.5145987637725071,
|
4523 |
+
"grad_norm": 77.0,
|
4524 |
+
"learning_rate": 9.989949247608178e-07,
|
4525 |
+
"loss": 131.0352,
|
4526 |
+
"step": 6430
|
4527 |
+
},
|
4528 |
+
{
|
4529 |
+
"epoch": 0.5153990728919045,
|
4530 |
+
"grad_norm": 77.8125,
|
4531 |
+
"learning_rate": 9.98993361657802e-07,
|
4532 |
+
"loss": 132.0469,
|
4533 |
+
"step": 6440
|
4534 |
+
},
|
4535 |
+
{
|
4536 |
+
"epoch": 0.5161993820113019,
|
4537 |
+
"grad_norm": 76.1875,
|
4538 |
+
"learning_rate": 9.989917985547862e-07,
|
4539 |
+
"loss": 133.7251,
|
4540 |
+
"step": 6450
|
4541 |
+
},
|
4542 |
+
{
|
4543 |
+
"epoch": 0.5169996911306992,
|
4544 |
+
"grad_norm": 70.6875,
|
4545 |
+
"learning_rate": 9.989902354517702e-07,
|
4546 |
+
"loss": 131.82,
|
4547 |
+
"step": 6460
|
4548 |
+
},
|
4549 |
+
{
|
4550 |
+
"epoch": 0.5178000002500966,
|
4551 |
+
"grad_norm": 76.8125,
|
4552 |
+
"learning_rate": 9.989886723487545e-07,
|
4553 |
+
"loss": 131.712,
|
4554 |
+
"step": 6470
|
4555 |
+
},
|
4556 |
+
{
|
4557 |
+
"epoch": 0.518600309369494,
|
4558 |
+
"grad_norm": 77.9375,
|
4559 |
+
"learning_rate": 9.989871092457387e-07,
|
4560 |
+
"loss": 132.1365,
|
4561 |
+
"step": 6480
|
4562 |
+
},
|
4563 |
+
{
|
4564 |
+
"epoch": 0.5194006184888913,
|
4565 |
+
"grad_norm": 74.0,
|
4566 |
+
"learning_rate": 9.989855461427227e-07,
|
4567 |
+
"loss": 131.9362,
|
4568 |
+
"step": 6490
|
4569 |
+
},
|
4570 |
+
{
|
4571 |
+
"epoch": 0.5202009276082887,
|
4572 |
+
"grad_norm": 77.3125,
|
4573 |
+
"learning_rate": 9.98983983039707e-07,
|
4574 |
+
"loss": 133.708,
|
4575 |
+
"step": 6500
|
4576 |
+
},
|
4577 |
+
{
|
4578 |
+
"epoch": 0.5210012367276861,
|
4579 |
+
"grad_norm": 77.75,
|
4580 |
+
"learning_rate": 9.989824199366911e-07,
|
4581 |
+
"loss": 132.9561,
|
4582 |
+
"step": 6510
|
4583 |
+
},
|
4584 |
+
{
|
4585 |
+
"epoch": 0.5218015458470834,
|
4586 |
+
"grad_norm": 76.8125,
|
4587 |
+
"learning_rate": 9.989808568336753e-07,
|
4588 |
+
"loss": 133.186,
|
4589 |
+
"step": 6520
|
4590 |
+
},
|
4591 |
+
{
|
4592 |
+
"epoch": 0.5226018549664808,
|
4593 |
+
"grad_norm": 79.125,
|
4594 |
+
"learning_rate": 9.989792937306596e-07,
|
4595 |
+
"loss": 131.5064,
|
4596 |
+
"step": 6530
|
4597 |
+
},
|
4598 |
+
{
|
4599 |
+
"epoch": 0.5234021640858781,
|
4600 |
+
"grad_norm": 74.5,
|
4601 |
+
"learning_rate": 9.989777306276438e-07,
|
4602 |
+
"loss": 131.7443,
|
4603 |
+
"step": 6540
|
4604 |
+
},
|
4605 |
+
{
|
4606 |
+
"epoch": 0.5242024732052756,
|
4607 |
+
"grad_norm": 75.9375,
|
4608 |
+
"learning_rate": 9.989761675246278e-07,
|
4609 |
+
"loss": 132.0182,
|
4610 |
+
"step": 6550
|
4611 |
+
},
|
4612 |
+
{
|
4613 |
+
"epoch": 0.5250027823246729,
|
4614 |
+
"grad_norm": 74.25,
|
4615 |
+
"learning_rate": 9.98974604421612e-07,
|
4616 |
+
"loss": 132.1322,
|
4617 |
+
"step": 6560
|
4618 |
+
},
|
4619 |
+
{
|
4620 |
+
"epoch": 0.5258030914440702,
|
4621 |
+
"grad_norm": 77.25,
|
4622 |
+
"learning_rate": 9.989730413185962e-07,
|
4623 |
+
"loss": 131.4318,
|
4624 |
+
"step": 6570
|
4625 |
+
},
|
4626 |
+
{
|
4627 |
+
"epoch": 0.5266034005634677,
|
4628 |
+
"grad_norm": 79.375,
|
4629 |
+
"learning_rate": 9.989714782155802e-07,
|
4630 |
+
"loss": 131.6486,
|
4631 |
+
"step": 6580
|
4632 |
+
},
|
4633 |
+
{
|
4634 |
+
"epoch": 0.527403709682865,
|
4635 |
+
"grad_norm": 72.875,
|
4636 |
+
"learning_rate": 9.989699151125644e-07,
|
4637 |
+
"loss": 132.6251,
|
4638 |
+
"step": 6590
|
4639 |
+
},
|
4640 |
+
{
|
4641 |
+
"epoch": 0.5282040188022624,
|
4642 |
+
"grad_norm": 77.75,
|
4643 |
+
"learning_rate": 9.989683520095487e-07,
|
4644 |
+
"loss": 130.9398,
|
4645 |
+
"step": 6600
|
4646 |
+
},
|
4647 |
+
{
|
4648 |
+
"epoch": 0.5290043279216597,
|
4649 |
+
"grad_norm": 78.0625,
|
4650 |
+
"learning_rate": 9.989667889065329e-07,
|
4651 |
+
"loss": 132.2289,
|
4652 |
+
"step": 6610
|
4653 |
+
},
|
4654 |
+
{
|
4655 |
+
"epoch": 0.5298046370410571,
|
4656 |
+
"grad_norm": 80.75,
|
4657 |
+
"learning_rate": 9.989652258035169e-07,
|
4658 |
+
"loss": 133.6902,
|
4659 |
+
"step": 6620
|
4660 |
+
},
|
4661 |
+
{
|
4662 |
+
"epoch": 0.5306049461604545,
|
4663 |
+
"grad_norm": 77.0,
|
4664 |
+
"learning_rate": 9.989636627005011e-07,
|
4665 |
+
"loss": 131.2565,
|
4666 |
+
"step": 6630
|
4667 |
+
},
|
4668 |
+
{
|
4669 |
+
"epoch": 0.5314052552798518,
|
4670 |
+
"grad_norm": 72.3125,
|
4671 |
+
"learning_rate": 9.989620995974853e-07,
|
4672 |
+
"loss": 133.0545,
|
4673 |
+
"step": 6640
|
4674 |
+
},
|
4675 |
+
{
|
4676 |
+
"epoch": 0.5322055643992493,
|
4677 |
+
"grad_norm": 76.125,
|
4678 |
+
"learning_rate": 9.989605364944693e-07,
|
4679 |
+
"loss": 131.2922,
|
4680 |
+
"step": 6650
|
4681 |
+
},
|
4682 |
+
{
|
4683 |
+
"epoch": 0.5330058735186466,
|
4684 |
+
"grad_norm": 70.8125,
|
4685 |
+
"learning_rate": 9.989589733914536e-07,
|
4686 |
+
"loss": 133.2325,
|
4687 |
+
"step": 6660
|
4688 |
+
},
|
4689 |
+
{
|
4690 |
+
"epoch": 0.5338061826380439,
|
4691 |
+
"grad_norm": 83.5625,
|
4692 |
+
"learning_rate": 9.989574102884378e-07,
|
4693 |
+
"loss": 133.1627,
|
4694 |
+
"step": 6670
|
4695 |
+
},
|
4696 |
+
{
|
4697 |
+
"epoch": 0.5346064917574413,
|
4698 |
+
"grad_norm": 73.4375,
|
4699 |
+
"learning_rate": 9.98955847185422e-07,
|
4700 |
+
"loss": 132.0812,
|
4701 |
+
"step": 6680
|
4702 |
+
},
|
4703 |
+
{
|
4704 |
+
"epoch": 0.5354068008768387,
|
4705 |
+
"grad_norm": 84.8125,
|
4706 |
+
"learning_rate": 9.989542840824062e-07,
|
4707 |
+
"loss": 131.5462,
|
4708 |
+
"step": 6690
|
4709 |
+
},
|
4710 |
+
{
|
4711 |
+
"epoch": 0.536207109996236,
|
4712 |
+
"grad_norm": 75.0,
|
4713 |
+
"learning_rate": 9.989527209793904e-07,
|
4714 |
+
"loss": 132.3298,
|
4715 |
+
"step": 6700
|
4716 |
+
},
|
4717 |
+
{
|
4718 |
+
"epoch": 0.5370074191156334,
|
4719 |
+
"grad_norm": 82.0625,
|
4720 |
+
"learning_rate": 9.989511578763744e-07,
|
4721 |
+
"loss": 133.5979,
|
4722 |
+
"step": 6710
|
4723 |
+
},
|
4724 |
+
{
|
4725 |
+
"epoch": 0.5378077282350308,
|
4726 |
+
"grad_norm": 73.8125,
|
4727 |
+
"learning_rate": 9.989495947733586e-07,
|
4728 |
+
"loss": 130.3924,
|
4729 |
+
"step": 6720
|
4730 |
+
},
|
4731 |
+
{
|
4732 |
+
"epoch": 0.5386080373544282,
|
4733 |
+
"grad_norm": 85.1875,
|
4734 |
+
"learning_rate": 9.989480316703429e-07,
|
4735 |
+
"loss": 130.9538,
|
4736 |
+
"step": 6730
|
4737 |
+
},
|
4738 |
+
{
|
4739 |
+
"epoch": 0.5394083464738255,
|
4740 |
+
"grad_norm": 77.4375,
|
4741 |
+
"learning_rate": 9.989464685673269e-07,
|
4742 |
+
"loss": 132.9888,
|
4743 |
+
"step": 6740
|
4744 |
+
},
|
4745 |
+
{
|
4746 |
+
"epoch": 0.5402086555932228,
|
4747 |
+
"grad_norm": 78.75,
|
4748 |
+
"learning_rate": 9.98944905464311e-07,
|
4749 |
+
"loss": 132.472,
|
4750 |
+
"step": 6750
|
4751 |
+
},
|
4752 |
+
{
|
4753 |
+
"epoch": 0.5410089647126203,
|
4754 |
+
"grad_norm": 73.6875,
|
4755 |
+
"learning_rate": 9.989433423612953e-07,
|
4756 |
+
"loss": 132.7828,
|
4757 |
+
"step": 6760
|
4758 |
+
},
|
4759 |
+
{
|
4760 |
+
"epoch": 0.5418092738320176,
|
4761 |
+
"grad_norm": 74.9375,
|
4762 |
+
"learning_rate": 9.989417792582795e-07,
|
4763 |
+
"loss": 131.1772,
|
4764 |
+
"step": 6770
|
4765 |
+
},
|
4766 |
+
{
|
4767 |
+
"epoch": 0.542609582951415,
|
4768 |
+
"grad_norm": 73.5625,
|
4769 |
+
"learning_rate": 9.989402161552635e-07,
|
4770 |
+
"loss": 132.5478,
|
4771 |
+
"step": 6780
|
4772 |
+
},
|
4773 |
+
{
|
4774 |
+
"epoch": 0.5434098920708124,
|
4775 |
+
"grad_norm": 79.5,
|
4776 |
+
"learning_rate": 9.989386530522478e-07,
|
4777 |
+
"loss": 130.6843,
|
4778 |
+
"step": 6790
|
4779 |
+
},
|
4780 |
+
{
|
4781 |
+
"epoch": 0.5442102011902097,
|
4782 |
+
"grad_norm": 73.5625,
|
4783 |
+
"learning_rate": 9.98937089949232e-07,
|
4784 |
+
"loss": 131.9659,
|
4785 |
+
"step": 6800
|
4786 |
+
},
|
4787 |
+
{
|
4788 |
+
"epoch": 0.5450105103096071,
|
4789 |
+
"grad_norm": 73.3125,
|
4790 |
+
"learning_rate": 9.98935526846216e-07,
|
4791 |
+
"loss": 131.0417,
|
4792 |
+
"step": 6810
|
4793 |
+
},
|
4794 |
+
{
|
4795 |
+
"epoch": 0.5458108194290044,
|
4796 |
+
"grad_norm": 74.3125,
|
4797 |
+
"learning_rate": 9.989339637432002e-07,
|
4798 |
+
"loss": 133.0612,
|
4799 |
+
"step": 6820
|
4800 |
+
},
|
4801 |
+
{
|
4802 |
+
"epoch": 0.5466111285484018,
|
4803 |
+
"grad_norm": 75.25,
|
4804 |
+
"learning_rate": 9.989324006401844e-07,
|
4805 |
+
"loss": 132.5644,
|
4806 |
+
"step": 6830
|
4807 |
+
},
|
4808 |
+
{
|
4809 |
+
"epoch": 0.5474114376677992,
|
4810 |
+
"grad_norm": 77.75,
|
4811 |
+
"learning_rate": 9.989308375371686e-07,
|
4812 |
+
"loss": 132.1365,
|
4813 |
+
"step": 6840
|
4814 |
+
},
|
4815 |
+
{
|
4816 |
+
"epoch": 0.5482117467871965,
|
4817 |
+
"grad_norm": 77.1875,
|
4818 |
+
"learning_rate": 9.989292744341529e-07,
|
4819 |
+
"loss": 132.7418,
|
4820 |
+
"step": 6850
|
4821 |
+
},
|
4822 |
+
{
|
4823 |
+
"epoch": 0.549012055906594,
|
4824 |
+
"grad_norm": 79.125,
|
4825 |
+
"learning_rate": 9.989277113311369e-07,
|
4826 |
+
"loss": 131.9871,
|
4827 |
+
"step": 6860
|
4828 |
+
},
|
4829 |
+
{
|
4830 |
+
"epoch": 0.5498123650259913,
|
4831 |
+
"grad_norm": 75.0,
|
4832 |
+
"learning_rate": 9.98926148228121e-07,
|
4833 |
+
"loss": 131.313,
|
4834 |
+
"step": 6870
|
4835 |
+
},
|
4836 |
+
{
|
4837 |
+
"epoch": 0.5506126741453886,
|
4838 |
+
"grad_norm": 74.6875,
|
4839 |
+
"learning_rate": 9.989245851251053e-07,
|
4840 |
+
"loss": 132.0703,
|
4841 |
+
"step": 6880
|
4842 |
+
},
|
4843 |
+
{
|
4844 |
+
"epoch": 0.551412983264786,
|
4845 |
+
"grad_norm": 73.5625,
|
4846 |
+
"learning_rate": 9.989230220220895e-07,
|
4847 |
+
"loss": 130.9999,
|
4848 |
+
"step": 6890
|
4849 |
+
},
|
4850 |
+
{
|
4851 |
+
"epoch": 0.5522132923841834,
|
4852 |
+
"grad_norm": 75.0625,
|
4853 |
+
"learning_rate": 9.989214589190735e-07,
|
4854 |
+
"loss": 132.2652,
|
4855 |
+
"step": 6900
|
4856 |
+
},
|
4857 |
+
{
|
4858 |
+
"epoch": 0.5530136015035808,
|
4859 |
+
"grad_norm": 74.125,
|
4860 |
+
"learning_rate": 9.989198958160577e-07,
|
4861 |
+
"loss": 131.5572,
|
4862 |
+
"step": 6910
|
4863 |
+
},
|
4864 |
+
{
|
4865 |
+
"epoch": 0.5538139106229781,
|
4866 |
+
"grad_norm": 75.8125,
|
4867 |
+
"learning_rate": 9.98918332713042e-07,
|
4868 |
+
"loss": 130.6713,
|
4869 |
+
"step": 6920
|
4870 |
+
},
|
4871 |
+
{
|
4872 |
+
"epoch": 0.5546142197423755,
|
4873 |
+
"grad_norm": 75.0625,
|
4874 |
+
"learning_rate": 9.98916769610026e-07,
|
4875 |
+
"loss": 132.5074,
|
4876 |
+
"step": 6930
|
4877 |
+
},
|
4878 |
+
{
|
4879 |
+
"epoch": 0.5554145288617729,
|
4880 |
+
"grad_norm": 83.5,
|
4881 |
+
"learning_rate": 9.989152065070102e-07,
|
4882 |
+
"loss": 132.7792,
|
4883 |
+
"step": 6940
|
4884 |
+
},
|
4885 |
+
{
|
4886 |
+
"epoch": 0.5562148379811702,
|
4887 |
+
"grad_norm": 78.6875,
|
4888 |
+
"learning_rate": 9.989136434039944e-07,
|
4889 |
+
"loss": 132.4211,
|
4890 |
+
"step": 6950
|
4891 |
+
},
|
4892 |
+
{
|
4893 |
+
"epoch": 0.5570151471005675,
|
4894 |
+
"grad_norm": 84.125,
|
4895 |
+
"learning_rate": 9.989120803009786e-07,
|
4896 |
+
"loss": 133.1216,
|
4897 |
+
"step": 6960
|
4898 |
+
},
|
4899 |
+
{
|
4900 |
+
"epoch": 0.557815456219965,
|
4901 |
+
"grad_norm": 80.8125,
|
4902 |
+
"learning_rate": 9.989105171979626e-07,
|
4903 |
+
"loss": 131.8983,
|
4904 |
+
"step": 6970
|
4905 |
+
},
|
4906 |
+
{
|
4907 |
+
"epoch": 0.5586157653393623,
|
4908 |
+
"grad_norm": 77.875,
|
4909 |
+
"learning_rate": 9.989089540949468e-07,
|
4910 |
+
"loss": 131.918,
|
4911 |
+
"step": 6980
|
4912 |
+
},
|
4913 |
+
{
|
4914 |
+
"epoch": 0.5594160744587597,
|
4915 |
+
"grad_norm": 73.1875,
|
4916 |
+
"learning_rate": 9.98907390991931e-07,
|
4917 |
+
"loss": 131.1051,
|
4918 |
+
"step": 6990
|
4919 |
+
},
|
4920 |
+
{
|
4921 |
+
"epoch": 0.5602163835781571,
|
4922 |
+
"grad_norm": 78.0625,
|
4923 |
+
"learning_rate": 9.989058278889153e-07,
|
4924 |
+
"loss": 132.776,
|
4925 |
+
"step": 7000
|
4926 |
+
},
|
4927 |
+
{
|
4928 |
+
"epoch": 0.5610166926975544,
|
4929 |
+
"grad_norm": 82.875,
|
4930 |
+
"learning_rate": 9.989042647858995e-07,
|
4931 |
+
"loss": 131.2556,
|
4932 |
+
"step": 7010
|
4933 |
+
},
|
4934 |
+
{
|
4935 |
+
"epoch": 0.5618170018169518,
|
4936 |
+
"grad_norm": 79.8125,
|
4937 |
+
"learning_rate": 9.989027016828835e-07,
|
4938 |
+
"loss": 131.6359,
|
4939 |
+
"step": 7020
|
4940 |
+
},
|
4941 |
+
{
|
4942 |
+
"epoch": 0.5626173109363491,
|
4943 |
+
"grad_norm": 78.875,
|
4944 |
+
"learning_rate": 9.989011385798677e-07,
|
4945 |
+
"loss": 132.1506,
|
4946 |
+
"step": 7030
|
4947 |
+
},
|
4948 |
+
{
|
4949 |
+
"epoch": 0.5634176200557466,
|
4950 |
+
"grad_norm": 74.5625,
|
4951 |
+
"learning_rate": 9.98899575476852e-07,
|
4952 |
+
"loss": 131.6759,
|
4953 |
+
"step": 7040
|
4954 |
+
},
|
4955 |
+
{
|
4956 |
+
"epoch": 0.5642179291751439,
|
4957 |
+
"grad_norm": 76.5,
|
4958 |
+
"learning_rate": 9.988980123738362e-07,
|
4959 |
+
"loss": 131.1462,
|
4960 |
+
"step": 7050
|
4961 |
+
},
|
4962 |
+
{
|
4963 |
+
"epoch": 0.5650182382945412,
|
4964 |
+
"grad_norm": 82.625,
|
4965 |
+
"learning_rate": 9.988964492708202e-07,
|
4966 |
+
"loss": 130.7009,
|
4967 |
+
"step": 7060
|
4968 |
+
},
|
4969 |
+
{
|
4970 |
+
"epoch": 0.5658185474139387,
|
4971 |
+
"grad_norm": 84.875,
|
4972 |
+
"learning_rate": 9.988948861678044e-07,
|
4973 |
+
"loss": 134.6625,
|
4974 |
+
"step": 7070
|
4975 |
+
},
|
4976 |
+
{
|
4977 |
+
"epoch": 0.566618856533336,
|
4978 |
+
"grad_norm": 77.1875,
|
4979 |
+
"learning_rate": 9.988933230647886e-07,
|
4980 |
+
"loss": 132.1597,
|
4981 |
+
"step": 7080
|
4982 |
+
},
|
4983 |
+
{
|
4984 |
+
"epoch": 0.5674191656527333,
|
4985 |
+
"grad_norm": 75.6875,
|
4986 |
+
"learning_rate": 9.988917599617726e-07,
|
4987 |
+
"loss": 131.5931,
|
4988 |
+
"step": 7090
|
4989 |
+
},
|
4990 |
+
{
|
4991 |
+
"epoch": 0.5682194747721308,
|
4992 |
+
"grad_norm": 73.6875,
|
4993 |
+
"learning_rate": 9.988901968587568e-07,
|
4994 |
+
"loss": 131.8727,
|
4995 |
+
"step": 7100
|
4996 |
+
},
|
4997 |
+
{
|
4998 |
+
"epoch": 0.5690197838915281,
|
4999 |
+
"grad_norm": 73.25,
|
5000 |
+
"learning_rate": 9.98888633755741e-07,
|
5001 |
+
"loss": 132.6044,
|
5002 |
+
"step": 7110
|
5003 |
+
},
|
5004 |
+
{
|
5005 |
+
"epoch": 0.5698200930109255,
|
5006 |
+
"grad_norm": 85.75,
|
5007 |
+
"learning_rate": 9.98887070652725e-07,
|
5008 |
+
"loss": 131.7462,
|
5009 |
+
"step": 7120
|
5010 |
+
},
|
5011 |
+
{
|
5012 |
+
"epoch": 0.5706204021303228,
|
5013 |
+
"grad_norm": 74.0625,
|
5014 |
+
"learning_rate": 9.988855075497093e-07,
|
5015 |
+
"loss": 132.0296,
|
5016 |
+
"step": 7130
|
5017 |
+
},
|
5018 |
+
{
|
5019 |
+
"epoch": 0.5714207112497202,
|
5020 |
+
"grad_norm": 80.6875,
|
5021 |
+
"learning_rate": 9.988839444466935e-07,
|
5022 |
+
"loss": 134.3069,
|
5023 |
+
"step": 7140
|
5024 |
+
},
|
5025 |
+
{
|
5026 |
+
"epoch": 0.5722210203691176,
|
5027 |
+
"grad_norm": 76.0,
|
5028 |
+
"learning_rate": 9.988823813436777e-07,
|
5029 |
+
"loss": 131.401,
|
5030 |
+
"step": 7150
|
5031 |
+
},
|
5032 |
+
{
|
5033 |
+
"epoch": 0.5730213294885149,
|
5034 |
+
"grad_norm": 81.5625,
|
5035 |
+
"learning_rate": 9.98880818240662e-07,
|
5036 |
+
"loss": 132.4782,
|
5037 |
+
"step": 7160
|
5038 |
+
},
|
5039 |
+
{
|
5040 |
+
"epoch": 0.5738216386079124,
|
5041 |
+
"grad_norm": 76.4375,
|
5042 |
+
"learning_rate": 9.988792551376461e-07,
|
5043 |
+
"loss": 132.7547,
|
5044 |
+
"step": 7170
|
5045 |
+
},
|
5046 |
+
{
|
5047 |
+
"epoch": 0.5746219477273097,
|
5048 |
+
"grad_norm": 72.9375,
|
5049 |
+
"learning_rate": 9.988776920346301e-07,
|
5050 |
+
"loss": 131.6157,
|
5051 |
+
"step": 7180
|
5052 |
+
},
|
5053 |
+
{
|
5054 |
+
"epoch": 0.575422256846707,
|
5055 |
+
"grad_norm": 75.5625,
|
5056 |
+
"learning_rate": 9.988761289316144e-07,
|
5057 |
+
"loss": 132.9658,
|
5058 |
+
"step": 7190
|
5059 |
+
},
|
5060 |
+
{
|
5061 |
+
"epoch": 0.5762225659661044,
|
5062 |
+
"grad_norm": 73.3125,
|
5063 |
+
"learning_rate": 9.988745658285986e-07,
|
5064 |
+
"loss": 131.4694,
|
5065 |
+
"step": 7200
|
5066 |
+
},
|
5067 |
+
{
|
5068 |
+
"epoch": 0.5770228750855018,
|
5069 |
+
"grad_norm": 76.6875,
|
5070 |
+
"learning_rate": 9.988730027255828e-07,
|
5071 |
+
"loss": 132.546,
|
5072 |
+
"step": 7210
|
5073 |
+
},
|
5074 |
+
{
|
5075 |
+
"epoch": 0.5778231842048991,
|
5076 |
+
"grad_norm": 77.6875,
|
5077 |
+
"learning_rate": 9.988714396225668e-07,
|
5078 |
+
"loss": 131.9255,
|
5079 |
+
"step": 7220
|
5080 |
+
},
|
5081 |
+
{
|
5082 |
+
"epoch": 0.5786234933242965,
|
5083 |
+
"grad_norm": 82.0625,
|
5084 |
+
"learning_rate": 9.98869876519551e-07,
|
5085 |
+
"loss": 133.1269,
|
5086 |
+
"step": 7230
|
5087 |
+
},
|
5088 |
+
{
|
5089 |
+
"epoch": 0.5794238024436938,
|
5090 |
+
"grad_norm": 75.5625,
|
5091 |
+
"learning_rate": 9.988683134165352e-07,
|
5092 |
+
"loss": 131.8465,
|
5093 |
+
"step": 7240
|
5094 |
+
},
|
5095 |
+
{
|
5096 |
+
"epoch": 0.5802241115630913,
|
5097 |
+
"grad_norm": 78.125,
|
5098 |
+
"learning_rate": 9.988667503135193e-07,
|
5099 |
+
"loss": 131.8865,
|
5100 |
+
"step": 7250
|
5101 |
+
},
|
5102 |
+
{
|
5103 |
+
"epoch": 0.5810244206824886,
|
5104 |
+
"grad_norm": 77.625,
|
5105 |
+
"learning_rate": 9.988651872105035e-07,
|
5106 |
+
"loss": 131.2077,
|
5107 |
+
"step": 7260
|
5108 |
+
},
|
5109 |
+
{
|
5110 |
+
"epoch": 0.5818247298018859,
|
5111 |
+
"grad_norm": 73.0,
|
5112 |
+
"learning_rate": 9.988636241074877e-07,
|
5113 |
+
"loss": 132.2729,
|
5114 |
+
"step": 7270
|
5115 |
+
},
|
5116 |
+
{
|
5117 |
+
"epoch": 0.5826250389212834,
|
5118 |
+
"grad_norm": 76.125,
|
5119 |
+
"learning_rate": 9.988620610044717e-07,
|
5120 |
+
"loss": 132.5173,
|
5121 |
+
"step": 7280
|
5122 |
+
},
|
5123 |
+
{
|
5124 |
+
"epoch": 0.5834253480406807,
|
5125 |
+
"grad_norm": 72.875,
|
5126 |
+
"learning_rate": 9.98860497901456e-07,
|
5127 |
+
"loss": 130.8854,
|
5128 |
+
"step": 7290
|
5129 |
+
},
|
5130 |
+
{
|
5131 |
+
"epoch": 0.5842256571600781,
|
5132 |
+
"grad_norm": 76.5625,
|
5133 |
+
"learning_rate": 9.988589347984401e-07,
|
5134 |
+
"loss": 130.3204,
|
5135 |
+
"step": 7300
|
5136 |
+
},
|
5137 |
+
{
|
5138 |
+
"epoch": 0.5850259662794755,
|
5139 |
+
"grad_norm": 79.375,
|
5140 |
+
"learning_rate": 9.988573716954244e-07,
|
5141 |
+
"loss": 131.2047,
|
5142 |
+
"step": 7310
|
5143 |
+
},
|
5144 |
+
{
|
5145 |
+
"epoch": 0.5858262753988728,
|
5146 |
+
"grad_norm": 76.6875,
|
5147 |
+
"learning_rate": 9.988558085924086e-07,
|
5148 |
+
"loss": 132.3948,
|
5149 |
+
"step": 7320
|
5150 |
+
},
|
5151 |
+
{
|
5152 |
+
"epoch": 0.5866265845182702,
|
5153 |
+
"grad_norm": 75.0625,
|
5154 |
+
"learning_rate": 9.988542454893926e-07,
|
5155 |
+
"loss": 131.0941,
|
5156 |
+
"step": 7330
|
5157 |
+
},
|
5158 |
+
{
|
5159 |
+
"epoch": 0.5874268936376675,
|
5160 |
+
"grad_norm": 80.375,
|
5161 |
+
"learning_rate": 9.988526823863768e-07,
|
5162 |
+
"loss": 132.392,
|
5163 |
+
"step": 7340
|
5164 |
+
},
|
5165 |
+
{
|
5166 |
+
"epoch": 0.588227202757065,
|
5167 |
+
"grad_norm": 75.3125,
|
5168 |
+
"learning_rate": 9.98851119283361e-07,
|
5169 |
+
"loss": 132.1595,
|
5170 |
+
"step": 7350
|
5171 |
+
},
|
5172 |
+
{
|
5173 |
+
"epoch": 0.5890275118764623,
|
5174 |
+
"grad_norm": 80.5,
|
5175 |
+
"learning_rate": 9.988495561803452e-07,
|
5176 |
+
"loss": 131.9138,
|
5177 |
+
"step": 7360
|
5178 |
+
},
|
5179 |
+
{
|
5180 |
+
"epoch": 0.5898278209958596,
|
5181 |
+
"grad_norm": 74.5625,
|
5182 |
+
"learning_rate": 9.988479930773294e-07,
|
5183 |
+
"loss": 129.6595,
|
5184 |
+
"step": 7370
|
5185 |
+
},
|
5186 |
+
{
|
5187 |
+
"epoch": 0.590628130115257,
|
5188 |
+
"grad_norm": 74.5625,
|
5189 |
+
"learning_rate": 9.988464299743135e-07,
|
5190 |
+
"loss": 130.2861,
|
5191 |
+
"step": 7380
|
5192 |
+
},
|
5193 |
+
{
|
5194 |
+
"epoch": 0.5914284392346544,
|
5195 |
+
"grad_norm": 71.625,
|
5196 |
+
"learning_rate": 9.988448668712977e-07,
|
5197 |
+
"loss": 132.6865,
|
5198 |
+
"step": 7390
|
5199 |
+
},
|
5200 |
+
{
|
5201 |
+
"epoch": 0.5922287483540517,
|
5202 |
+
"grad_norm": 80.25,
|
5203 |
+
"learning_rate": 9.988433037682819e-07,
|
5204 |
+
"loss": 132.5223,
|
5205 |
+
"step": 7400
|
5206 |
+
},
|
5207 |
+
{
|
5208 |
+
"epoch": 0.5930290574734491,
|
5209 |
+
"grad_norm": 70.5,
|
5210 |
+
"learning_rate": 9.98841740665266e-07,
|
5211 |
+
"loss": 132.5188,
|
5212 |
+
"step": 7410
|
5213 |
+
},
|
5214 |
+
{
|
5215 |
+
"epoch": 0.5938293665928465,
|
5216 |
+
"grad_norm": 71.4375,
|
5217 |
+
"learning_rate": 9.988401775622501e-07,
|
5218 |
+
"loss": 132.2418,
|
5219 |
+
"step": 7420
|
5220 |
+
},
|
5221 |
+
{
|
5222 |
+
"epoch": 0.5946296757122439,
|
5223 |
+
"grad_norm": 75.25,
|
5224 |
+
"learning_rate": 9.988386144592343e-07,
|
5225 |
+
"loss": 130.8538,
|
5226 |
+
"step": 7430
|
5227 |
+
},
|
5228 |
+
{
|
5229 |
+
"epoch": 0.5954299848316412,
|
5230 |
+
"grad_norm": 75.8125,
|
5231 |
+
"learning_rate": 9.988370513562183e-07,
|
5232 |
+
"loss": 131.213,
|
5233 |
+
"step": 7440
|
5234 |
+
},
|
5235 |
+
{
|
5236 |
+
"epoch": 0.5962302939510385,
|
5237 |
+
"grad_norm": 74.0,
|
5238 |
+
"learning_rate": 9.988354882532026e-07,
|
5239 |
+
"loss": 132.8653,
|
5240 |
+
"step": 7450
|
5241 |
+
},
|
5242 |
+
{
|
5243 |
+
"epoch": 0.597030603070436,
|
5244 |
+
"grad_norm": 74.25,
|
5245 |
+
"learning_rate": 9.988339251501868e-07,
|
5246 |
+
"loss": 131.1389,
|
5247 |
+
"step": 7460
|
5248 |
+
},
|
5249 |
+
{
|
5250 |
+
"epoch": 0.5978309121898333,
|
5251 |
+
"grad_norm": 79.0625,
|
5252 |
+
"learning_rate": 9.98832362047171e-07,
|
5253 |
+
"loss": 131.8387,
|
5254 |
+
"step": 7470
|
5255 |
+
},
|
5256 |
+
{
|
5257 |
+
"epoch": 0.5986312213092307,
|
5258 |
+
"grad_norm": 92.9375,
|
5259 |
+
"learning_rate": 9.988307989441552e-07,
|
5260 |
+
"loss": 132.9031,
|
5261 |
+
"step": 7480
|
5262 |
+
},
|
5263 |
+
{
|
5264 |
+
"epoch": 0.5994315304286281,
|
5265 |
+
"grad_norm": 75.75,
|
5266 |
+
"learning_rate": 9.988292358411392e-07,
|
5267 |
+
"loss": 130.4166,
|
5268 |
+
"step": 7490
|
5269 |
+
},
|
5270 |
+
{
|
5271 |
+
"epoch": 0.6002318395480254,
|
5272 |
+
"grad_norm": 75.25,
|
5273 |
+
"learning_rate": 9.988276727381234e-07,
|
5274 |
+
"loss": 131.6747,
|
5275 |
+
"step": 7500
|
5276 |
+
},
|
5277 |
+
{
|
5278 |
+
"epoch": 0.6002318395480254,
|
5279 |
+
"eval_loss": 2.0600602626800537,
|
5280 |
+
"eval_runtime": 418.6157,
|
5281 |
+
"eval_samples_per_second": 1567.447,
|
5282 |
+
"eval_steps_per_second": 48.983,
|
5283 |
+
"step": 7500
|
5284 |
}
|
5285 |
],
|
5286 |
"logging_steps": 10,
|
|
|
5300 |
"attributes": {}
|
5301 |
}
|
5302 |
},
|
5303 |
+
"total_flos": 2.0708374333095936e+19,
|
5304 |
"train_batch_size": 4,
|
5305 |
"trial_name": null,
|
5306 |
"trial_params": null
|