Training in progress, step 1389, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6159f6e7e904bfec19903400d3a3481453cbf88b8e5b75fcc9d6ed7e25f5379d
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:89b2ceb23199d4ac2e2e3d0868a2ec5d8c431569dc6d6075dfc563be95e44ada
|
3 |
size 43123028
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a3600e0e149644d8141635da3bc070c04eb57d99d50d63d93ea0b79077ab6e8e
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cbb9dafd3c48beed5e8df7fc3a6cd3b0e9d3d9db9628e4eeb4da2ef5df67d040
|
3 |
size 1064
|
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": 348,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7347,6 +7347,2421 @@
|
|
7347 |
"eval_samples_per_second": 27.23,
|
7348 |
"eval_steps_per_second": 13.615,
|
7349 |
"step": 1044
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7350 |
}
|
7351 |
],
|
7352 |
"logging_steps": 1,
|
@@ -7361,12 +9776,12 @@
|
|
7361 |
"should_evaluate": false,
|
7362 |
"should_log": false,
|
7363 |
"should_save": true,
|
7364 |
-
"should_training_stop":
|
7365 |
},
|
7366 |
"attributes": {}
|
7367 |
}
|
7368 |
},
|
7369 |
-
"total_flos": 2.
|
7370 |
"train_batch_size": 2,
|
7371 |
"trial_name": null,
|
7372 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.27237964506324147,
|
5 |
"eval_steps": 348,
|
6 |
+
"global_step": 1389,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7347 |
"eval_samples_per_second": 27.23,
|
7348 |
"eval_steps_per_second": 13.615,
|
7349 |
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.20492205118148837,
|
7353 |
+
"grad_norm": 4.138796329498291,
|
7354 |
+
"learning_rate": 2.9168607092569845e-05,
|
7355 |
+
"loss": 0.6768,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.20511814883812138,
|
7360 |
+
"grad_norm": 4.2951250076293945,
|
7361 |
+
"learning_rate": 2.9007975767533714e-05,
|
7362 |
+
"loss": 1.0667,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.20531424649475438,
|
7367 |
+
"grad_norm": 6.767472743988037,
|
7368 |
+
"learning_rate": 2.8847712894214686e-05,
|
7369 |
+
"loss": 2.2965,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.2055103441513874,
|
7374 |
+
"grad_norm": 5.017336845397949,
|
7375 |
+
"learning_rate": 2.8687819304384066e-05,
|
7376 |
+
"loss": 1.457,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.2057064418080204,
|
7381 |
+
"grad_norm": 5.76704740524292,
|
7382 |
+
"learning_rate": 2.852829582789669e-05,
|
7383 |
+
"loss": 1.3418,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.2059025394646534,
|
7388 |
+
"grad_norm": 5.123453617095947,
|
7389 |
+
"learning_rate": 2.8369143292686306e-05,
|
7390 |
+
"loss": 1.0328,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.2060986371212864,
|
7395 |
+
"grad_norm": 6.851526260375977,
|
7396 |
+
"learning_rate": 2.821036252476156e-05,
|
7397 |
+
"loss": 2.0026,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.2062947347779194,
|
7402 |
+
"grad_norm": 5.268173694610596,
|
7403 |
+
"learning_rate": 2.8051954348201613e-05,
|
7404 |
+
"loss": 0.69,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.2064908324345524,
|
7409 |
+
"grad_norm": 6.25408935546875,
|
7410 |
+
"learning_rate": 2.789391958515183e-05,
|
7411 |
+
"loss": 1.2385,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.2066869300911854,
|
7416 |
+
"grad_norm": 4.305760383605957,
|
7417 |
+
"learning_rate": 2.7736259055819568e-05,
|
7418 |
+
"loss": 0.747,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.20688302774781842,
|
7423 |
+
"grad_norm": 9.248473167419434,
|
7424 |
+
"learning_rate": 2.757897357846988e-05,
|
7425 |
+
"loss": 1.4235,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.20707912540445142,
|
7430 |
+
"grad_norm": 4.8141984939575195,
|
7431 |
+
"learning_rate": 2.7422063969421285e-05,
|
7432 |
+
"loss": 0.5879,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.20727522306108442,
|
7437 |
+
"grad_norm": 4.95810604095459,
|
7438 |
+
"learning_rate": 2.7265531043041535e-05,
|
7439 |
+
"loss": 0.8899,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.20747132071771743,
|
7444 |
+
"grad_norm": 8.220407485961914,
|
7445 |
+
"learning_rate": 2.710937561174337e-05,
|
7446 |
+
"loss": 1.5904,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.20766741837435043,
|
7451 |
+
"grad_norm": 5.406398296356201,
|
7452 |
+
"learning_rate": 2.6953598485980336e-05,
|
7453 |
+
"loss": 2.2579,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.20786351603098344,
|
7458 |
+
"grad_norm": 7.9236273765563965,
|
7459 |
+
"learning_rate": 2.679820047424253e-05,
|
7460 |
+
"loss": 1.6289,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.20805961368761644,
|
7465 |
+
"grad_norm": 11.548074722290039,
|
7466 |
+
"learning_rate": 2.6643182383052446e-05,
|
7467 |
+
"loss": 1.3829,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.20825571134424944,
|
7472 |
+
"grad_norm": 6.566219329833984,
|
7473 |
+
"learning_rate": 2.6488545016960776e-05,
|
7474 |
+
"loss": 1.5911,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.20845180900088245,
|
7479 |
+
"grad_norm": 6.300790786743164,
|
7480 |
+
"learning_rate": 2.6334289178542226e-05,
|
7481 |
+
"loss": 0.8637,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.20864790665751545,
|
7486 |
+
"grad_norm": 5.563969135284424,
|
7487 |
+
"learning_rate": 2.6180415668391356e-05,
|
7488 |
+
"loss": 1.388,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.20884400431414846,
|
7493 |
+
"grad_norm": 6.617629528045654,
|
7494 |
+
"learning_rate": 2.602692528511843e-05,
|
7495 |
+
"loss": 1.1707,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.20904010197078146,
|
7500 |
+
"grad_norm": 6.338384628295898,
|
7501 |
+
"learning_rate": 2.5873818825345254e-05,
|
7502 |
+
"loss": 1.3016,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.20923619962741447,
|
7507 |
+
"grad_norm": 5.3434271812438965,
|
7508 |
+
"learning_rate": 2.5721097083701084e-05,
|
7509 |
+
"loss": 1.5446,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.20943229728404747,
|
7514 |
+
"grad_norm": 28.035884857177734,
|
7515 |
+
"learning_rate": 2.556876085281843e-05,
|
7516 |
+
"loss": 1.168,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.20962839494068045,
|
7521 |
+
"grad_norm": 14.401495933532715,
|
7522 |
+
"learning_rate": 2.5416810923329028e-05,
|
7523 |
+
"loss": 1.1242,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.20982449259731345,
|
7528 |
+
"grad_norm": 6.766130447387695,
|
7529 |
+
"learning_rate": 2.5265248083859648e-05,
|
7530 |
+
"loss": 1.4922,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.21002059025394645,
|
7535 |
+
"grad_norm": 5.8070292472839355,
|
7536 |
+
"learning_rate": 2.5114073121028093e-05,
|
7537 |
+
"loss": 1.7635,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.21021668791057946,
|
7542 |
+
"grad_norm": 5.450509548187256,
|
7543 |
+
"learning_rate": 2.4963286819439037e-05,
|
7544 |
+
"loss": 0.7839,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.21041278556721246,
|
7549 |
+
"grad_norm": 4.60319185256958,
|
7550 |
+
"learning_rate": 2.4812889961679986e-05,
|
7551 |
+
"loss": 1.8491,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.21060888322384547,
|
7556 |
+
"grad_norm": 3.9338133335113525,
|
7557 |
+
"learning_rate": 2.4662883328317222e-05,
|
7558 |
+
"loss": 0.9618,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.21080498088047847,
|
7563 |
+
"grad_norm": 7.811015605926514,
|
7564 |
+
"learning_rate": 2.451326769789176e-05,
|
7565 |
+
"loss": 1.5204,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.21100107853711147,
|
7570 |
+
"grad_norm": 6.122691631317139,
|
7571 |
+
"learning_rate": 2.4364043846915274e-05,
|
7572 |
+
"loss": 1.0331,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.21119717619374448,
|
7577 |
+
"grad_norm": 8.609393119812012,
|
7578 |
+
"learning_rate": 2.4215212549866116e-05,
|
7579 |
+
"loss": 1.9155,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.21139327385037748,
|
7584 |
+
"grad_norm": 7.292634963989258,
|
7585 |
+
"learning_rate": 2.4066774579185158e-05,
|
7586 |
+
"loss": 1.2971,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.2115893715070105,
|
7591 |
+
"grad_norm": 5.975107192993164,
|
7592 |
+
"learning_rate": 2.3918730705272064e-05,
|
7593 |
+
"loss": 1.4156,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.2117854691636435,
|
7598 |
+
"grad_norm": 14.805880546569824,
|
7599 |
+
"learning_rate": 2.377108169648098e-05,
|
7600 |
+
"loss": 1.0787,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.2119815668202765,
|
7605 |
+
"grad_norm": 5.720880031585693,
|
7606 |
+
"learning_rate": 2.3623828319116748e-05,
|
7607 |
+
"loss": 1.3329,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.2121776644769095,
|
7612 |
+
"grad_norm": 7.515237331390381,
|
7613 |
+
"learning_rate": 2.3476971337430875e-05,
|
7614 |
+
"loss": 1.7147,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.2123737621335425,
|
7619 |
+
"grad_norm": 6.027990341186523,
|
7620 |
+
"learning_rate": 2.3330511513617448e-05,
|
7621 |
+
"loss": 1.9171,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.2125698597901755,
|
7626 |
+
"grad_norm": 4.526289463043213,
|
7627 |
+
"learning_rate": 2.318444960780949e-05,
|
7628 |
+
"loss": 0.6467,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.2127659574468085,
|
7633 |
+
"grad_norm": 9.184370040893555,
|
7634 |
+
"learning_rate": 2.3038786378074574e-05,
|
7635 |
+
"loss": 1.6125,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.21296205510344152,
|
7640 |
+
"grad_norm": 7.263781547546387,
|
7641 |
+
"learning_rate": 2.289352258041133e-05,
|
7642 |
+
"loss": 1.6612,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.21315815276007452,
|
7647 |
+
"grad_norm": 9.886373519897461,
|
7648 |
+
"learning_rate": 2.274865896874523e-05,
|
7649 |
+
"loss": 2.4533,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.21335425041670752,
|
7654 |
+
"grad_norm": 9.56613540649414,
|
7655 |
+
"learning_rate": 2.2604196294924694e-05,
|
7656 |
+
"loss": 2.0031,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.21355034807334053,
|
7661 |
+
"grad_norm": 4.8680925369262695,
|
7662 |
+
"learning_rate": 2.2460135308717445e-05,
|
7663 |
+
"loss": 1.2974,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.21374644572997353,
|
7668 |
+
"grad_norm": 4.032529830932617,
|
7669 |
+
"learning_rate": 2.231647675780619e-05,
|
7670 |
+
"loss": 0.7491,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.21394254338660654,
|
7675 |
+
"grad_norm": 5.186695098876953,
|
7676 |
+
"learning_rate": 2.2173221387785216e-05,
|
7677 |
+
"loss": 1.2152,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.21413864104323954,
|
7682 |
+
"grad_norm": 5.538721561431885,
|
7683 |
+
"learning_rate": 2.2030369942156072e-05,
|
7684 |
+
"loss": 1.6134,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.21433473869987255,
|
7689 |
+
"grad_norm": 8.213644027709961,
|
7690 |
+
"learning_rate": 2.1887923162324097e-05,
|
7691 |
+
"loss": 1.5288,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.21453083635650555,
|
7696 |
+
"grad_norm": 8.110743522644043,
|
7697 |
+
"learning_rate": 2.1745881787594334e-05,
|
7698 |
+
"loss": 0.8424,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.21472693401313855,
|
7703 |
+
"grad_norm": 6.066911697387695,
|
7704 |
+
"learning_rate": 2.1604246555167638e-05,
|
7705 |
+
"loss": 1.9787,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.21492303166977156,
|
7710 |
+
"grad_norm": 3.873046636581421,
|
7711 |
+
"learning_rate": 2.1463018200137196e-05,
|
7712 |
+
"loss": 1.7871,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.21511912932640456,
|
7717 |
+
"grad_norm": 7.206579208374023,
|
7718 |
+
"learning_rate": 2.1322197455484248e-05,
|
7719 |
+
"loss": 1.3871,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.21531522698303757,
|
7724 |
+
"grad_norm": 7.201257705688477,
|
7725 |
+
"learning_rate": 2.1181785052074756e-05,
|
7726 |
+
"loss": 1.2177,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.21551132463967054,
|
7731 |
+
"grad_norm": 5.7325239181518555,
|
7732 |
+
"learning_rate": 2.104178171865513e-05,
|
7733 |
+
"loss": 1.0547,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.21570742229630355,
|
7738 |
+
"grad_norm": 4.294105052947998,
|
7739 |
+
"learning_rate": 2.0902188181848838e-05,
|
7740 |
+
"loss": 1.9177,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.21590351995293655,
|
7745 |
+
"grad_norm": 4.344897270202637,
|
7746 |
+
"learning_rate": 2.0763005166152517e-05,
|
7747 |
+
"loss": 1.2101,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.21609961760956956,
|
7752 |
+
"grad_norm": 10.920205116271973,
|
7753 |
+
"learning_rate": 2.0624233393932024e-05,
|
7754 |
+
"loss": 1.0783,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.21629571526620256,
|
7759 |
+
"grad_norm": 6.155653953552246,
|
7760 |
+
"learning_rate": 2.0485873585419035e-05,
|
7761 |
+
"loss": 2.6903,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.21649181292283556,
|
7766 |
+
"grad_norm": 7.211439609527588,
|
7767 |
+
"learning_rate": 2.0347926458706945e-05,
|
7768 |
+
"loss": 1.8343,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.21668791057946857,
|
7773 |
+
"grad_norm": 8.261204719543457,
|
7774 |
+
"learning_rate": 2.021039272974742e-05,
|
7775 |
+
"loss": 2.3539,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.21688400823610157,
|
7780 |
+
"grad_norm": 5.709137916564941,
|
7781 |
+
"learning_rate": 2.0073273112346526e-05,
|
7782 |
+
"loss": 1.0312,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.21708010589273458,
|
7787 |
+
"grad_norm": 6.048385143280029,
|
7788 |
+
"learning_rate": 1.9936568318161076e-05,
|
7789 |
+
"loss": 1.086,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.21727620354936758,
|
7794 |
+
"grad_norm": 4.2964677810668945,
|
7795 |
+
"learning_rate": 1.9800279056695005e-05,
|
7796 |
+
"loss": 0.7691,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.21747230120600058,
|
7801 |
+
"grad_norm": 4.8451972007751465,
|
7802 |
+
"learning_rate": 1.966440603529549e-05,
|
7803 |
+
"loss": 1.4864,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.2176683988626336,
|
7808 |
+
"grad_norm": 5.61348295211792,
|
7809 |
+
"learning_rate": 1.952894995914949e-05,
|
7810 |
+
"loss": 1.5675,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.2178644965192666,
|
7815 |
+
"grad_norm": 6.666314601898193,
|
7816 |
+
"learning_rate": 1.9393911531279974e-05,
|
7817 |
+
"loss": 1.3273,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.2180605941758996,
|
7822 |
+
"grad_norm": 5.22393798828125,
|
7823 |
+
"learning_rate": 1.9259291452542293e-05,
|
7824 |
+
"loss": 1.4175,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.2182566918325326,
|
7829 |
+
"grad_norm": 7.283446311950684,
|
7830 |
+
"learning_rate": 1.9125090421620574e-05,
|
7831 |
+
"loss": 2.5381,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.2184527894891656,
|
7836 |
+
"grad_norm": 8.553428649902344,
|
7837 |
+
"learning_rate": 1.8991309135024004e-05,
|
7838 |
+
"loss": 2.0767,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.2186488871457986,
|
7843 |
+
"grad_norm": 6.280401706695557,
|
7844 |
+
"learning_rate": 1.8857948287083416e-05,
|
7845 |
+
"loss": 0.9069,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.2188449848024316,
|
7850 |
+
"grad_norm": 14.387880325317383,
|
7851 |
+
"learning_rate": 1.8725008569947365e-05,
|
7852 |
+
"loss": 1.6501,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.21904108245906462,
|
7857 |
+
"grad_norm": 8.282727241516113,
|
7858 |
+
"learning_rate": 1.8592490673578843e-05,
|
7859 |
+
"loss": 1.645,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.21923718011569762,
|
7864 |
+
"grad_norm": 7.460145950317383,
|
7865 |
+
"learning_rate": 1.8460395285751542e-05,
|
7866 |
+
"loss": 2.3436,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.21943327777233063,
|
7871 |
+
"grad_norm": 4.364814758300781,
|
7872 |
+
"learning_rate": 1.8328723092046317e-05,
|
7873 |
+
"loss": 1.1313,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.21962937542896363,
|
7878 |
+
"grad_norm": 16.233600616455078,
|
7879 |
+
"learning_rate": 1.8197474775847613e-05,
|
7880 |
+
"loss": 2.5969,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.21982547308559663,
|
7885 |
+
"grad_norm": 7.081477165222168,
|
7886 |
+
"learning_rate": 1.806665101833994e-05,
|
7887 |
+
"loss": 2.4986,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.22002157074222964,
|
7892 |
+
"grad_norm": 4.212589740753174,
|
7893 |
+
"learning_rate": 1.7936252498504356e-05,
|
7894 |
+
"loss": 1.3044,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.22021766839886264,
|
7899 |
+
"grad_norm": 6.28343391418457,
|
7900 |
+
"learning_rate": 1.7806279893114875e-05,
|
7901 |
+
"loss": 1.7792,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.22041376605549565,
|
7906 |
+
"grad_norm": 6.392016410827637,
|
7907 |
+
"learning_rate": 1.7676733876735018e-05,
|
7908 |
+
"loss": 1.6957,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.22060986371212865,
|
7913 |
+
"grad_norm": 8.44856071472168,
|
7914 |
+
"learning_rate": 1.754761512171429e-05,
|
7915 |
+
"loss": 2.5809,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.22080596136876166,
|
7920 |
+
"grad_norm": 6.063395977020264,
|
7921 |
+
"learning_rate": 1.741892429818468e-05,
|
7922 |
+
"loss": 1.1537,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.22100205902539466,
|
7927 |
+
"grad_norm": 22.53998374938965,
|
7928 |
+
"learning_rate": 1.729066207405722e-05,
|
7929 |
+
"loss": 2.6686,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.22119815668202766,
|
7934 |
+
"grad_norm": 4.210419178009033,
|
7935 |
+
"learning_rate": 1.7162829115018452e-05,
|
7936 |
+
"loss": 1.2635,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.22139425433866064,
|
7941 |
+
"grad_norm": 9.418660163879395,
|
7942 |
+
"learning_rate": 1.7035426084527062e-05,
|
7943 |
+
"loss": 2.5926,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.22159035199529364,
|
7948 |
+
"grad_norm": 8.299819946289062,
|
7949 |
+
"learning_rate": 1.690845364381034e-05,
|
7950 |
+
"loss": 1.5026,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.22178644965192665,
|
7955 |
+
"grad_norm": 7.302555084228516,
|
7956 |
+
"learning_rate": 1.6781912451860827e-05,
|
7957 |
+
"loss": 1.5762,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.22198254730855965,
|
7962 |
+
"grad_norm": 8.947975158691406,
|
7963 |
+
"learning_rate": 1.665580316543286e-05,
|
7964 |
+
"loss": 0.9516,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.22217864496519266,
|
7969 |
+
"grad_norm": 7.167786121368408,
|
7970 |
+
"learning_rate": 1.653012643903915e-05,
|
7971 |
+
"loss": 2.3246,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.22237474262182566,
|
7976 |
+
"grad_norm": 8.101836204528809,
|
7977 |
+
"learning_rate": 1.640488292494743e-05,
|
7978 |
+
"loss": 1.5492,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.22257084027845866,
|
7983 |
+
"grad_norm": 7.395213603973389,
|
7984 |
+
"learning_rate": 1.628007327317701e-05,
|
7985 |
+
"loss": 1.7188,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.22276693793509167,
|
7990 |
+
"grad_norm": 3.9345545768737793,
|
7991 |
+
"learning_rate": 1.6155698131495454e-05,
|
7992 |
+
"loss": 0.7864,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.22296303559172467,
|
7997 |
+
"grad_norm": 6.147872447967529,
|
7998 |
+
"learning_rate": 1.603175814541522e-05,
|
7999 |
+
"loss": 1.7044,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.22315913324835768,
|
8004 |
+
"grad_norm": 5.625443458557129,
|
8005 |
+
"learning_rate": 1.5908253958190256e-05,
|
8006 |
+
"loss": 2.0135,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.22335523090499068,
|
8011 |
+
"grad_norm": 8.224568367004395,
|
8012 |
+
"learning_rate": 1.5785186210812698e-05,
|
8013 |
+
"loss": 1.4209,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.22355132856162369,
|
8018 |
+
"grad_norm": 5.80567741394043,
|
8019 |
+
"learning_rate": 1.566255554200955e-05,
|
8020 |
+
"loss": 0.8693,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.2237474262182567,
|
8025 |
+
"grad_norm": 7.716653347015381,
|
8026 |
+
"learning_rate": 1.5540362588239364e-05,
|
8027 |
+
"loss": 2.0515,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.2239435238748897,
|
8032 |
+
"grad_norm": 5.845829963684082,
|
8033 |
+
"learning_rate": 1.5418607983688927e-05,
|
8034 |
+
"loss": 1.3831,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.2241396215315227,
|
8039 |
+
"grad_norm": 5.728200435638428,
|
8040 |
+
"learning_rate": 1.529729236026999e-05,
|
8041 |
+
"loss": 1.6435,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.2243357191881557,
|
8046 |
+
"grad_norm": 9.004356384277344,
|
8047 |
+
"learning_rate": 1.5176416347615885e-05,
|
8048 |
+
"loss": 2.4094,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.2245318168447887,
|
8053 |
+
"grad_norm": 7.60123872756958,
|
8054 |
+
"learning_rate": 1.5055980573078487e-05,
|
8055 |
+
"loss": 1.6575,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.2247279145014217,
|
8060 |
+
"grad_norm": 6.572809219360352,
|
8061 |
+
"learning_rate": 1.4935985661724727e-05,
|
8062 |
+
"loss": 1.5638,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.22492401215805471,
|
8067 |
+
"grad_norm": 5.095000267028809,
|
8068 |
+
"learning_rate": 1.4816432236333444e-05,
|
8069 |
+
"loss": 1.6261,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.22512010981468772,
|
8074 |
+
"grad_norm": 5.581939220428467,
|
8075 |
+
"learning_rate": 1.4697320917392188e-05,
|
8076 |
+
"loss": 1.0154,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.22531620747132072,
|
8081 |
+
"grad_norm": 7.922572135925293,
|
8082 |
+
"learning_rate": 1.4578652323093855e-05,
|
8083 |
+
"loss": 3.1376,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.22551230512795373,
|
8088 |
+
"grad_norm": 8.814900398254395,
|
8089 |
+
"learning_rate": 1.4460427069333726e-05,
|
8090 |
+
"loss": 1.8381,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.22570840278458673,
|
8095 |
+
"grad_norm": 8.666311264038086,
|
8096 |
+
"learning_rate": 1.4342645769705977e-05,
|
8097 |
+
"loss": 0.7864,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.22590450044121974,
|
8102 |
+
"grad_norm": 4.692161560058594,
|
8103 |
+
"learning_rate": 1.4225309035500778e-05,
|
8104 |
+
"loss": 1.2963,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.22610059809785274,
|
8109 |
+
"grad_norm": 4.93988561630249,
|
8110 |
+
"learning_rate": 1.4108417475700908e-05,
|
8111 |
+
"loss": 0.8273,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.22629669575448574,
|
8116 |
+
"grad_norm": 5.998587131500244,
|
8117 |
+
"learning_rate": 1.3991971696978645e-05,
|
8118 |
+
"loss": 2.5023,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.22649279341111875,
|
8123 |
+
"grad_norm": 7.898712635040283,
|
8124 |
+
"learning_rate": 1.3875972303692752e-05,
|
8125 |
+
"loss": 1.7302,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.22668889106775175,
|
8130 |
+
"grad_norm": 4.3550190925598145,
|
8131 |
+
"learning_rate": 1.376041989788508e-05,
|
8132 |
+
"loss": 0.6245,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.22688498872438476,
|
8137 |
+
"grad_norm": 9.550539016723633,
|
8138 |
+
"learning_rate": 1.3645315079277765e-05,
|
8139 |
+
"loss": 1.7358,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.22708108638101776,
|
8144 |
+
"grad_norm": 7.5146164894104,
|
8145 |
+
"learning_rate": 1.3530658445269783e-05,
|
8146 |
+
"loss": 1.5159,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.22727718403765074,
|
8151 |
+
"grad_norm": 8.449292182922363,
|
8152 |
+
"learning_rate": 1.341645059093415e-05,
|
8153 |
+
"loss": 2.5644,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.22747328169428374,
|
8158 |
+
"grad_norm": 7.365999698638916,
|
8159 |
+
"learning_rate": 1.3302692109014625e-05,
|
8160 |
+
"loss": 1.3292,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.22766937935091675,
|
8165 |
+
"grad_norm": 4.936117172241211,
|
8166 |
+
"learning_rate": 1.3189383589922665e-05,
|
8167 |
+
"loss": 0.636,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.22786547700754975,
|
8172 |
+
"grad_norm": 22.70414161682129,
|
8173 |
+
"learning_rate": 1.3076525621734526e-05,
|
8174 |
+
"loss": 1.5574,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.22806157466418275,
|
8179 |
+
"grad_norm": 6.079709053039551,
|
8180 |
+
"learning_rate": 1.2964118790187929e-05,
|
8181 |
+
"loss": 1.8188,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.22825767232081576,
|
8186 |
+
"grad_norm": 6.864228248596191,
|
8187 |
+
"learning_rate": 1.2852163678679341e-05,
|
8188 |
+
"loss": 1.2751,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.22845376997744876,
|
8193 |
+
"grad_norm": 3.384364366531372,
|
8194 |
+
"learning_rate": 1.2740660868260633e-05,
|
8195 |
+
"loss": 1.3381,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.22864986763408177,
|
8200 |
+
"grad_norm": 4.77728796005249,
|
8201 |
+
"learning_rate": 1.2629610937636283e-05,
|
8202 |
+
"loss": 1.9661,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.22884596529071477,
|
8207 |
+
"grad_norm": 4.837019443511963,
|
8208 |
+
"learning_rate": 1.251901446316035e-05,
|
8209 |
+
"loss": 1.7805,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.22904206294734777,
|
8214 |
+
"grad_norm": 4.927963733673096,
|
8215 |
+
"learning_rate": 1.2408872018833296e-05,
|
8216 |
+
"loss": 0.8884,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.22923816060398078,
|
8221 |
+
"grad_norm": 7.3830647468566895,
|
8222 |
+
"learning_rate": 1.2299184176299339e-05,
|
8223 |
+
"loss": 1.1489,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.22943425826061378,
|
8228 |
+
"grad_norm": 4.659049987792969,
|
8229 |
+
"learning_rate": 1.2189951504843112e-05,
|
8230 |
+
"loss": 0.6723,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.2296303559172468,
|
8235 |
+
"grad_norm": 10.427599906921387,
|
8236 |
+
"learning_rate": 1.2081174571386989e-05,
|
8237 |
+
"loss": 1.1733,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.2298264535738798,
|
8242 |
+
"grad_norm": 13.781340599060059,
|
8243 |
+
"learning_rate": 1.1972853940488015e-05,
|
8244 |
+
"loss": 2.0666,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.2300225512305128,
|
8249 |
+
"grad_norm": 4.749293804168701,
|
8250 |
+
"learning_rate": 1.1864990174335012e-05,
|
8251 |
+
"loss": 1.657,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.2302186488871458,
|
8256 |
+
"grad_norm": 5.438106536865234,
|
8257 |
+
"learning_rate": 1.17575838327457e-05,
|
8258 |
+
"loss": 2.2866,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.2304147465437788,
|
8263 |
+
"grad_norm": 5.253787517547607,
|
8264 |
+
"learning_rate": 1.165063547316363e-05,
|
8265 |
+
"loss": 0.7286,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.2306108442004118,
|
8270 |
+
"grad_norm": 5.084885120391846,
|
8271 |
+
"learning_rate": 1.1544145650655514e-05,
|
8272 |
+
"loss": 1.5842,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.2308069418570448,
|
8277 |
+
"grad_norm": 6.295192718505859,
|
8278 |
+
"learning_rate": 1.1438114917908193e-05,
|
8279 |
+
"loss": 1.4375,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.23100303951367782,
|
8284 |
+
"grad_norm": 7.397315502166748,
|
8285 |
+
"learning_rate": 1.1332543825225806e-05,
|
8286 |
+
"loss": 1.5273,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.23119913717031082,
|
8291 |
+
"grad_norm": 4.620631217956543,
|
8292 |
+
"learning_rate": 1.122743292052697e-05,
|
8293 |
+
"loss": 2.3138,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.23139523482694382,
|
8298 |
+
"grad_norm": 6.895712852478027,
|
8299 |
+
"learning_rate": 1.1122782749341843e-05,
|
8300 |
+
"loss": 2.7047,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.23159133248357683,
|
8305 |
+
"grad_norm": 5.031332969665527,
|
8306 |
+
"learning_rate": 1.1018593854809478e-05,
|
8307 |
+
"loss": 1.3471,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.23178743014020983,
|
8312 |
+
"grad_norm": 5.385929584503174,
|
8313 |
+
"learning_rate": 1.0914866777674737e-05,
|
8314 |
+
"loss": 1.7638,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.23198352779684284,
|
8319 |
+
"grad_norm": 6.311648368835449,
|
8320 |
+
"learning_rate": 1.081160205628572e-05,
|
8321 |
+
"loss": 1.4686,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.23217962545347584,
|
8326 |
+
"grad_norm": 7.770483016967773,
|
8327 |
+
"learning_rate": 1.0708800226590854e-05,
|
8328 |
+
"loss": 0.9346,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.23237572311010884,
|
8333 |
+
"grad_norm": 7.618231773376465,
|
8334 |
+
"learning_rate": 1.0606461822136137e-05,
|
8335 |
+
"loss": 1.0662,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.23257182076674185,
|
8340 |
+
"grad_norm": 8.606494903564453,
|
8341 |
+
"learning_rate": 1.0504587374062391e-05,
|
8342 |
+
"loss": 1.5704,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.23276791842337485,
|
8347 |
+
"grad_norm": 25.36232566833496,
|
8348 |
+
"learning_rate": 1.0403177411102438e-05,
|
8349 |
+
"loss": 2.207,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.23296401608000786,
|
8354 |
+
"grad_norm": 8.098196983337402,
|
8355 |
+
"learning_rate": 1.0302232459578454e-05,
|
8356 |
+
"loss": 2.2944,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.23316011373664083,
|
8361 |
+
"grad_norm": 6.22314977645874,
|
8362 |
+
"learning_rate": 1.0201753043399143e-05,
|
8363 |
+
"loss": 1.62,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.23335621139327384,
|
8368 |
+
"grad_norm": 9.491933822631836,
|
8369 |
+
"learning_rate": 1.0101739684057098e-05,
|
8370 |
+
"loss": 1.392,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.23355230904990684,
|
8375 |
+
"grad_norm": 6.178860187530518,
|
8376 |
+
"learning_rate": 1.0002192900626028e-05,
|
8377 |
+
"loss": 1.6095,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.23374840670653985,
|
8382 |
+
"grad_norm": 5.492360591888428,
|
8383 |
+
"learning_rate": 9.903113209758096e-06,
|
8384 |
+
"loss": 2.4164,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.23394450436317285,
|
8389 |
+
"grad_norm": 7.380622386932373,
|
8390 |
+
"learning_rate": 9.804501125681243e-06,
|
8391 |
+
"loss": 1.8485,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.23414060201980585,
|
8396 |
+
"grad_norm": 6.217007637023926,
|
8397 |
+
"learning_rate": 9.70635716019651e-06,
|
8398 |
+
"loss": 1.5716,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.23433669967643886,
|
8403 |
+
"grad_norm": 3.790039539337158,
|
8404 |
+
"learning_rate": 9.608681822675381e-06,
|
8405 |
+
"loss": 1.5046,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.23453279733307186,
|
8410 |
+
"grad_norm": 4.153253555297852,
|
8411 |
+
"learning_rate": 9.51147562005713e-06,
|
8412 |
+
"loss": 0.7896,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.23472889498970487,
|
8417 |
+
"grad_norm": 5.173237323760986,
|
8418 |
+
"learning_rate": 9.414739056846222e-06,
|
8419 |
+
"loss": 1.1483,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.23492499264633787,
|
8424 |
+
"grad_norm": 6.67634916305542,
|
8425 |
+
"learning_rate": 9.318472635109653e-06,
|
8426 |
+
"loss": 1.0958,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.23512109030297088,
|
8431 |
+
"grad_norm": 5.527042388916016,
|
8432 |
+
"learning_rate": 9.222676854474365e-06,
|
8433 |
+
"loss": 2.349,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.23531718795960388,
|
8438 |
+
"grad_norm": 6.437062740325928,
|
8439 |
+
"learning_rate": 9.127352212124662e-06,
|
8440 |
+
"loss": 2.2577,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.23551328561623688,
|
8445 |
+
"grad_norm": 6.333834171295166,
|
8446 |
+
"learning_rate": 9.032499202799628e-06,
|
8447 |
+
"loss": 1.5634,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.2357093832728699,
|
8452 |
+
"grad_norm": 8.251330375671387,
|
8453 |
+
"learning_rate": 8.938118318790522e-06,
|
8454 |
+
"loss": 1.6758,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.2359054809295029,
|
8459 |
+
"grad_norm": 11.161762237548828,
|
8460 |
+
"learning_rate": 8.844210049938262e-06,
|
8461 |
+
"loss": 0.9043,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.2361015785861359,
|
8466 |
+
"grad_norm": 5.800900936126709,
|
8467 |
+
"learning_rate": 8.750774883630908e-06,
|
8468 |
+
"loss": 2.1929,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.2362976762427689,
|
8473 |
+
"grad_norm": 4.731690883636475,
|
8474 |
+
"learning_rate": 8.657813304801043e-06,
|
8475 |
+
"loss": 0.5826,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.2364937738994019,
|
8480 |
+
"grad_norm": 4.208858966827393,
|
8481 |
+
"learning_rate": 8.565325795923341e-06,
|
8482 |
+
"loss": 0.9173,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.2366898715560349,
|
8487 |
+
"grad_norm": 4.98720645904541,
|
8488 |
+
"learning_rate": 8.473312837012026e-06,
|
8489 |
+
"loss": 1.7575,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.2368859692126679,
|
8494 |
+
"grad_norm": 6.606943130493164,
|
8495 |
+
"learning_rate": 8.3817749056184e-06,
|
8496 |
+
"loss": 1.6911,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.23708206686930092,
|
8501 |
+
"grad_norm": 6.240574359893799,
|
8502 |
+
"learning_rate": 8.290712476828332e-06,
|
8503 |
+
"loss": 2.235,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.23727816452593392,
|
8508 |
+
"grad_norm": 7.585755348205566,
|
8509 |
+
"learning_rate": 8.200126023259791e-06,
|
8510 |
+
"loss": 1.4902,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.23747426218256693,
|
8515 |
+
"grad_norm": 7.400862693786621,
|
8516 |
+
"learning_rate": 8.110016015060484e-06,
|
8517 |
+
"loss": 1.9178,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.23767035983919993,
|
8522 |
+
"grad_norm": 9.042954444885254,
|
8523 |
+
"learning_rate": 8.020382919905278e-06,
|
8524 |
+
"loss": 2.6052,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.23786645749583293,
|
8529 |
+
"grad_norm": 15.022893905639648,
|
8530 |
+
"learning_rate": 7.931227202993873e-06,
|
8531 |
+
"loss": 1.697,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.23806255515246594,
|
8536 |
+
"grad_norm": 11.485679626464844,
|
8537 |
+
"learning_rate": 7.842549327048365e-06,
|
8538 |
+
"loss": 2.7337,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.23825865280909894,
|
8543 |
+
"grad_norm": 4.79671049118042,
|
8544 |
+
"learning_rate": 7.754349752310752e-06,
|
8545 |
+
"loss": 1.7505,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.23845475046573195,
|
8550 |
+
"grad_norm": 4.849589824676514,
|
8551 |
+
"learning_rate": 7.666628936540776e-06,
|
8552 |
+
"loss": 0.7821,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.23865084812236495,
|
8557 |
+
"grad_norm": 4.220393180847168,
|
8558 |
+
"learning_rate": 7.579387335013255e-06,
|
8559 |
+
"loss": 0.7812,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.23884694577899795,
|
8564 |
+
"grad_norm": 12.55904769897461,
|
8565 |
+
"learning_rate": 7.492625400515951e-06,
|
8566 |
+
"loss": 1.8666,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.23904304343563093,
|
8571 |
+
"grad_norm": 5.080347537994385,
|
8572 |
+
"learning_rate": 7.406343583347119e-06,
|
8573 |
+
"loss": 0.7878,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.23923914109226393,
|
8578 |
+
"grad_norm": 4.184628486633301,
|
8579 |
+
"learning_rate": 7.320542331313118e-06,
|
8580 |
+
"loss": 1.193,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.23943523874889694,
|
8585 |
+
"grad_norm": 4.931305885314941,
|
8586 |
+
"learning_rate": 7.235222089726279e-06,
|
8587 |
+
"loss": 0.9402,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.23963133640552994,
|
8592 |
+
"grad_norm": 8.1219482421875,
|
8593 |
+
"learning_rate": 7.15038330140233e-06,
|
8594 |
+
"loss": 2.1749,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.23982743406216295,
|
8599 |
+
"grad_norm": 5.320800304412842,
|
8600 |
+
"learning_rate": 7.066026406658355e-06,
|
8601 |
+
"loss": 1.0909,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.24002353171879595,
|
8606 |
+
"grad_norm": 5.9109907150268555,
|
8607 |
+
"learning_rate": 6.982151843310281e-06,
|
8608 |
+
"loss": 2.0466,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.24021962937542896,
|
8613 |
+
"grad_norm": 4.978448390960693,
|
8614 |
+
"learning_rate": 6.898760046670815e-06,
|
8615 |
+
"loss": 2.8843,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.24041572703206196,
|
8620 |
+
"grad_norm": 6.749833106994629,
|
8621 |
+
"learning_rate": 6.815851449547028e-06,
|
8622 |
+
"loss": 0.8198,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.24061182468869496,
|
8627 |
+
"grad_norm": 3.7781665325164795,
|
8628 |
+
"learning_rate": 6.7334264822381254e-06,
|
8629 |
+
"loss": 0.9111,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.24080792234532797,
|
8634 |
+
"grad_norm": 6.719789028167725,
|
8635 |
+
"learning_rate": 6.651485572533378e-06,
|
8636 |
+
"loss": 1.9275,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.24100402000196097,
|
8641 |
+
"grad_norm": 7.115839004516602,
|
8642 |
+
"learning_rate": 6.570029145709622e-06,
|
8643 |
+
"loss": 1.2663,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.24120011765859398,
|
8648 |
+
"grad_norm": 7.567049503326416,
|
8649 |
+
"learning_rate": 6.489057624529349e-06,
|
8650 |
+
"loss": 2.1002,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.24139621531522698,
|
8655 |
+
"grad_norm": 7.6808180809021,
|
8656 |
+
"learning_rate": 6.408571429238253e-06,
|
8657 |
+
"loss": 1.0844,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.24159231297185998,
|
8662 |
+
"grad_norm": 15.046584129333496,
|
8663 |
+
"learning_rate": 6.328570977563208e-06,
|
8664 |
+
"loss": 2.3513,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.241788410628493,
|
8669 |
+
"grad_norm": 9.350245475769043,
|
8670 |
+
"learning_rate": 6.24905668471013e-06,
|
8671 |
+
"loss": 1.81,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.241984508285126,
|
8676 |
+
"grad_norm": 5.964272499084473,
|
8677 |
+
"learning_rate": 6.170028963361618e-06,
|
8678 |
+
"loss": 2.0853,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.242180605941759,
|
8683 |
+
"grad_norm": 5.296453475952148,
|
8684 |
+
"learning_rate": 6.091488223675057e-06,
|
8685 |
+
"loss": 0.8561,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.242376703598392,
|
8690 |
+
"grad_norm": 20.43738555908203,
|
8691 |
+
"learning_rate": 6.013434873280288e-06,
|
8692 |
+
"loss": 3.0297,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.242572801255025,
|
8697 |
+
"grad_norm": 5.732899188995361,
|
8698 |
+
"learning_rate": 5.935869317277643e-06,
|
8699 |
+
"loss": 1.4933,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.242768898911658,
|
8704 |
+
"grad_norm": 3.8854384422302246,
|
8705 |
+
"learning_rate": 5.858791958235754e-06,
|
8706 |
+
"loss": 0.8324,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.24296499656829101,
|
8711 |
+
"grad_norm": 6.640622615814209,
|
8712 |
+
"learning_rate": 5.782203196189461e-06,
|
8713 |
+
"loss": 1.5419,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.24316109422492402,
|
8718 |
+
"grad_norm": 8.89739990234375,
|
8719 |
+
"learning_rate": 5.706103428637865e-06,
|
8720 |
+
"loss": 0.7287,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.24335719188155702,
|
8725 |
+
"grad_norm": 3.595503568649292,
|
8726 |
+
"learning_rate": 5.630493050542041e-06,
|
8727 |
+
"loss": 0.8245,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.24355328953819003,
|
8732 |
+
"grad_norm": 8.166557312011719,
|
8733 |
+
"learning_rate": 5.5553724543231825e-06,
|
8734 |
+
"loss": 1.5169,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.24374938719482303,
|
8739 |
+
"grad_norm": 11.863232612609863,
|
8740 |
+
"learning_rate": 5.480742029860464e-06,
|
8741 |
+
"loss": 2.4468,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.24394548485145603,
|
8746 |
+
"grad_norm": 4.747961044311523,
|
8747 |
+
"learning_rate": 5.406602164489072e-06,
|
8748 |
+
"loss": 1.1186,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.24414158250808904,
|
8753 |
+
"grad_norm": 3.378335952758789,
|
8754 |
+
"learning_rate": 5.332953242998151e-06,
|
8755 |
+
"loss": 1.222,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.24433768016472204,
|
8760 |
+
"grad_norm": 8.078283309936523,
|
8761 |
+
"learning_rate": 5.259795647628818e-06,
|
8762 |
+
"loss": 2.1882,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.24453377782135505,
|
8767 |
+
"grad_norm": 6.306114196777344,
|
8768 |
+
"learning_rate": 5.1871297580722515e-06,
|
8769 |
+
"loss": 1.9767,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.24472987547798805,
|
8774 |
+
"grad_norm": 4.042963981628418,
|
8775 |
+
"learning_rate": 5.114955951467537e-06,
|
8776 |
+
"loss": 1.6329,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.24492597313462103,
|
8781 |
+
"grad_norm": 4.9465179443359375,
|
8782 |
+
"learning_rate": 5.043274602399939e-06,
|
8783 |
+
"loss": 1.4165,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.24512207079125403,
|
8788 |
+
"grad_norm": 7.358532428741455,
|
8789 |
+
"learning_rate": 4.972086082898775e-06,
|
8790 |
+
"loss": 1.5899,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.24531816844788704,
|
8795 |
+
"grad_norm": 4.857394218444824,
|
8796 |
+
"learning_rate": 4.901390762435587e-06,
|
8797 |
+
"loss": 0.5367,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.24551426610452004,
|
8802 |
+
"grad_norm": 7.886226177215576,
|
8803 |
+
"learning_rate": 4.831189007922199e-06,
|
8804 |
+
"loss": 2.0356,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.24571036376115304,
|
8809 |
+
"grad_norm": 6.1324872970581055,
|
8810 |
+
"learning_rate": 4.761481183708783e-06,
|
8811 |
+
"loss": 1.1557,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.24590646141778605,
|
8816 |
+
"grad_norm": 4.222672939300537,
|
8817 |
+
"learning_rate": 4.692267651581994e-06,
|
8818 |
+
"loss": 1.5599,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.24610255907441905,
|
8823 |
+
"grad_norm": 6.940048694610596,
|
8824 |
+
"learning_rate": 4.6235487707631085e-06,
|
8825 |
+
"loss": 1.9772,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.24629865673105206,
|
8830 |
+
"grad_norm": 7.745510578155518,
|
8831 |
+
"learning_rate": 4.555324897906132e-06,
|
8832 |
+
"loss": 1.7293,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.24649475438768506,
|
8837 |
+
"grad_norm": 4.879486560821533,
|
8838 |
+
"learning_rate": 4.48759638709596e-06,
|
8839 |
+
"loss": 0.5805,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.24669085204431807,
|
8844 |
+
"grad_norm": 8.680879592895508,
|
8845 |
+
"learning_rate": 4.42036358984651e-06,
|
8846 |
+
"loss": 1.9445,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.24688694970095107,
|
8851 |
+
"grad_norm": 9.775550842285156,
|
8852 |
+
"learning_rate": 4.353626855098958e-06,
|
8853 |
+
"loss": 1.8064,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.24708304735758407,
|
8858 |
+
"grad_norm": 4.5769805908203125,
|
8859 |
+
"learning_rate": 4.287386529219894e-06,
|
8860 |
+
"loss": 1.3112,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.24727914501421708,
|
8865 |
+
"grad_norm": 9.00042724609375,
|
8866 |
+
"learning_rate": 4.221642955999494e-06,
|
8867 |
+
"loss": 1.9532,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.24747524267085008,
|
8872 |
+
"grad_norm": 5.064240455627441,
|
8873 |
+
"learning_rate": 4.156396476649782e-06,
|
8874 |
+
"loss": 1.262,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.24767134032748309,
|
8879 |
+
"grad_norm": 5.231726169586182,
|
8880 |
+
"learning_rate": 4.091647429802869e-06,
|
8881 |
+
"loss": 0.9197,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.2478674379841161,
|
8886 |
+
"grad_norm": 4.577012062072754,
|
8887 |
+
"learning_rate": 4.027396151509133e-06,
|
8888 |
+
"loss": 0.9262,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.2480635356407491,
|
8893 |
+
"grad_norm": 11.125844955444336,
|
8894 |
+
"learning_rate": 3.963642975235515e-06,
|
8895 |
+
"loss": 0.8612,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.2482596332973821,
|
8900 |
+
"grad_norm": 6.266839981079102,
|
8901 |
+
"learning_rate": 3.900388231863805e-06,
|
8902 |
+
"loss": 1.8149,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.2484557309540151,
|
8907 |
+
"grad_norm": 11.235099792480469,
|
8908 |
+
"learning_rate": 3.8376322496888825e-06,
|
8909 |
+
"loss": 1.9527,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.2486518286106481,
|
8914 |
+
"grad_norm": 5.78483247756958,
|
8915 |
+
"learning_rate": 3.7753753544170655e-06,
|
8916 |
+
"loss": 1.5316,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.2488479262672811,
|
8921 |
+
"grad_norm": 7.571983814239502,
|
8922 |
+
"learning_rate": 3.7136178691643433e-06,
|
8923 |
+
"loss": 1.8357,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.24904402392391412,
|
8928 |
+
"grad_norm": 10.283147811889648,
|
8929 |
+
"learning_rate": 3.6523601144548003e-06,
|
8930 |
+
"loss": 1.9926,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.24924012158054712,
|
8935 |
+
"grad_norm": 4.631470203399658,
|
8936 |
+
"learning_rate": 3.5916024082188414e-06,
|
8937 |
+
"loss": 0.7877,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.24943621923718012,
|
8942 |
+
"grad_norm": 5.498098373413086,
|
8943 |
+
"learning_rate": 3.531345065791636e-06,
|
8944 |
+
"loss": 3.0685,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.24963231689381313,
|
8949 |
+
"grad_norm": 5.340700149536133,
|
8950 |
+
"learning_rate": 3.471588399911441e-06,
|
8951 |
+
"loss": 1.7174,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.24982841455044613,
|
8956 |
+
"grad_norm": 8.994786262512207,
|
8957 |
+
"learning_rate": 3.4123327207179477e-06,
|
8958 |
+
"loss": 1.722,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.2500245122070791,
|
8963 |
+
"grad_norm": 4.499628067016602,
|
8964 |
+
"learning_rate": 3.3535783357507624e-06,
|
8965 |
+
"loss": 1.3553,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.25022060986371214,
|
8970 |
+
"grad_norm": 13.06672477722168,
|
8971 |
+
"learning_rate": 3.29532554994767e-06,
|
8972 |
+
"loss": 2.8721,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.2504167075203451,
|
8977 |
+
"grad_norm": 4.79543399810791,
|
8978 |
+
"learning_rate": 3.2375746656432284e-06,
|
8979 |
+
"loss": 1.0701,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.25061280517697815,
|
8984 |
+
"grad_norm": 5.686835765838623,
|
8985 |
+
"learning_rate": 3.180325982567034e-06,
|
8986 |
+
"loss": 1.2648,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.2508089028336111,
|
8991 |
+
"grad_norm": 4.657541275024414,
|
8992 |
+
"learning_rate": 3.1235797978422687e-06,
|
8993 |
+
"loss": 1.3918,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.25100500049024416,
|
8998 |
+
"grad_norm": 10.92974853515625,
|
8999 |
+
"learning_rate": 3.0673364059841338e-06,
|
9000 |
+
"loss": 2.4811,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.25120109814687713,
|
9005 |
+
"grad_norm": 5.082363128662109,
|
9006 |
+
"learning_rate": 3.0115960988982506e-06,
|
9007 |
+
"loss": 2.2096,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.25139719580351017,
|
9012 |
+
"grad_norm": 3.3826422691345215,
|
9013 |
+
"learning_rate": 2.9563591658793076e-06,
|
9014 |
+
"loss": 0.8372,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.25159329346014314,
|
9019 |
+
"grad_norm": 5.3609113693237305,
|
9020 |
+
"learning_rate": 2.901625893609361e-06,
|
9021 |
+
"loss": 1.2737,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.2517893911167762,
|
9026 |
+
"grad_norm": 4.7862043380737305,
|
9027 |
+
"learning_rate": 2.8473965661565347e-06,
|
9028 |
+
"loss": 0.7095,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.25198548877340915,
|
9033 |
+
"grad_norm": 8.444502830505371,
|
9034 |
+
"learning_rate": 2.793671464973413e-06,
|
9035 |
+
"loss": 2.1887,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.2521815864300422,
|
9040 |
+
"grad_norm": 5.742269039154053,
|
9041 |
+
"learning_rate": 2.740450868895583e-06,
|
9042 |
+
"loss": 1.1412,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.25237768408667516,
|
9047 |
+
"grad_norm": 7.201744079589844,
|
9048 |
+
"learning_rate": 2.687735054140317e-06,
|
9049 |
+
"loss": 1.3327,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.2525737817433082,
|
9054 |
+
"grad_norm": 7.618087291717529,
|
9055 |
+
"learning_rate": 2.63552429430497e-06,
|
9056 |
+
"loss": 1.9268,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.25276987939994117,
|
9061 |
+
"grad_norm": 7.360147953033447,
|
9062 |
+
"learning_rate": 2.5838188603657056e-06,
|
9063 |
+
"loss": 1.0334,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.2529659770565742,
|
9068 |
+
"grad_norm": 5.17711067199707,
|
9069 |
+
"learning_rate": 2.5326190206759527e-06,
|
9070 |
+
"loss": 1.784,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.2531620747132072,
|
9075 |
+
"grad_norm": 6.017335891723633,
|
9076 |
+
"learning_rate": 2.4819250409651607e-06,
|
9077 |
+
"loss": 0.9696,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.2533581723698402,
|
9082 |
+
"grad_norm": 6.2475433349609375,
|
9083 |
+
"learning_rate": 2.4317371843372904e-06,
|
9084 |
+
"loss": 2.4048,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.2535542700264732,
|
9089 |
+
"grad_norm": 6.189507484436035,
|
9090 |
+
"learning_rate": 2.3820557112695153e-06,
|
9091 |
+
"loss": 1.3096,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.25375036768310616,
|
9096 |
+
"grad_norm": 7.4594011306762695,
|
9097 |
+
"learning_rate": 2.3328808796108657e-06,
|
9098 |
+
"loss": 0.8887,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.2539464653397392,
|
9103 |
+
"grad_norm": 4.671147346496582,
|
9104 |
+
"learning_rate": 2.2842129445808546e-06,
|
9105 |
+
"loss": 1.3748,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.25414256299637217,
|
9110 |
+
"grad_norm": 4.567478179931641,
|
9111 |
+
"learning_rate": 2.2360521587682313e-06,
|
9112 |
+
"loss": 0.8291,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.2543386606530052,
|
9117 |
+
"grad_norm": 7.609241008758545,
|
9118 |
+
"learning_rate": 2.188398772129552e-06,
|
9119 |
+
"loss": 2.9765,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.2545347583096382,
|
9124 |
+
"grad_norm": 9.003718376159668,
|
9125 |
+
"learning_rate": 2.1412530319879887e-06,
|
9126 |
+
"loss": 3.6435,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.2547308559662712,
|
9131 |
+
"grad_norm": 6.3176164627075195,
|
9132 |
+
"learning_rate": 2.0946151830320224e-06,
|
9133 |
+
"loss": 1.2652,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.2549269536229042,
|
9138 |
+
"grad_norm": 12.08651065826416,
|
9139 |
+
"learning_rate": 2.0484854673140983e-06,
|
9140 |
+
"loss": 2.2016,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.2551230512795372,
|
9145 |
+
"grad_norm": 3.733414649963379,
|
9146 |
+
"learning_rate": 2.002864124249504e-06,
|
9147 |
+
"loss": 2.3059,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.2553191489361702,
|
9152 |
+
"grad_norm": 8.422008514404297,
|
9153 |
+
"learning_rate": 1.9577513906149702e-06,
|
9154 |
+
"loss": 1.5297,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.2555152465928032,
|
9159 |
+
"grad_norm": 7.98868989944458,
|
9160 |
+
"learning_rate": 1.913147500547574e-06,
|
9161 |
+
"loss": 2.1541,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.2557113442494362,
|
9166 |
+
"grad_norm": 4.629753589630127,
|
9167 |
+
"learning_rate": 1.869052685543471e-06,
|
9168 |
+
"loss": 1.617,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.25590744190606923,
|
9173 |
+
"grad_norm": 4.7464823722839355,
|
9174 |
+
"learning_rate": 1.825467174456652e-06,
|
9175 |
+
"loss": 0.7396,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.2561035395627022,
|
9180 |
+
"grad_norm": 6.383626937866211,
|
9181 |
+
"learning_rate": 1.7823911934978898e-06,
|
9182 |
+
"loss": 2.6406,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.25629963721933524,
|
9187 |
+
"grad_norm": 3.751051902770996,
|
9188 |
+
"learning_rate": 1.73982496623335e-06,
|
9189 |
+
"loss": 2.1239,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.2564957348759682,
|
9194 |
+
"grad_norm": 8.627437591552734,
|
9195 |
+
"learning_rate": 1.6977687135836584e-06,
|
9196 |
+
"loss": 1.5667,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.25669183253260125,
|
9201 |
+
"grad_norm": 5.191834926605225,
|
9202 |
+
"learning_rate": 1.656222653822581e-06,
|
9203 |
+
"loss": 1.7599,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.2568879301892342,
|
9208 |
+
"grad_norm": 11.265420913696289,
|
9209 |
+
"learning_rate": 1.615187002576013e-06,
|
9210 |
+
"loss": 1.6896,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.25708402784586726,
|
9215 |
+
"grad_norm": 7.610350608825684,
|
9216 |
+
"learning_rate": 1.574661972820779e-06,
|
9217 |
+
"loss": 1.0143,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.25728012550250023,
|
9222 |
+
"grad_norm": 2.748887538909912,
|
9223 |
+
"learning_rate": 1.5346477748835354e-06,
|
9224 |
+
"loss": 0.3839,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.25747622315913327,
|
9229 |
+
"grad_norm": 9.816802024841309,
|
9230 |
+
"learning_rate": 1.4951446164397587e-06,
|
9231 |
+
"loss": 1.659,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.25767232081576624,
|
9236 |
+
"grad_norm": 11.259073257446289,
|
9237 |
+
"learning_rate": 1.4561527025125476e-06,
|
9238 |
+
"loss": 1.9394,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.2578684184723993,
|
9243 |
+
"grad_norm": 5.475277423858643,
|
9244 |
+
"learning_rate": 1.4176722354716455e-06,
|
9245 |
+
"loss": 1.1844,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.25806451612903225,
|
9250 |
+
"grad_norm": 3.6779868602752686,
|
9251 |
+
"learning_rate": 1.379703415032374e-06,
|
9252 |
+
"loss": 1.5793,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.2582606137856653,
|
9257 |
+
"grad_norm": 7.077935695648193,
|
9258 |
+
"learning_rate": 1.3422464382545797e-06,
|
9259 |
+
"loss": 2.019,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.25845671144229826,
|
9264 |
+
"grad_norm": 3.54400634765625,
|
9265 |
+
"learning_rate": 1.3053014995415891e-06,
|
9266 |
+
"loss": 1.0317,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.2586528090989313,
|
9271 |
+
"grad_norm": 8.784947395324707,
|
9272 |
+
"learning_rate": 1.268868790639277e-06,
|
9273 |
+
"loss": 0.6858,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.25884890675556427,
|
9278 |
+
"grad_norm": 3.767857789993286,
|
9279 |
+
"learning_rate": 1.2329485006349895e-06,
|
9280 |
+
"loss": 1.6571,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.2590450044121973,
|
9285 |
+
"grad_norm": 2.997652769088745,
|
9286 |
+
"learning_rate": 1.1975408159566103e-06,
|
9287 |
+
"loss": 0.7133,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.2592411020688303,
|
9292 |
+
"grad_norm": 13.322317123413086,
|
9293 |
+
"learning_rate": 1.1626459203715633e-06,
|
9294 |
+
"loss": 2.6333,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.2594371997254633,
|
9299 |
+
"grad_norm": 7.647657871246338,
|
9300 |
+
"learning_rate": 1.128263994985901e-06,
|
9301 |
+
"loss": 1.3197,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.2596332973820963,
|
9306 |
+
"grad_norm": 5.545785903930664,
|
9307 |
+
"learning_rate": 1.0943952182433048e-06,
|
9308 |
+
"loss": 1.8365,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.25982939503872926,
|
9313 |
+
"grad_norm": 6.008249759674072,
|
9314 |
+
"learning_rate": 1.0610397659242322e-06,
|
9315 |
+
"loss": 0.7947,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.2600254926953623,
|
9320 |
+
"grad_norm": 6.118869304656982,
|
9321 |
+
"learning_rate": 1.0281978111449375e-06,
|
9322 |
+
"loss": 2.1696,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.26022159035199527,
|
9327 |
+
"grad_norm": 5.802155017852783,
|
9328 |
+
"learning_rate": 9.958695243565853e-07,
|
9329 |
+
"loss": 1.3875,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.2604176880086283,
|
9334 |
+
"grad_norm": 5.157081604003906,
|
9335 |
+
"learning_rate": 9.640550733444275e-07,
|
9336 |
+
"loss": 1.7365,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.2606137856652613,
|
9341 |
+
"grad_norm": 4.770791053771973,
|
9342 |
+
"learning_rate": 9.327546232268392e-07,
|
9343 |
+
"loss": 1.0385,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.2608098833218943,
|
9348 |
+
"grad_norm": 6.321690559387207,
|
9349 |
+
"learning_rate": 9.019683364545395e-07,
|
9350 |
+
"loss": 1.4739,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.2610059809785273,
|
9355 |
+
"grad_norm": 10.451473236083984,
|
9356 |
+
"learning_rate": 8.71696372809705e-07,
|
9357 |
+
"loss": 2.4985,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.2612020786351603,
|
9362 |
+
"grad_norm": 4.130318641662598,
|
9363 |
+
"learning_rate": 8.419388894051472e-07,
|
9364 |
+
"loss": 0.9982,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.2613981762917933,
|
9369 |
+
"grad_norm": 6.775696277618408,
|
9370 |
+
"learning_rate": 8.126960406835249e-07,
|
9371 |
+
"loss": 1.3855,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.2615942739484263,
|
9376 |
+
"grad_norm": 4.182593822479248,
|
9377 |
+
"learning_rate": 7.839679784164778e-07,
|
9378 |
+
"loss": 1.0039,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.2617903716050593,
|
9383 |
+
"grad_norm": 8.561737060546875,
|
9384 |
+
"learning_rate": 7.557548517039381e-07,
|
9385 |
+
"loss": 2.7707,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.26198646926169233,
|
9390 |
+
"grad_norm": 7.187052249908447,
|
9391 |
+
"learning_rate": 7.28056806973243e-07,
|
9392 |
+
"loss": 2.3727,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.2621825669183253,
|
9397 |
+
"grad_norm": 6.421212673187256,
|
9398 |
+
"learning_rate": 7.008739879784787e-07,
|
9399 |
+
"loss": 1.4712,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.26237866457495834,
|
9404 |
+
"grad_norm": 5.969448089599609,
|
9405 |
+
"learning_rate": 6.742065357996486e-07,
|
9406 |
+
"loss": 2.0104,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.2625747622315913,
|
9411 |
+
"grad_norm": 3.251781463623047,
|
9412 |
+
"learning_rate": 6.480545888420176e-07,
|
9413 |
+
"loss": 0.7331,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.26277085988822435,
|
9418 |
+
"grad_norm": 8.010222434997559,
|
9419 |
+
"learning_rate": 6.224182828353242e-07,
|
9420 |
+
"loss": 2.0611,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.2629669575448573,
|
9425 |
+
"grad_norm": 14.670138359069824,
|
9426 |
+
"learning_rate": 5.972977508331368e-07,
|
9427 |
+
"loss": 1.6784,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.26316305520149036,
|
9432 |
+
"grad_norm": 3.959542751312256,
|
9433 |
+
"learning_rate": 5.726931232120869e-07,
|
9434 |
+
"loss": 1.1185,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.26335915285812334,
|
9439 |
+
"grad_norm": 3.680546998977661,
|
9440 |
+
"learning_rate": 5.486045276712926e-07,
|
9441 |
+
"loss": 1.0173,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.26355525051475637,
|
9446 |
+
"grad_norm": 8.655952453613281,
|
9447 |
+
"learning_rate": 5.250320892316252e-07,
|
9448 |
+
"loss": 1.4111,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.26375134817138934,
|
9453 |
+
"grad_norm": 5.929543495178223,
|
9454 |
+
"learning_rate": 5.019759302350547e-07,
|
9455 |
+
"loss": 1.7795,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.2639474458280224,
|
9460 |
+
"grad_norm": 6.462841987609863,
|
9461 |
+
"learning_rate": 4.794361703440719e-07,
|
9462 |
+
"loss": 0.7667,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.26414354348465535,
|
9467 |
+
"grad_norm": 11.463775634765625,
|
9468 |
+
"learning_rate": 4.57412926541001e-07,
|
9469 |
+
"loss": 1.8102,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.2643396411412884,
|
9474 |
+
"grad_norm": 8.908552169799805,
|
9475 |
+
"learning_rate": 4.3590631312746545e-07,
|
9476 |
+
"loss": 1.5951,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.26453573879792136,
|
9481 |
+
"grad_norm": 5.707728862762451,
|
9482 |
+
"learning_rate": 4.149164417237117e-07,
|
9483 |
+
"loss": 1.287,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.2647318364545544,
|
9488 |
+
"grad_norm": 18.28376579284668,
|
9489 |
+
"learning_rate": 3.944434212680981e-07,
|
9490 |
+
"loss": 3.0229,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.26492793411118737,
|
9495 |
+
"grad_norm": 7.537192344665527,
|
9496 |
+
"learning_rate": 3.744873580165176e-07,
|
9497 |
+
"loss": 1.6773,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.2651240317678204,
|
9502 |
+
"grad_norm": 5.802631378173828,
|
9503 |
+
"learning_rate": 3.5504835554177605e-07,
|
9504 |
+
"loss": 1.5217,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.2653201294244534,
|
9509 |
+
"grad_norm": 5.430902004241943,
|
9510 |
+
"learning_rate": 3.361265147331816e-07,
|
9511 |
+
"loss": 1.668,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.26551622708108635,
|
9516 |
+
"grad_norm": 8.240734100341797,
|
9517 |
+
"learning_rate": 3.177219337958892e-07,
|
9518 |
+
"loss": 1.1068,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.2657123247377194,
|
9523 |
+
"grad_norm": 10.376667976379395,
|
9524 |
+
"learning_rate": 2.998347082505126e-07,
|
9525 |
+
"loss": 2.2403,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.26590842239435236,
|
9530 |
+
"grad_norm": 6.063961505889893,
|
9531 |
+
"learning_rate": 2.8246493093250227e-07,
|
9532 |
+
"loss": 1.3754,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.2661045200509854,
|
9537 |
+
"grad_norm": 5.9845733642578125,
|
9538 |
+
"learning_rate": 2.6561269199179006e-07,
|
9539 |
+
"loss": 1.5861,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.26630061770761837,
|
9544 |
+
"grad_norm": 6.585875511169434,
|
9545 |
+
"learning_rate": 2.492780788922344e-07,
|
9546 |
+
"loss": 2.882,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.2664967153642514,
|
9551 |
+
"grad_norm": 5.5460968017578125,
|
9552 |
+
"learning_rate": 2.3346117641116494e-07,
|
9553 |
+
"loss": 1.7507,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.2666928130208844,
|
9558 |
+
"grad_norm": 6.179254531860352,
|
9559 |
+
"learning_rate": 2.1816206663902717e-07,
|
9560 |
+
"loss": 0.9413,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.2668889106775174,
|
9565 |
+
"grad_norm": 18.120824813842773,
|
9566 |
+
"learning_rate": 2.0338082897886079e-07,
|
9567 |
+
"loss": 2.7844,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.2670850083341504,
|
9572 |
+
"grad_norm": 4.4856157302856445,
|
9573 |
+
"learning_rate": 1.891175401459444e-07,
|
9574 |
+
"loss": 0.7925,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.2672811059907834,
|
9579 |
+
"grad_norm": 5.911020755767822,
|
9580 |
+
"learning_rate": 1.7537227416735135e-07,
|
9581 |
+
"loss": 1.7331,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.2674772036474164,
|
9586 |
+
"grad_norm": 4.898580551147461,
|
9587 |
+
"learning_rate": 1.6214510238163893e-07,
|
9588 |
+
"loss": 1.5288,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.2676733013040494,
|
9593 |
+
"grad_norm": 4.032687664031982,
|
9594 |
+
"learning_rate": 1.4943609343839316e-07,
|
9595 |
+
"loss": 1.6682,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.2678693989606824,
|
9600 |
+
"grad_norm": 4.88344144821167,
|
9601 |
+
"learning_rate": 1.37245313297929e-07,
|
9602 |
+
"loss": 0.8977,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.26806549661731544,
|
9607 |
+
"grad_norm": 5.699201583862305,
|
9608 |
+
"learning_rate": 1.2557282523094627e-07,
|
9609 |
+
"loss": 1.4562,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.2682615942739484,
|
9614 |
+
"grad_norm": 6.58184814453125,
|
9615 |
+
"learning_rate": 1.1441868981815207e-07,
|
9616 |
+
"loss": 0.9233,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.26845769193058144,
|
9621 |
+
"grad_norm": 7.048121929168701,
|
9622 |
+
"learning_rate": 1.0378296494999439e-07,
|
9623 |
+
"loss": 2.6469,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.2686537895872144,
|
9628 |
+
"grad_norm": 6.19679069519043,
|
9629 |
+
"learning_rate": 9.366570582637346e-08,
|
9630 |
+
"loss": 1.6399,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.26884988724384745,
|
9635 |
+
"grad_norm": 6.541812896728516,
|
9636 |
+
"learning_rate": 8.406696495627531e-08,
|
9637 |
+
"loss": 2.3217,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.26904598490048043,
|
9642 |
+
"grad_norm": 7.404893398284912,
|
9643 |
+
"learning_rate": 7.498679215761639e-08,
|
9644 |
+
"loss": 1.32,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.26924208255711346,
|
9649 |
+
"grad_norm": 6.199706077575684,
|
9650 |
+
"learning_rate": 6.642523455687721e-08,
|
9651 |
+
"loss": 1.0584,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.26943818021374644,
|
9656 |
+
"grad_norm": 4.55602502822876,
|
9657 |
+
"learning_rate": 5.838233658892467e-08,
|
9658 |
+
"loss": 1.0957,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.26963427787037947,
|
9663 |
+
"grad_norm": 8.639842987060547,
|
9664 |
+
"learning_rate": 5.08581399967345e-08,
|
9665 |
+
"loss": 1.0236,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.26983037552701244,
|
9670 |
+
"grad_norm": 4.639615535736084,
|
9671 |
+
"learning_rate": 4.3852683831235866e-08,
|
9672 |
+
"loss": 0.8647,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.2700264731836455,
|
9677 |
+
"grad_norm": 4.805739402770996,
|
9678 |
+
"learning_rate": 3.736600445104488e-08,
|
9679 |
+
"loss": 1.9424,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.27022257084027845,
|
9684 |
+
"grad_norm": 4.504196643829346,
|
9685 |
+
"learning_rate": 3.139813552230919e-08,
|
9686 |
+
"loss": 0.815,
|
9687 |
+
"step": 1378
|
9688 |
+
},
|
9689 |
+
{
|
9690 |
+
"epoch": 0.2704186684969115,
|
9691 |
+
"grad_norm": 8.973424911499023,
|
9692 |
+
"learning_rate": 2.5949108018530342e-08,
|
9693 |
+
"loss": 1.7934,
|
9694 |
+
"step": 1379
|
9695 |
+
},
|
9696 |
+
{
|
9697 |
+
"epoch": 0.27061476615354446,
|
9698 |
+
"grad_norm": 5.073940277099609,
|
9699 |
+
"learning_rate": 2.101895022040834e-08,
|
9700 |
+
"loss": 1.6873,
|
9701 |
+
"step": 1380
|
9702 |
+
},
|
9703 |
+
{
|
9704 |
+
"epoch": 0.2708108638101775,
|
9705 |
+
"grad_norm": 5.69566535949707,
|
9706 |
+
"learning_rate": 1.6607687715675113e-08,
|
9707 |
+
"loss": 1.5254,
|
9708 |
+
"step": 1381
|
9709 |
+
},
|
9710 |
+
{
|
9711 |
+
"epoch": 0.27100696146681047,
|
9712 |
+
"grad_norm": 2.8405473232269287,
|
9713 |
+
"learning_rate": 1.2715343398972402e-08,
|
9714 |
+
"loss": 1.5803,
|
9715 |
+
"step": 1382
|
9716 |
+
},
|
9717 |
+
{
|
9718 |
+
"epoch": 0.2712030591234435,
|
9719 |
+
"grad_norm": 3.1316022872924805,
|
9720 |
+
"learning_rate": 9.341937471740724e-09,
|
9721 |
+
"loss": 0.5618,
|
9722 |
+
"step": 1383
|
9723 |
+
},
|
9724 |
+
{
|
9725 |
+
"epoch": 0.2713991567800765,
|
9726 |
+
"grad_norm": 6.8218793869018555,
|
9727 |
+
"learning_rate": 6.487487442097262e-09,
|
9728 |
+
"loss": 0.5681,
|
9729 |
+
"step": 1384
|
9730 |
+
},
|
9731 |
+
{
|
9732 |
+
"epoch": 0.27159525443670945,
|
9733 |
+
"grad_norm": 6.514251232147217,
|
9734 |
+
"learning_rate": 4.1520081247803375e-09,
|
9735 |
+
"loss": 2.2744,
|
9736 |
+
"step": 1385
|
9737 |
+
},
|
9738 |
+
{
|
9739 |
+
"epoch": 0.2717913520933425,
|
9740 |
+
"grad_norm": 8.452823638916016,
|
9741 |
+
"learning_rate": 2.335511641005095e-09,
|
9742 |
+
"loss": 1.4333,
|
9743 |
+
"step": 1386
|
9744 |
+
},
|
9745 |
+
{
|
9746 |
+
"epoch": 0.27198744974997546,
|
9747 |
+
"grad_norm": 5.929966449737549,
|
9748 |
+
"learning_rate": 1.0380074184856981e-09,
|
9749 |
+
"loss": 1.6609,
|
9750 |
+
"step": 1387
|
9751 |
+
},
|
9752 |
+
{
|
9753 |
+
"epoch": 0.2721835474066085,
|
9754 |
+
"grad_norm": 3.682739019393921,
|
9755 |
+
"learning_rate": 2.595021913243123e-10,
|
9756 |
+
"loss": 1.7354,
|
9757 |
+
"step": 1388
|
9758 |
+
},
|
9759 |
+
{
|
9760 |
+
"epoch": 0.27237964506324147,
|
9761 |
+
"grad_norm": 6.1388397216796875,
|
9762 |
+
"learning_rate": 0.0,
|
9763 |
+
"loss": 0.7588,
|
9764 |
+
"step": 1389
|
9765 |
}
|
9766 |
],
|
9767 |
"logging_steps": 1,
|
|
|
9776 |
"should_evaluate": false,
|
9777 |
"should_log": false,
|
9778 |
"should_save": true,
|
9779 |
+
"should_training_stop": true
|
9780 |
},
|
9781 |
"attributes": {}
|
9782 |
}
|
9783 |
},
|
9784 |
+
"total_flos": 2.976036715065508e+17,
|
9785 |
"train_batch_size": 2,
|
9786 |
"trial_name": null,
|
9787 |
"trial_params": null
|