Fanucci
commited on
Training in progress, step 800, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671149168
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:356d90ca6784d8ad896343b590031a4d760212691309e5fa44e94d9bd333476d
|
| 3 |
size 671149168
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1342555602
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07f70391f1d92cd7a39e5c59693cbf6577b5fec9bf7ca1b7d90b99c64d2fd956
|
| 3 |
size 1342555602
|
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 sha256:8b69db2d3e137bfc4fdab5a7f1d0c0af3e74fe5a5f426e92a44235c4a4bbad81
|
| 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 sha256:30632263e1c1835085a75879d50f43d028777df4547e51105124fa5b76c67ad2
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 3.1351470947265625,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4239,6 +4239,1414 @@
|
|
| 4239 |
"eval_samples_per_second": 5.632,
|
| 4240 |
"eval_steps_per_second": 1.408,
|
| 4241 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4242 |
}
|
| 4243 |
],
|
| 4244 |
"logging_steps": 1,
|
|
@@ -4253,7 +5661,7 @@
|
|
| 4253 |
"early_stopping_threshold": 0.0
|
| 4254 |
},
|
| 4255 |
"attributes": {
|
| 4256 |
-
"early_stopping_patience_counter":
|
| 4257 |
}
|
| 4258 |
},
|
| 4259 |
"TrainerControl": {
|
|
@@ -4267,7 +5675,7 @@
|
|
| 4267 |
"attributes": {}
|
| 4268 |
}
|
| 4269 |
},
|
| 4270 |
-
"total_flos":
|
| 4271 |
"train_batch_size": 4,
|
| 4272 |
"trial_name": null,
|
| 4273 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 3.1351470947265625,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 4 |
+
"epoch": 0.2559590465525516,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 800,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4239 |
"eval_samples_per_second": 5.632,
|
| 4240 |
"eval_steps_per_second": 1.408,
|
| 4241 |
"step": 600
|
| 4242 |
+
},
|
| 4243 |
+
{
|
| 4244 |
+
"epoch": 0.1922892337226044,
|
| 4245 |
+
"grad_norm": 5.824222087860107,
|
| 4246 |
+
"learning_rate": 0.00036586928005909067,
|
| 4247 |
+
"loss": 6.3228,
|
| 4248 |
+
"step": 601
|
| 4249 |
+
},
|
| 4250 |
+
{
|
| 4251 |
+
"epoch": 0.19260918253079506,
|
| 4252 |
+
"grad_norm": 5.025035381317139,
|
| 4253 |
+
"learning_rate": 0.00036575728485310305,
|
| 4254 |
+
"loss": 6.0095,
|
| 4255 |
+
"step": 602
|
| 4256 |
+
},
|
| 4257 |
+
{
|
| 4258 |
+
"epoch": 0.19292913133898576,
|
| 4259 |
+
"grad_norm": 7.741343975067139,
|
| 4260 |
+
"learning_rate": 0.0003656451243944793,
|
| 4261 |
+
"loss": 6.4401,
|
| 4262 |
+
"step": 603
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 0.19324908014717646,
|
| 4266 |
+
"grad_norm": 12.790258407592773,
|
| 4267 |
+
"learning_rate": 0.00036553279880221736,
|
| 4268 |
+
"loss": 6.0217,
|
| 4269 |
+
"step": 604
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 0.19356902895536715,
|
| 4273 |
+
"grad_norm": 8.1022310256958,
|
| 4274 |
+
"learning_rate": 0.00036542030819549054,
|
| 4275 |
+
"loss": 6.2374,
|
| 4276 |
+
"step": 605
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 0.19388897776355782,
|
| 4280 |
+
"grad_norm": 10.174038887023926,
|
| 4281 |
+
"learning_rate": 0.0003653076526936471,
|
| 4282 |
+
"loss": 6.1404,
|
| 4283 |
+
"step": 606
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 0.1942089265717485,
|
| 4287 |
+
"grad_norm": 5.43154239654541,
|
| 4288 |
+
"learning_rate": 0.0003651948324162104,
|
| 4289 |
+
"loss": 6.0869,
|
| 4290 |
+
"step": 607
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 0.1945288753799392,
|
| 4294 |
+
"grad_norm": 7.523396015167236,
|
| 4295 |
+
"learning_rate": 0.0003650818474828783,
|
| 4296 |
+
"loss": 6.244,
|
| 4297 |
+
"step": 608
|
| 4298 |
+
},
|
| 4299 |
+
{
|
| 4300 |
+
"epoch": 0.1948488241881299,
|
| 4301 |
+
"grad_norm": 9.378880500793457,
|
| 4302 |
+
"learning_rate": 0.0003649686980135238,
|
| 4303 |
+
"loss": 6.3951,
|
| 4304 |
+
"step": 609
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 0.1951687729963206,
|
| 4308 |
+
"grad_norm": 9.061972618103027,
|
| 4309 |
+
"learning_rate": 0.0003648553841281942,
|
| 4310 |
+
"loss": 6.3282,
|
| 4311 |
+
"step": 610
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 0.19548872180451127,
|
| 4315 |
+
"grad_norm": 6.108553409576416,
|
| 4316 |
+
"learning_rate": 0.0003647419059471112,
|
| 4317 |
+
"loss": 6.2572,
|
| 4318 |
+
"step": 611
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 0.19580867061270196,
|
| 4322 |
+
"grad_norm": 5.526744842529297,
|
| 4323 |
+
"learning_rate": 0.00036462826359067097,
|
| 4324 |
+
"loss": 6.2983,
|
| 4325 |
+
"step": 612
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.19612861942089266,
|
| 4329 |
+
"grad_norm": 15.063833236694336,
|
| 4330 |
+
"learning_rate": 0.00036451445717944364,
|
| 4331 |
+
"loss": 6.0462,
|
| 4332 |
+
"step": 613
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 0.19644856822908335,
|
| 4336 |
+
"grad_norm": 8.600305557250977,
|
| 4337 |
+
"learning_rate": 0.00036440048683417374,
|
| 4338 |
+
"loss": 6.0926,
|
| 4339 |
+
"step": 614
|
| 4340 |
+
},
|
| 4341 |
+
{
|
| 4342 |
+
"epoch": 0.19676851703727405,
|
| 4343 |
+
"grad_norm": 6.220649242401123,
|
| 4344 |
+
"learning_rate": 0.00036428635267577944,
|
| 4345 |
+
"loss": 6.4756,
|
| 4346 |
+
"step": 615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.19708846584546472,
|
| 4350 |
+
"grad_norm": 10.122354507446289,
|
| 4351 |
+
"learning_rate": 0.0003641720548253528,
|
| 4352 |
+
"loss": 6.3049,
|
| 4353 |
+
"step": 616
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 0.1974084146536554,
|
| 4357 |
+
"grad_norm": 8.642034530639648,
|
| 4358 |
+
"learning_rate": 0.00036405759340415956,
|
| 4359 |
+
"loss": 5.996,
|
| 4360 |
+
"step": 617
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.1977283634618461,
|
| 4364 |
+
"grad_norm": 7.122521877288818,
|
| 4365 |
+
"learning_rate": 0.0003639429685336391,
|
| 4366 |
+
"loss": 6.2142,
|
| 4367 |
+
"step": 618
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.1980483122700368,
|
| 4371 |
+
"grad_norm": 13.584242820739746,
|
| 4372 |
+
"learning_rate": 0.0003638281803354039,
|
| 4373 |
+
"loss": 6.2225,
|
| 4374 |
+
"step": 619
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 0.19836826107822747,
|
| 4378 |
+
"grad_norm": 15.066466331481934,
|
| 4379 |
+
"learning_rate": 0.00036371322893124,
|
| 4380 |
+
"loss": 6.4064,
|
| 4381 |
+
"step": 620
|
| 4382 |
+
},
|
| 4383 |
+
{
|
| 4384 |
+
"epoch": 0.19868820988641817,
|
| 4385 |
+
"grad_norm": 13.614229202270508,
|
| 4386 |
+
"learning_rate": 0.00036359811444310665,
|
| 4387 |
+
"loss": 6.5411,
|
| 4388 |
+
"step": 621
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 0.19900815869460886,
|
| 4392 |
+
"grad_norm": 21.728307723999023,
|
| 4393 |
+
"learning_rate": 0.0003634828369931358,
|
| 4394 |
+
"loss": 6.1478,
|
| 4395 |
+
"step": 622
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 0.19932810750279956,
|
| 4399 |
+
"grad_norm": 13.957999229431152,
|
| 4400 |
+
"learning_rate": 0.0003633673967036327,
|
| 4401 |
+
"loss": 6.4172,
|
| 4402 |
+
"step": 623
|
| 4403 |
+
},
|
| 4404 |
+
{
|
| 4405 |
+
"epoch": 0.19964805631099025,
|
| 4406 |
+
"grad_norm": 8.37833023071289,
|
| 4407 |
+
"learning_rate": 0.0003632517936970751,
|
| 4408 |
+
"loss": 6.4588,
|
| 4409 |
+
"step": 624
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.19996800511918092,
|
| 4413 |
+
"grad_norm": 10.713350296020508,
|
| 4414 |
+
"learning_rate": 0.0003631360280961134,
|
| 4415 |
+
"loss": 6.2626,
|
| 4416 |
+
"step": 625
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 0.20028795392737161,
|
| 4420 |
+
"grad_norm": 10.606658935546875,
|
| 4421 |
+
"learning_rate": 0.00036302010002357057,
|
| 4422 |
+
"loss": 6.4442,
|
| 4423 |
+
"step": 626
|
| 4424 |
+
},
|
| 4425 |
+
{
|
| 4426 |
+
"epoch": 0.2006079027355623,
|
| 4427 |
+
"grad_norm": 6.849085330963135,
|
| 4428 |
+
"learning_rate": 0.00036290400960244204,
|
| 4429 |
+
"loss": 6.1653,
|
| 4430 |
+
"step": 627
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.200927851543753,
|
| 4434 |
+
"grad_norm": 9.328271865844727,
|
| 4435 |
+
"learning_rate": 0.0003627877569558953,
|
| 4436 |
+
"loss": 6.2049,
|
| 4437 |
+
"step": 628
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 0.2012478003519437,
|
| 4441 |
+
"grad_norm": 20.28767967224121,
|
| 4442 |
+
"learning_rate": 0.00036267134220727,
|
| 4443 |
+
"loss": 6.4698,
|
| 4444 |
+
"step": 629
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.20156774916013437,
|
| 4448 |
+
"grad_norm": 9.141938209533691,
|
| 4449 |
+
"learning_rate": 0.00036255476548007794,
|
| 4450 |
+
"loss": 6.1397,
|
| 4451 |
+
"step": 630
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.20188769796832506,
|
| 4455 |
+
"grad_norm": 7.7828369140625,
|
| 4456 |
+
"learning_rate": 0.00036243802689800257,
|
| 4457 |
+
"loss": 6.0992,
|
| 4458 |
+
"step": 631
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 0.20220764677651576,
|
| 4462 |
+
"grad_norm": 9.355428695678711,
|
| 4463 |
+
"learning_rate": 0.0003623211265848993,
|
| 4464 |
+
"loss": 6.1,
|
| 4465 |
+
"step": 632
|
| 4466 |
+
},
|
| 4467 |
+
{
|
| 4468 |
+
"epoch": 0.20252759558470645,
|
| 4469 |
+
"grad_norm": 6.362244129180908,
|
| 4470 |
+
"learning_rate": 0.0003622040646647947,
|
| 4471 |
+
"loss": 6.1492,
|
| 4472 |
+
"step": 633
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.20284754439289715,
|
| 4476 |
+
"grad_norm": 15.05521011352539,
|
| 4477 |
+
"learning_rate": 0.0003620868412618873,
|
| 4478 |
+
"loss": 6.4308,
|
| 4479 |
+
"step": 634
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 0.20316749320108782,
|
| 4483 |
+
"grad_norm": 10.501233100891113,
|
| 4484 |
+
"learning_rate": 0.00036196945650054674,
|
| 4485 |
+
"loss": 6.0315,
|
| 4486 |
+
"step": 635
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.2034874420092785,
|
| 4490 |
+
"grad_norm": 10.360550880432129,
|
| 4491 |
+
"learning_rate": 0.00036185191050531367,
|
| 4492 |
+
"loss": 6.0577,
|
| 4493 |
+
"step": 636
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.2038073908174692,
|
| 4497 |
+
"grad_norm": 13.611196517944336,
|
| 4498 |
+
"learning_rate": 0.0003617342034009001,
|
| 4499 |
+
"loss": 6.3088,
|
| 4500 |
+
"step": 637
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 0.2041273396256599,
|
| 4504 |
+
"grad_norm": 19.629377365112305,
|
| 4505 |
+
"learning_rate": 0.0003616163353121889,
|
| 4506 |
+
"loss": 6.4983,
|
| 4507 |
+
"step": 638
|
| 4508 |
+
},
|
| 4509 |
+
{
|
| 4510 |
+
"epoch": 0.20444728843385057,
|
| 4511 |
+
"grad_norm": 9.073545455932617,
|
| 4512 |
+
"learning_rate": 0.0003614983063642336,
|
| 4513 |
+
"loss": 6.2678,
|
| 4514 |
+
"step": 639
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.20476723724204127,
|
| 4518 |
+
"grad_norm": 15.663934707641602,
|
| 4519 |
+
"learning_rate": 0.0003613801166822585,
|
| 4520 |
+
"loss": 6.2223,
|
| 4521 |
+
"step": 640
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 0.20508718605023196,
|
| 4525 |
+
"grad_norm": 13.69797134399414,
|
| 4526 |
+
"learning_rate": 0.00036126176639165845,
|
| 4527 |
+
"loss": 6.1535,
|
| 4528 |
+
"step": 641
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.20540713485842266,
|
| 4532 |
+
"grad_norm": 10.088582038879395,
|
| 4533 |
+
"learning_rate": 0.0003611432556179986,
|
| 4534 |
+
"loss": 6.0588,
|
| 4535 |
+
"step": 642
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.20572708366661335,
|
| 4539 |
+
"grad_norm": 52.31211853027344,
|
| 4540 |
+
"learning_rate": 0.0003610245844870145,
|
| 4541 |
+
"loss": 6.5132,
|
| 4542 |
+
"step": 643
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 0.20604703247480402,
|
| 4546 |
+
"grad_norm": 15.596002578735352,
|
| 4547 |
+
"learning_rate": 0.0003609057531246118,
|
| 4548 |
+
"loss": 6.3505,
|
| 4549 |
+
"step": 644
|
| 4550 |
+
},
|
| 4551 |
+
{
|
| 4552 |
+
"epoch": 0.20636698128299472,
|
| 4553 |
+
"grad_norm": 14.089200019836426,
|
| 4554 |
+
"learning_rate": 0.00036078676165686603,
|
| 4555 |
+
"loss": 6.1755,
|
| 4556 |
+
"step": 645
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.2066869300911854,
|
| 4560 |
+
"grad_norm": 22.81130599975586,
|
| 4561 |
+
"learning_rate": 0.00036066761021002276,
|
| 4562 |
+
"loss": 6.7358,
|
| 4563 |
+
"step": 646
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 0.2070068788993761,
|
| 4567 |
+
"grad_norm": 27.079946517944336,
|
| 4568 |
+
"learning_rate": 0.0003605482989104972,
|
| 4569 |
+
"loss": 6.1623,
|
| 4570 |
+
"step": 647
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.2073268277075668,
|
| 4574 |
+
"grad_norm": 90.47372436523438,
|
| 4575 |
+
"learning_rate": 0.00036042882788487403,
|
| 4576 |
+
"loss": 6.2518,
|
| 4577 |
+
"step": 648
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.20764677651575747,
|
| 4581 |
+
"grad_norm": 12.399378776550293,
|
| 4582 |
+
"learning_rate": 0.00036030919725990764,
|
| 4583 |
+
"loss": 6.1738,
|
| 4584 |
+
"step": 649
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 0.20796672532394817,
|
| 4588 |
+
"grad_norm": 24.64794158935547,
|
| 4589 |
+
"learning_rate": 0.00036018940716252165,
|
| 4590 |
+
"loss": 6.2701,
|
| 4591 |
+
"step": 650
|
| 4592 |
+
},
|
| 4593 |
+
{
|
| 4594 |
+
"epoch": 0.20828667413213886,
|
| 4595 |
+
"grad_norm": 31.136415481567383,
|
| 4596 |
+
"learning_rate": 0.0003600694577198088,
|
| 4597 |
+
"loss": 6.4434,
|
| 4598 |
+
"step": 651
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 0.20860662294032956,
|
| 4602 |
+
"grad_norm": 11.99364948272705,
|
| 4603 |
+
"learning_rate": 0.000359949349059031,
|
| 4604 |
+
"loss": 6.2209,
|
| 4605 |
+
"step": 652
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 0.20892657174852022,
|
| 4609 |
+
"grad_norm": 22.257827758789062,
|
| 4610 |
+
"learning_rate": 0.000359829081307619,
|
| 4611 |
+
"loss": 6.0483,
|
| 4612 |
+
"step": 653
|
| 4613 |
+
},
|
| 4614 |
+
{
|
| 4615 |
+
"epoch": 0.20924652055671092,
|
| 4616 |
+
"grad_norm": 42.29777145385742,
|
| 4617 |
+
"learning_rate": 0.0003597086545931724,
|
| 4618 |
+
"loss": 6.1647,
|
| 4619 |
+
"step": 654
|
| 4620 |
+
},
|
| 4621 |
+
{
|
| 4622 |
+
"epoch": 0.20956646936490161,
|
| 4623 |
+
"grad_norm": 22.489221572875977,
|
| 4624 |
+
"learning_rate": 0.00035958806904345936,
|
| 4625 |
+
"loss": 6.2579,
|
| 4626 |
+
"step": 655
|
| 4627 |
+
},
|
| 4628 |
+
{
|
| 4629 |
+
"epoch": 0.2098864181730923,
|
| 4630 |
+
"grad_norm": 28.098785400390625,
|
| 4631 |
+
"learning_rate": 0.0003594673247864167,
|
| 4632 |
+
"loss": 6.2493,
|
| 4633 |
+
"step": 656
|
| 4634 |
+
},
|
| 4635 |
+
{
|
| 4636 |
+
"epoch": 0.210206366981283,
|
| 4637 |
+
"grad_norm": 13.18166732788086,
|
| 4638 |
+
"learning_rate": 0.00035934642195014954,
|
| 4639 |
+
"loss": 6.3545,
|
| 4640 |
+
"step": 657
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 0.21052631578947367,
|
| 4644 |
+
"grad_norm": 19.027376174926758,
|
| 4645 |
+
"learning_rate": 0.0003592253606629312,
|
| 4646 |
+
"loss": 6.3131,
|
| 4647 |
+
"step": 658
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 0.21084626459766437,
|
| 4651 |
+
"grad_norm": 17.17510414123535,
|
| 4652 |
+
"learning_rate": 0.0003591041410532032,
|
| 4653 |
+
"loss": 6.2641,
|
| 4654 |
+
"step": 659
|
| 4655 |
+
},
|
| 4656 |
+
{
|
| 4657 |
+
"epoch": 0.21116621340585506,
|
| 4658 |
+
"grad_norm": 24.58960723876953,
|
| 4659 |
+
"learning_rate": 0.000358982763249575,
|
| 4660 |
+
"loss": 6.1744,
|
| 4661 |
+
"step": 660
|
| 4662 |
+
},
|
| 4663 |
+
{
|
| 4664 |
+
"epoch": 0.21148616221404576,
|
| 4665 |
+
"grad_norm": 74.80911254882812,
|
| 4666 |
+
"learning_rate": 0.0003588612273808239,
|
| 4667 |
+
"loss": 6.2601,
|
| 4668 |
+
"step": 661
|
| 4669 |
+
},
|
| 4670 |
+
{
|
| 4671 |
+
"epoch": 0.21180611102223645,
|
| 4672 |
+
"grad_norm": 24.598215103149414,
|
| 4673 |
+
"learning_rate": 0.0003587395335758949,
|
| 4674 |
+
"loss": 6.0995,
|
| 4675 |
+
"step": 662
|
| 4676 |
+
},
|
| 4677 |
+
{
|
| 4678 |
+
"epoch": 0.21212605983042712,
|
| 4679 |
+
"grad_norm": 27.990663528442383,
|
| 4680 |
+
"learning_rate": 0.00035861768196390054,
|
| 4681 |
+
"loss": 6.3451,
|
| 4682 |
+
"step": 663
|
| 4683 |
+
},
|
| 4684 |
+
{
|
| 4685 |
+
"epoch": 0.21244600863861782,
|
| 4686 |
+
"grad_norm": 30.33852767944336,
|
| 4687 |
+
"learning_rate": 0.0003584956726741208,
|
| 4688 |
+
"loss": 6.1803,
|
| 4689 |
+
"step": 664
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 0.2127659574468085,
|
| 4693 |
+
"grad_norm": 21.035871505737305,
|
| 4694 |
+
"learning_rate": 0.00035837350583600296,
|
| 4695 |
+
"loss": 6.0933,
|
| 4696 |
+
"step": 665
|
| 4697 |
+
},
|
| 4698 |
+
{
|
| 4699 |
+
"epoch": 0.2130859062549992,
|
| 4700 |
+
"grad_norm": 18.981107711791992,
|
| 4701 |
+
"learning_rate": 0.0003582511815791615,
|
| 4702 |
+
"loss": 6.2382,
|
| 4703 |
+
"step": 666
|
| 4704 |
+
},
|
| 4705 |
+
{
|
| 4706 |
+
"epoch": 0.2134058550631899,
|
| 4707 |
+
"grad_norm": 15.4559965133667,
|
| 4708 |
+
"learning_rate": 0.0003581287000333779,
|
| 4709 |
+
"loss": 6.2025,
|
| 4710 |
+
"step": 667
|
| 4711 |
+
},
|
| 4712 |
+
{
|
| 4713 |
+
"epoch": 0.21372580387138057,
|
| 4714 |
+
"grad_norm": 42.094425201416016,
|
| 4715 |
+
"learning_rate": 0.00035800606132860036,
|
| 4716 |
+
"loss": 6.2438,
|
| 4717 |
+
"step": 668
|
| 4718 |
+
},
|
| 4719 |
+
{
|
| 4720 |
+
"epoch": 0.21404575267957127,
|
| 4721 |
+
"grad_norm": 394.44744873046875,
|
| 4722 |
+
"learning_rate": 0.000357883265594944,
|
| 4723 |
+
"loss": 6.1518,
|
| 4724 |
+
"step": 669
|
| 4725 |
+
},
|
| 4726 |
+
{
|
| 4727 |
+
"epoch": 0.21436570148776196,
|
| 4728 |
+
"grad_norm": 23.389293670654297,
|
| 4729 |
+
"learning_rate": 0.00035776031296269053,
|
| 4730 |
+
"loss": 6.2643,
|
| 4731 |
+
"step": 670
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 0.21468565029595266,
|
| 4735 |
+
"grad_norm": 26.32927703857422,
|
| 4736 |
+
"learning_rate": 0.00035763720356228807,
|
| 4737 |
+
"loss": 6.1562,
|
| 4738 |
+
"step": 671
|
| 4739 |
+
},
|
| 4740 |
+
{
|
| 4741 |
+
"epoch": 0.21500559910414332,
|
| 4742 |
+
"grad_norm": 51.09504318237305,
|
| 4743 |
+
"learning_rate": 0.000357513937524351,
|
| 4744 |
+
"loss": 6.1792,
|
| 4745 |
+
"step": 672
|
| 4746 |
+
},
|
| 4747 |
+
{
|
| 4748 |
+
"epoch": 0.21532554791233402,
|
| 4749 |
+
"grad_norm": 55.16264724731445,
|
| 4750 |
+
"learning_rate": 0.00035739051497966,
|
| 4751 |
+
"loss": 6.2488,
|
| 4752 |
+
"step": 673
|
| 4753 |
+
},
|
| 4754 |
+
{
|
| 4755 |
+
"epoch": 0.21564549672052472,
|
| 4756 |
+
"grad_norm": 13.69837760925293,
|
| 4757 |
+
"learning_rate": 0.00035726693605916184,
|
| 4758 |
+
"loss": 6.0671,
|
| 4759 |
+
"step": 674
|
| 4760 |
+
},
|
| 4761 |
+
{
|
| 4762 |
+
"epoch": 0.2159654455287154,
|
| 4763 |
+
"grad_norm": 30.63555908203125,
|
| 4764 |
+
"learning_rate": 0.00035714320089396903,
|
| 4765 |
+
"loss": 6.208,
|
| 4766 |
+
"step": 675
|
| 4767 |
+
},
|
| 4768 |
+
{
|
| 4769 |
+
"epoch": 0.2162853943369061,
|
| 4770 |
+
"grad_norm": 13.71101188659668,
|
| 4771 |
+
"learning_rate": 0.00035701930961536,
|
| 4772 |
+
"loss": 6.1028,
|
| 4773 |
+
"step": 676
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 0.21660534314509677,
|
| 4777 |
+
"grad_norm": 16.373701095581055,
|
| 4778 |
+
"learning_rate": 0.00035689526235477867,
|
| 4779 |
+
"loss": 6.2419,
|
| 4780 |
+
"step": 677
|
| 4781 |
+
},
|
| 4782 |
+
{
|
| 4783 |
+
"epoch": 0.21692529195328747,
|
| 4784 |
+
"grad_norm": 19.588153839111328,
|
| 4785 |
+
"learning_rate": 0.0003567710592438346,
|
| 4786 |
+
"loss": 6.1758,
|
| 4787 |
+
"step": 678
|
| 4788 |
+
},
|
| 4789 |
+
{
|
| 4790 |
+
"epoch": 0.21724524076147816,
|
| 4791 |
+
"grad_norm": 31.715362548828125,
|
| 4792 |
+
"learning_rate": 0.0003566467004143025,
|
| 4793 |
+
"loss": 6.2205,
|
| 4794 |
+
"step": 679
|
| 4795 |
+
},
|
| 4796 |
+
{
|
| 4797 |
+
"epoch": 0.21756518956966886,
|
| 4798 |
+
"grad_norm": 32.14853286743164,
|
| 4799 |
+
"learning_rate": 0.0003565221859981226,
|
| 4800 |
+
"loss": 6.2068,
|
| 4801 |
+
"step": 680
|
| 4802 |
+
},
|
| 4803 |
+
{
|
| 4804 |
+
"epoch": 0.21788513837785956,
|
| 4805 |
+
"grad_norm": 28.47088623046875,
|
| 4806 |
+
"learning_rate": 0.0003563975161274,
|
| 4807 |
+
"loss": 6.3612,
|
| 4808 |
+
"step": 681
|
| 4809 |
+
},
|
| 4810 |
+
{
|
| 4811 |
+
"epoch": 0.21820508718605022,
|
| 4812 |
+
"grad_norm": 27.772294998168945,
|
| 4813 |
+
"learning_rate": 0.00035627269093440467,
|
| 4814 |
+
"loss": 6.3503,
|
| 4815 |
+
"step": 682
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 0.21852503599424092,
|
| 4819 |
+
"grad_norm": 27.19485855102539,
|
| 4820 |
+
"learning_rate": 0.00035614771055157143,
|
| 4821 |
+
"loss": 6.222,
|
| 4822 |
+
"step": 683
|
| 4823 |
+
},
|
| 4824 |
+
{
|
| 4825 |
+
"epoch": 0.2188449848024316,
|
| 4826 |
+
"grad_norm": 14.713149070739746,
|
| 4827 |
+
"learning_rate": 0.00035602257511149993,
|
| 4828 |
+
"loss": 6.3099,
|
| 4829 |
+
"step": 684
|
| 4830 |
+
},
|
| 4831 |
+
{
|
| 4832 |
+
"epoch": 0.2191649336106223,
|
| 4833 |
+
"grad_norm": 10.625816345214844,
|
| 4834 |
+
"learning_rate": 0.00035589728474695403,
|
| 4835 |
+
"loss": 6.1701,
|
| 4836 |
+
"step": 685
|
| 4837 |
+
},
|
| 4838 |
+
{
|
| 4839 |
+
"epoch": 0.21948488241881298,
|
| 4840 |
+
"grad_norm": 17.59633445739746,
|
| 4841 |
+
"learning_rate": 0.0003557718395908622,
|
| 4842 |
+
"loss": 6.1059,
|
| 4843 |
+
"step": 686
|
| 4844 |
+
},
|
| 4845 |
+
{
|
| 4846 |
+
"epoch": 0.21980483122700367,
|
| 4847 |
+
"grad_norm": 40.79133987426758,
|
| 4848 |
+
"learning_rate": 0.000355646239776317,
|
| 4849 |
+
"loss": 6.2505,
|
| 4850 |
+
"step": 687
|
| 4851 |
+
},
|
| 4852 |
+
{
|
| 4853 |
+
"epoch": 0.22012478003519437,
|
| 4854 |
+
"grad_norm": 21.789743423461914,
|
| 4855 |
+
"learning_rate": 0.0003555204854365751,
|
| 4856 |
+
"loss": 6.2508,
|
| 4857 |
+
"step": 688
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 0.22044472884338506,
|
| 4861 |
+
"grad_norm": 287.25286865234375,
|
| 4862 |
+
"learning_rate": 0.0003553945767050572,
|
| 4863 |
+
"loss": 6.3592,
|
| 4864 |
+
"step": 689
|
| 4865 |
+
},
|
| 4866 |
+
{
|
| 4867 |
+
"epoch": 0.22076467765157576,
|
| 4868 |
+
"grad_norm": 122.58528900146484,
|
| 4869 |
+
"learning_rate": 0.0003552685137153477,
|
| 4870 |
+
"loss": 6.5598,
|
| 4871 |
+
"step": 690
|
| 4872 |
+
},
|
| 4873 |
+
{
|
| 4874 |
+
"epoch": 0.22108462645976643,
|
| 4875 |
+
"grad_norm": 65.58609771728516,
|
| 4876 |
+
"learning_rate": 0.00035514229660119475,
|
| 4877 |
+
"loss": 6.2963,
|
| 4878 |
+
"step": 691
|
| 4879 |
+
},
|
| 4880 |
+
{
|
| 4881 |
+
"epoch": 0.22140457526795712,
|
| 4882 |
+
"grad_norm": 78.26929473876953,
|
| 4883 |
+
"learning_rate": 0.00035501592549650986,
|
| 4884 |
+
"loss": 6.6145,
|
| 4885 |
+
"step": 692
|
| 4886 |
+
},
|
| 4887 |
+
{
|
| 4888 |
+
"epoch": 0.22172452407614782,
|
| 4889 |
+
"grad_norm": 468.92840576171875,
|
| 4890 |
+
"learning_rate": 0.0003548894005353682,
|
| 4891 |
+
"loss": 6.5893,
|
| 4892 |
+
"step": 693
|
| 4893 |
+
},
|
| 4894 |
+
{
|
| 4895 |
+
"epoch": 0.2220444728843385,
|
| 4896 |
+
"grad_norm": 29858.107421875,
|
| 4897 |
+
"learning_rate": 0.00035476272185200786,
|
| 4898 |
+
"loss": 6.8783,
|
| 4899 |
+
"step": 694
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 0.2223644216925292,
|
| 4903 |
+
"grad_norm": 567.8423461914062,
|
| 4904 |
+
"learning_rate": 0.00035463588958083023,
|
| 4905 |
+
"loss": 7.506,
|
| 4906 |
+
"step": 695
|
| 4907 |
+
},
|
| 4908 |
+
{
|
| 4909 |
+
"epoch": 0.22268437050071987,
|
| 4910 |
+
"grad_norm": 269.0973815917969,
|
| 4911 |
+
"learning_rate": 0.00035450890385639957,
|
| 4912 |
+
"loss": 6.98,
|
| 4913 |
+
"step": 696
|
| 4914 |
+
},
|
| 4915 |
+
{
|
| 4916 |
+
"epoch": 0.22300431930891057,
|
| 4917 |
+
"grad_norm": 70.39237976074219,
|
| 4918 |
+
"learning_rate": 0.00035438176481344307,
|
| 4919 |
+
"loss": 6.6005,
|
| 4920 |
+
"step": 697
|
| 4921 |
+
},
|
| 4922 |
+
{
|
| 4923 |
+
"epoch": 0.22332426811710127,
|
| 4924 |
+
"grad_norm": 35.886253356933594,
|
| 4925 |
+
"learning_rate": 0.0003542544725868503,
|
| 4926 |
+
"loss": 6.3986,
|
| 4927 |
+
"step": 698
|
| 4928 |
+
},
|
| 4929 |
+
{
|
| 4930 |
+
"epoch": 0.22364421692529196,
|
| 4931 |
+
"grad_norm": 132.87338256835938,
|
| 4932 |
+
"learning_rate": 0.0003541270273116737,
|
| 4933 |
+
"loss": 6.4347,
|
| 4934 |
+
"step": 699
|
| 4935 |
+
},
|
| 4936 |
+
{
|
| 4937 |
+
"epoch": 0.22396416573348266,
|
| 4938 |
+
"grad_norm": 243.24110412597656,
|
| 4939 |
+
"learning_rate": 0.00035399942912312784,
|
| 4940 |
+
"loss": 6.6825,
|
| 4941 |
+
"step": 700
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 0.22428411454167332,
|
| 4945 |
+
"grad_norm": 47.076541900634766,
|
| 4946 |
+
"learning_rate": 0.00035387167815658955,
|
| 4947 |
+
"loss": 6.5573,
|
| 4948 |
+
"step": 701
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.22460406334986402,
|
| 4952 |
+
"grad_norm": 47.393280029296875,
|
| 4953 |
+
"learning_rate": 0.0003537437745475981,
|
| 4954 |
+
"loss": 6.5841,
|
| 4955 |
+
"step": 702
|
| 4956 |
+
},
|
| 4957 |
+
{
|
| 4958 |
+
"epoch": 0.22492401215805471,
|
| 4959 |
+
"grad_norm": 173.40023803710938,
|
| 4960 |
+
"learning_rate": 0.0003536157184318541,
|
| 4961 |
+
"loss": 6.5547,
|
| 4962 |
+
"step": 703
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 0.2252439609662454,
|
| 4966 |
+
"grad_norm": 100.88816833496094,
|
| 4967 |
+
"learning_rate": 0.00035348750994522037,
|
| 4968 |
+
"loss": 6.5479,
|
| 4969 |
+
"step": 704
|
| 4970 |
+
},
|
| 4971 |
+
{
|
| 4972 |
+
"epoch": 0.22556390977443608,
|
| 4973 |
+
"grad_norm": 708.522216796875,
|
| 4974 |
+
"learning_rate": 0.00035335914922372135,
|
| 4975 |
+
"loss": 6.4429,
|
| 4976 |
+
"step": 705
|
| 4977 |
+
},
|
| 4978 |
+
{
|
| 4979 |
+
"epoch": 0.22588385858262677,
|
| 4980 |
+
"grad_norm": 5974.12744140625,
|
| 4981 |
+
"learning_rate": 0.000353230636403543,
|
| 4982 |
+
"loss": 6.8338,
|
| 4983 |
+
"step": 706
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 0.22620380739081747,
|
| 4987 |
+
"grad_norm": 1240.005615234375,
|
| 4988 |
+
"learning_rate": 0.0003531019716210326,
|
| 4989 |
+
"loss": 7.7215,
|
| 4990 |
+
"step": 707
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.22652375619900816,
|
| 4994 |
+
"grad_norm": 3238.5185546875,
|
| 4995 |
+
"learning_rate": 0.00035297315501269845,
|
| 4996 |
+
"loss": 10.7748,
|
| 4997 |
+
"step": 708
|
| 4998 |
+
},
|
| 4999 |
+
{
|
| 5000 |
+
"epoch": 0.22684370500719886,
|
| 5001 |
+
"grad_norm": 6019.43310546875,
|
| 5002 |
+
"learning_rate": 0.00035284418671521035,
|
| 5003 |
+
"loss": 10.382,
|
| 5004 |
+
"step": 709
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 0.22716365381538953,
|
| 5008 |
+
"grad_norm": 4291.9306640625,
|
| 5009 |
+
"learning_rate": 0.00035271506686539877,
|
| 5010 |
+
"loss": 8.5623,
|
| 5011 |
+
"step": 710
|
| 5012 |
+
},
|
| 5013 |
+
{
|
| 5014 |
+
"epoch": 0.22748360262358022,
|
| 5015 |
+
"grad_norm": 1000.52294921875,
|
| 5016 |
+
"learning_rate": 0.00035258579560025504,
|
| 5017 |
+
"loss": 6.9676,
|
| 5018 |
+
"step": 711
|
| 5019 |
+
},
|
| 5020 |
+
{
|
| 5021 |
+
"epoch": 0.22780355143177092,
|
| 5022 |
+
"grad_norm": 653.64990234375,
|
| 5023 |
+
"learning_rate": 0.0003524563730569311,
|
| 5024 |
+
"loss": 6.8677,
|
| 5025 |
+
"step": 712
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 0.2281235002399616,
|
| 5029 |
+
"grad_norm": 226.84967041015625,
|
| 5030 |
+
"learning_rate": 0.0003523267993727394,
|
| 5031 |
+
"loss": 6.5646,
|
| 5032 |
+
"step": 713
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.2284434490481523,
|
| 5036 |
+
"grad_norm": 69.58618927001953,
|
| 5037 |
+
"learning_rate": 0.00035219707468515276,
|
| 5038 |
+
"loss": 6.8388,
|
| 5039 |
+
"step": 714
|
| 5040 |
+
},
|
| 5041 |
+
{
|
| 5042 |
+
"epoch": 0.22876339785634298,
|
| 5043 |
+
"grad_norm": 723.3787841796875,
|
| 5044 |
+
"learning_rate": 0.00035206719913180427,
|
| 5045 |
+
"loss": 6.8472,
|
| 5046 |
+
"step": 715
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 0.22908334666453367,
|
| 5050 |
+
"grad_norm": 97.80406951904297,
|
| 5051 |
+
"learning_rate": 0.00035193717285048694,
|
| 5052 |
+
"loss": 7.5312,
|
| 5053 |
+
"step": 716
|
| 5054 |
+
},
|
| 5055 |
+
{
|
| 5056 |
+
"epoch": 0.22940329547272437,
|
| 5057 |
+
"grad_norm": 114.102294921875,
|
| 5058 |
+
"learning_rate": 0.00035180699597915383,
|
| 5059 |
+
"loss": 8.069,
|
| 5060 |
+
"step": 717
|
| 5061 |
+
},
|
| 5062 |
+
{
|
| 5063 |
+
"epoch": 0.22972324428091506,
|
| 5064 |
+
"grad_norm": 236.4305419921875,
|
| 5065 |
+
"learning_rate": 0.0003516766686559177,
|
| 5066 |
+
"loss": 8.6465,
|
| 5067 |
+
"step": 718
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 0.23004319308910573,
|
| 5071 |
+
"grad_norm": 99.95079040527344,
|
| 5072 |
+
"learning_rate": 0.0003515461910190509,
|
| 5073 |
+
"loss": 7.5548,
|
| 5074 |
+
"step": 719
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.23036314189729643,
|
| 5078 |
+
"grad_norm": 84.75758361816406,
|
| 5079 |
+
"learning_rate": 0.0003514155632069854,
|
| 5080 |
+
"loss": 6.442,
|
| 5081 |
+
"step": 720
|
| 5082 |
+
},
|
| 5083 |
+
{
|
| 5084 |
+
"epoch": 0.23068309070548712,
|
| 5085 |
+
"grad_norm": 212.29820251464844,
|
| 5086 |
+
"learning_rate": 0.00035128478535831234,
|
| 5087 |
+
"loss": 8.096,
|
| 5088 |
+
"step": 721
|
| 5089 |
+
},
|
| 5090 |
+
{
|
| 5091 |
+
"epoch": 0.23100303951367782,
|
| 5092 |
+
"grad_norm": 57.954811096191406,
|
| 5093 |
+
"learning_rate": 0.0003511538576117821,
|
| 5094 |
+
"loss": 6.924,
|
| 5095 |
+
"step": 722
|
| 5096 |
+
},
|
| 5097 |
+
{
|
| 5098 |
+
"epoch": 0.2313229883218685,
|
| 5099 |
+
"grad_norm": 51.535804748535156,
|
| 5100 |
+
"learning_rate": 0.0003510227801063042,
|
| 5101 |
+
"loss": 6.6562,
|
| 5102 |
+
"step": 723
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 0.23164293713005918,
|
| 5106 |
+
"grad_norm": 1651.22705078125,
|
| 5107 |
+
"learning_rate": 0.0003508915529809469,
|
| 5108 |
+
"loss": 7.1549,
|
| 5109 |
+
"step": 724
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 0.23196288593824987,
|
| 5113 |
+
"grad_norm": 103.06959533691406,
|
| 5114 |
+
"learning_rate": 0.0003507601763749372,
|
| 5115 |
+
"loss": 7.1778,
|
| 5116 |
+
"step": 725
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.23228283474644057,
|
| 5120 |
+
"grad_norm": 127.42375183105469,
|
| 5121 |
+
"learning_rate": 0.00035062865042766086,
|
| 5122 |
+
"loss": 7.0844,
|
| 5123 |
+
"step": 726
|
| 5124 |
+
},
|
| 5125 |
+
{
|
| 5126 |
+
"epoch": 0.23260278355463126,
|
| 5127 |
+
"grad_norm": 57.47138214111328,
|
| 5128 |
+
"learning_rate": 0.0003504969752786619,
|
| 5129 |
+
"loss": 6.7034,
|
| 5130 |
+
"step": 727
|
| 5131 |
+
},
|
| 5132 |
+
{
|
| 5133 |
+
"epoch": 0.23292273236282196,
|
| 5134 |
+
"grad_norm": 21.57773208618164,
|
| 5135 |
+
"learning_rate": 0.00035036515106764277,
|
| 5136 |
+
"loss": 6.5883,
|
| 5137 |
+
"step": 728
|
| 5138 |
+
},
|
| 5139 |
+
{
|
| 5140 |
+
"epoch": 0.23324268117101263,
|
| 5141 |
+
"grad_norm": 43.577693939208984,
|
| 5142 |
+
"learning_rate": 0.00035023317793446396,
|
| 5143 |
+
"loss": 7.1802,
|
| 5144 |
+
"step": 729
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 0.23356262997920332,
|
| 5148 |
+
"grad_norm": 47.857906341552734,
|
| 5149 |
+
"learning_rate": 0.000350101056019144,
|
| 5150 |
+
"loss": 7.1532,
|
| 5151 |
+
"step": 730
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 0.23388257878739402,
|
| 5155 |
+
"grad_norm": 41.412353515625,
|
| 5156 |
+
"learning_rate": 0.0003499687854618593,
|
| 5157 |
+
"loss": 6.8486,
|
| 5158 |
+
"step": 731
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.23420252759558471,
|
| 5162 |
+
"grad_norm": 194.344482421875,
|
| 5163 |
+
"learning_rate": 0.000349836366402944,
|
| 5164 |
+
"loss": 6.6898,
|
| 5165 |
+
"step": 732
|
| 5166 |
+
},
|
| 5167 |
+
{
|
| 5168 |
+
"epoch": 0.2345224764037754,
|
| 5169 |
+
"grad_norm": 97.9305419921875,
|
| 5170 |
+
"learning_rate": 0.00034970379898288976,
|
| 5171 |
+
"loss": 7.0291,
|
| 5172 |
+
"step": 733
|
| 5173 |
+
},
|
| 5174 |
+
{
|
| 5175 |
+
"epoch": 0.23484242521196608,
|
| 5176 |
+
"grad_norm": 617602.9375,
|
| 5177 |
+
"learning_rate": 0.00034957108334234563,
|
| 5178 |
+
"loss": 6.7315,
|
| 5179 |
+
"step": 734
|
| 5180 |
+
},
|
| 5181 |
+
{
|
| 5182 |
+
"epoch": 0.23516237402015677,
|
| 5183 |
+
"grad_norm": 82.41645050048828,
|
| 5184 |
+
"learning_rate": 0.00034943821962211784,
|
| 5185 |
+
"loss": 6.4541,
|
| 5186 |
+
"step": 735
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 0.23548232282834747,
|
| 5190 |
+
"grad_norm": 112.80008697509766,
|
| 5191 |
+
"learning_rate": 0.0003493052079631699,
|
| 5192 |
+
"loss": 6.5458,
|
| 5193 |
+
"step": 736
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 0.23580227163653816,
|
| 5197 |
+
"grad_norm": 57.59135437011719,
|
| 5198 |
+
"learning_rate": 0.00034917204850662225,
|
| 5199 |
+
"loss": 6.7299,
|
| 5200 |
+
"step": 737
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.23612222044472883,
|
| 5204 |
+
"grad_norm": 183.47340393066406,
|
| 5205 |
+
"learning_rate": 0.0003490387413937519,
|
| 5206 |
+
"loss": 6.6954,
|
| 5207 |
+
"step": 738
|
| 5208 |
+
},
|
| 5209 |
+
{
|
| 5210 |
+
"epoch": 0.23644216925291953,
|
| 5211 |
+
"grad_norm": 95.19715118408203,
|
| 5212 |
+
"learning_rate": 0.00034890528676599284,
|
| 5213 |
+
"loss": 6.6315,
|
| 5214 |
+
"step": 739
|
| 5215 |
+
},
|
| 5216 |
+
{
|
| 5217 |
+
"epoch": 0.23676211806111022,
|
| 5218 |
+
"grad_norm": 77.85346221923828,
|
| 5219 |
+
"learning_rate": 0.00034877168476493535,
|
| 5220 |
+
"loss": 6.9589,
|
| 5221 |
+
"step": 740
|
| 5222 |
+
},
|
| 5223 |
+
{
|
| 5224 |
+
"epoch": 0.23708206686930092,
|
| 5225 |
+
"grad_norm": 208.43527221679688,
|
| 5226 |
+
"learning_rate": 0.00034863793553232614,
|
| 5227 |
+
"loss": 6.5637,
|
| 5228 |
+
"step": 741
|
| 5229 |
+
},
|
| 5230 |
+
{
|
| 5231 |
+
"epoch": 0.2374020156774916,
|
| 5232 |
+
"grad_norm": 156.19229125976562,
|
| 5233 |
+
"learning_rate": 0.0003485040392100682,
|
| 5234 |
+
"loss": 7.0049,
|
| 5235 |
+
"step": 742
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 0.23772196448568228,
|
| 5239 |
+
"grad_norm": 199.3235321044922,
|
| 5240 |
+
"learning_rate": 0.00034836999594022044,
|
| 5241 |
+
"loss": 6.2481,
|
| 5242 |
+
"step": 743
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.23804191329387298,
|
| 5246 |
+
"grad_norm": 67.07147979736328,
|
| 5247 |
+
"learning_rate": 0.0003482358058649978,
|
| 5248 |
+
"loss": 6.3019,
|
| 5249 |
+
"step": 744
|
| 5250 |
+
},
|
| 5251 |
+
{
|
| 5252 |
+
"epoch": 0.23836186210206367,
|
| 5253 |
+
"grad_norm": 166.8501434326172,
|
| 5254 |
+
"learning_rate": 0.00034810146912677087,
|
| 5255 |
+
"loss": 6.3432,
|
| 5256 |
+
"step": 745
|
| 5257 |
+
},
|
| 5258 |
+
{
|
| 5259 |
+
"epoch": 0.23868181091025437,
|
| 5260 |
+
"grad_norm": 97.20511627197266,
|
| 5261 |
+
"learning_rate": 0.0003479669858680659,
|
| 5262 |
+
"loss": 6.3981,
|
| 5263 |
+
"step": 746
|
| 5264 |
+
},
|
| 5265 |
+
{
|
| 5266 |
+
"epoch": 0.23900175971844506,
|
| 5267 |
+
"grad_norm": 57.774749755859375,
|
| 5268 |
+
"learning_rate": 0.0003478323562315646,
|
| 5269 |
+
"loss": 6.3404,
|
| 5270 |
+
"step": 747
|
| 5271 |
+
},
|
| 5272 |
+
{
|
| 5273 |
+
"epoch": 0.23932170852663573,
|
| 5274 |
+
"grad_norm": 90.69790649414062,
|
| 5275 |
+
"learning_rate": 0.0003476975803601039,
|
| 5276 |
+
"loss": 6.4574,
|
| 5277 |
+
"step": 748
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 0.23964165733482642,
|
| 5281 |
+
"grad_norm": 170.02110290527344,
|
| 5282 |
+
"learning_rate": 0.000347562658396676,
|
| 5283 |
+
"loss": 6.6967,
|
| 5284 |
+
"step": 749
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.23996160614301712,
|
| 5288 |
+
"grad_norm": 80.30382537841797,
|
| 5289 |
+
"learning_rate": 0.000347427590484428,
|
| 5290 |
+
"loss": 6.3286,
|
| 5291 |
+
"step": 750
|
| 5292 |
+
},
|
| 5293 |
+
{
|
| 5294 |
+
"epoch": 0.24028155495120782,
|
| 5295 |
+
"grad_norm": 78.86782836914062,
|
| 5296 |
+
"learning_rate": 0.00034729237676666194,
|
| 5297 |
+
"loss": 6.2999,
|
| 5298 |
+
"step": 751
|
| 5299 |
+
},
|
| 5300 |
+
{
|
| 5301 |
+
"epoch": 0.24060150375939848,
|
| 5302 |
+
"grad_norm": 73.93916320800781,
|
| 5303 |
+
"learning_rate": 0.0003471570173868345,
|
| 5304 |
+
"loss": 6.4426,
|
| 5305 |
+
"step": 752
|
| 5306 |
+
},
|
| 5307 |
+
{
|
| 5308 |
+
"epoch": 0.24092145256758918,
|
| 5309 |
+
"grad_norm": 140.0988311767578,
|
| 5310 |
+
"learning_rate": 0.0003470215124885569,
|
| 5311 |
+
"loss": 6.4757,
|
| 5312 |
+
"step": 753
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 0.24124140137577987,
|
| 5316 |
+
"grad_norm": 86.3206558227539,
|
| 5317 |
+
"learning_rate": 0.00034688586221559467,
|
| 5318 |
+
"loss": 6.3304,
|
| 5319 |
+
"step": 754
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 0.24156135018397057,
|
| 5323 |
+
"grad_norm": 338.0812072753906,
|
| 5324 |
+
"learning_rate": 0.0003467500667118678,
|
| 5325 |
+
"loss": 6.3679,
|
| 5326 |
+
"step": 755
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.24188129899216126,
|
| 5330 |
+
"grad_norm": 412.3385314941406,
|
| 5331 |
+
"learning_rate": 0.0003466141261214501,
|
| 5332 |
+
"loss": 6.5072,
|
| 5333 |
+
"step": 756
|
| 5334 |
+
},
|
| 5335 |
+
{
|
| 5336 |
+
"epoch": 0.24220124780035193,
|
| 5337 |
+
"grad_norm": 205.6704864501953,
|
| 5338 |
+
"learning_rate": 0.0003464780405885695,
|
| 5339 |
+
"loss": 6.3448,
|
| 5340 |
+
"step": 757
|
| 5341 |
+
},
|
| 5342 |
+
{
|
| 5343 |
+
"epoch": 0.24252119660854263,
|
| 5344 |
+
"grad_norm": 256.4744873046875,
|
| 5345 |
+
"learning_rate": 0.00034634181025760765,
|
| 5346 |
+
"loss": 6.3772,
|
| 5347 |
+
"step": 758
|
| 5348 |
+
},
|
| 5349 |
+
{
|
| 5350 |
+
"epoch": 0.24284114541673332,
|
| 5351 |
+
"grad_norm": 461.74176025390625,
|
| 5352 |
+
"learning_rate": 0.00034620543527309983,
|
| 5353 |
+
"loss": 6.2274,
|
| 5354 |
+
"step": 759
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 0.24316109422492402,
|
| 5358 |
+
"grad_norm": 165.07127380371094,
|
| 5359 |
+
"learning_rate": 0.00034606891577973475,
|
| 5360 |
+
"loss": 6.3089,
|
| 5361 |
+
"step": 760
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 0.2434810430331147,
|
| 5365 |
+
"grad_norm": 1090.9415283203125,
|
| 5366 |
+
"learning_rate": 0.00034593225192235447,
|
| 5367 |
+
"loss": 6.5381,
|
| 5368 |
+
"step": 761
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.24380099184130538,
|
| 5372 |
+
"grad_norm": 1298.724365234375,
|
| 5373 |
+
"learning_rate": 0.00034579544384595427,
|
| 5374 |
+
"loss": 6.4875,
|
| 5375 |
+
"step": 762
|
| 5376 |
+
},
|
| 5377 |
+
{
|
| 5378 |
+
"epoch": 0.24412094064949608,
|
| 5379 |
+
"grad_norm": 2974.205322265625,
|
| 5380 |
+
"learning_rate": 0.0003456584916956823,
|
| 5381 |
+
"loss": 6.609,
|
| 5382 |
+
"step": 763
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 0.24444088945768677,
|
| 5386 |
+
"grad_norm": 506.7083435058594,
|
| 5387 |
+
"learning_rate": 0.0003455213956168397,
|
| 5388 |
+
"loss": 6.5206,
|
| 5389 |
+
"step": 764
|
| 5390 |
+
},
|
| 5391 |
+
{
|
| 5392 |
+
"epoch": 0.24476083826587747,
|
| 5393 |
+
"grad_norm": 2491.966796875,
|
| 5394 |
+
"learning_rate": 0.0003453841557548802,
|
| 5395 |
+
"loss": 6.3059,
|
| 5396 |
+
"step": 765
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.24508078707406816,
|
| 5400 |
+
"grad_norm": 558.4273681640625,
|
| 5401 |
+
"learning_rate": 0.00034524677225541034,
|
| 5402 |
+
"loss": 6.3385,
|
| 5403 |
+
"step": 766
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 0.24540073588225883,
|
| 5407 |
+
"grad_norm": 2481.155029296875,
|
| 5408 |
+
"learning_rate": 0.00034510924526418864,
|
| 5409 |
+
"loss": 6.4753,
|
| 5410 |
+
"step": 767
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.24572068469044953,
|
| 5414 |
+
"grad_norm": 6455.59375,
|
| 5415 |
+
"learning_rate": 0.00034497157492712616,
|
| 5416 |
+
"loss": 6.2157,
|
| 5417 |
+
"step": 768
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.24604063349864022,
|
| 5421 |
+
"grad_norm": 4837.32080078125,
|
| 5422 |
+
"learning_rate": 0.000344833761390286,
|
| 5423 |
+
"loss": 6.5173,
|
| 5424 |
+
"step": 769
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.24636058230683092,
|
| 5428 |
+
"grad_norm": 1091.15869140625,
|
| 5429 |
+
"learning_rate": 0.00034469580479988315,
|
| 5430 |
+
"loss": 6.4935,
|
| 5431 |
+
"step": 770
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.24668053111502158,
|
| 5435 |
+
"grad_norm": 2150.258544921875,
|
| 5436 |
+
"learning_rate": 0.0003445577053022843,
|
| 5437 |
+
"loss": 6.5266,
|
| 5438 |
+
"step": 771
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.24700047992321228,
|
| 5442 |
+
"grad_norm": 3155.990478515625,
|
| 5443 |
+
"learning_rate": 0.0003444194630440079,
|
| 5444 |
+
"loss": 6.4712,
|
| 5445 |
+
"step": 772
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 0.24732042873140297,
|
| 5449 |
+
"grad_norm": 745.5328979492188,
|
| 5450 |
+
"learning_rate": 0.0003442810781717238,
|
| 5451 |
+
"loss": 6.4966,
|
| 5452 |
+
"step": 773
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.24764037753959367,
|
| 5456 |
+
"grad_norm": 2718.595947265625,
|
| 5457 |
+
"learning_rate": 0.00034414255083225315,
|
| 5458 |
+
"loss": 6.3514,
|
| 5459 |
+
"step": 774
|
| 5460 |
+
},
|
| 5461 |
+
{
|
| 5462 |
+
"epoch": 0.24796032634778437,
|
| 5463 |
+
"grad_norm": 1184.953369140625,
|
| 5464 |
+
"learning_rate": 0.0003440038811725683,
|
| 5465 |
+
"loss": 6.5564,
|
| 5466 |
+
"step": 775
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 0.24828027515597503,
|
| 5470 |
+
"grad_norm": 1916.226318359375,
|
| 5471 |
+
"learning_rate": 0.00034386506933979243,
|
| 5472 |
+
"loss": 6.3485,
|
| 5473 |
+
"step": 776
|
| 5474 |
+
},
|
| 5475 |
+
{
|
| 5476 |
+
"epoch": 0.24860022396416573,
|
| 5477 |
+
"grad_norm": 2737.73095703125,
|
| 5478 |
+
"learning_rate": 0.0003437261154811997,
|
| 5479 |
+
"loss": 6.4106,
|
| 5480 |
+
"step": 777
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 0.24892017277235642,
|
| 5484 |
+
"grad_norm": 821.8926391601562,
|
| 5485 |
+
"learning_rate": 0.00034358701974421523,
|
| 5486 |
+
"loss": 6.3447,
|
| 5487 |
+
"step": 778
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 0.24924012158054712,
|
| 5491 |
+
"grad_norm": 781.9716796875,
|
| 5492 |
+
"learning_rate": 0.000343447782276414,
|
| 5493 |
+
"loss": 6.2201,
|
| 5494 |
+
"step": 779
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.24956007038873781,
|
| 5498 |
+
"grad_norm": 2429.21484375,
|
| 5499 |
+
"learning_rate": 0.0003433084032255219,
|
| 5500 |
+
"loss": 6.4331,
|
| 5501 |
+
"step": 780
|
| 5502 |
+
},
|
| 5503 |
+
{
|
| 5504 |
+
"epoch": 0.24988001919692848,
|
| 5505 |
+
"grad_norm": 7888.99853515625,
|
| 5506 |
+
"learning_rate": 0.0003431688827394149,
|
| 5507 |
+
"loss": 6.4502,
|
| 5508 |
+
"step": 781
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 0.2501999680051192,
|
| 5512 |
+
"grad_norm": 728.1339721679688,
|
| 5513 |
+
"learning_rate": 0.000343029220966119,
|
| 5514 |
+
"loss": 6.5406,
|
| 5515 |
+
"step": 782
|
| 5516 |
+
},
|
| 5517 |
+
{
|
| 5518 |
+
"epoch": 0.2505199168133099,
|
| 5519 |
+
"grad_norm": 1479.98095703125,
|
| 5520 |
+
"learning_rate": 0.0003428894180538101,
|
| 5521 |
+
"loss": 6.636,
|
| 5522 |
+
"step": 783
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 0.25083986562150057,
|
| 5526 |
+
"grad_norm": 322.9638366699219,
|
| 5527 |
+
"learning_rate": 0.0003427494741508138,
|
| 5528 |
+
"loss": 6.4123,
|
| 5529 |
+
"step": 784
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 0.25115981442969126,
|
| 5533 |
+
"grad_norm": 567.7588500976562,
|
| 5534 |
+
"learning_rate": 0.00034260938940560527,
|
| 5535 |
+
"loss": 6.5927,
|
| 5536 |
+
"step": 785
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.25147976323788196,
|
| 5540 |
+
"grad_norm": 912.3056640625,
|
| 5541 |
+
"learning_rate": 0.00034246916396680934,
|
| 5542 |
+
"loss": 6.3635,
|
| 5543 |
+
"step": 786
|
| 5544 |
+
},
|
| 5545 |
+
{
|
| 5546 |
+
"epoch": 0.25179971204607265,
|
| 5547 |
+
"grad_norm": 867.4058227539062,
|
| 5548 |
+
"learning_rate": 0.00034232879798319984,
|
| 5549 |
+
"loss": 6.4766,
|
| 5550 |
+
"step": 787
|
| 5551 |
+
},
|
| 5552 |
+
{
|
| 5553 |
+
"epoch": 0.2521196608542633,
|
| 5554 |
+
"grad_norm": 3383.78759765625,
|
| 5555 |
+
"learning_rate": 0.0003421882916036998,
|
| 5556 |
+
"loss": 6.5765,
|
| 5557 |
+
"step": 788
|
| 5558 |
+
},
|
| 5559 |
+
{
|
| 5560 |
+
"epoch": 0.252439609662454,
|
| 5561 |
+
"grad_norm": 551.5289916992188,
|
| 5562 |
+
"learning_rate": 0.0003420476449773813,
|
| 5563 |
+
"loss": 6.1571,
|
| 5564 |
+
"step": 789
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 0.2527595584706447,
|
| 5568 |
+
"grad_norm": 3835.2880859375,
|
| 5569 |
+
"learning_rate": 0.00034190685825346504,
|
| 5570 |
+
"loss": 6.3987,
|
| 5571 |
+
"step": 790
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 0.2530795072788354,
|
| 5575 |
+
"grad_norm": 299.2125549316406,
|
| 5576 |
+
"learning_rate": 0.00034176593158132055,
|
| 5577 |
+
"loss": 6.3278,
|
| 5578 |
+
"step": 791
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.2533994560870261,
|
| 5582 |
+
"grad_norm": 542.0026245117188,
|
| 5583 |
+
"learning_rate": 0.0003416248651104656,
|
| 5584 |
+
"loss": 6.4289,
|
| 5585 |
+
"step": 792
|
| 5586 |
+
},
|
| 5587 |
+
{
|
| 5588 |
+
"epoch": 0.25371940489521677,
|
| 5589 |
+
"grad_norm": 3202.345947265625,
|
| 5590 |
+
"learning_rate": 0.00034148365899056655,
|
| 5591 |
+
"loss": 6.3677,
|
| 5592 |
+
"step": 793
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 0.25403935370340747,
|
| 5596 |
+
"grad_norm": 2746.37109375,
|
| 5597 |
+
"learning_rate": 0.00034134231337143775,
|
| 5598 |
+
"loss": 6.3684,
|
| 5599 |
+
"step": 794
|
| 5600 |
+
},
|
| 5601 |
+
{
|
| 5602 |
+
"epoch": 0.25435930251159816,
|
| 5603 |
+
"grad_norm": 289.69354248046875,
|
| 5604 |
+
"learning_rate": 0.00034120082840304164,
|
| 5605 |
+
"loss": 6.25,
|
| 5606 |
+
"step": 795
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 0.25467925131978886,
|
| 5610 |
+
"grad_norm": 5020.22607421875,
|
| 5611 |
+
"learning_rate": 0.00034105920423548833,
|
| 5612 |
+
"loss": 6.4192,
|
| 5613 |
+
"step": 796
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 0.2549992001279795,
|
| 5617 |
+
"grad_norm": 21526.37109375,
|
| 5618 |
+
"learning_rate": 0.00034091744101903595,
|
| 5619 |
+
"loss": 6.7782,
|
| 5620 |
+
"step": 797
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.2553191489361702,
|
| 5624 |
+
"grad_norm": 96381.3203125,
|
| 5625 |
+
"learning_rate": 0.0003407755389040898,
|
| 5626 |
+
"loss": 6.4677,
|
| 5627 |
+
"step": 798
|
| 5628 |
+
},
|
| 5629 |
+
{
|
| 5630 |
+
"epoch": 0.2556390977443609,
|
| 5631 |
+
"grad_norm": 383685.59375,
|
| 5632 |
+
"learning_rate": 0.00034063349804120275,
|
| 5633 |
+
"loss": 6.3452,
|
| 5634 |
+
"step": 799
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 0.2559590465525516,
|
| 5638 |
+
"grad_norm": 1976569.875,
|
| 5639 |
+
"learning_rate": 0.0003404913185810748,
|
| 5640 |
+
"loss": 7.6792,
|
| 5641 |
+
"step": 800
|
| 5642 |
+
},
|
| 5643 |
+
{
|
| 5644 |
+
"epoch": 0.2559590465525516,
|
| 5645 |
+
"eval_loss": 9.59782886505127,
|
| 5646 |
+
"eval_runtime": 233.5229,
|
| 5647 |
+
"eval_samples_per_second": 5.635,
|
| 5648 |
+
"eval_steps_per_second": 1.409,
|
| 5649 |
+
"step": 800
|
| 5650 |
}
|
| 5651 |
],
|
| 5652 |
"logging_steps": 1,
|
|
|
|
| 5661 |
"early_stopping_threshold": 0.0
|
| 5662 |
},
|
| 5663 |
"attributes": {
|
| 5664 |
+
"early_stopping_patience_counter": 1
|
| 5665 |
}
|
| 5666 |
},
|
| 5667 |
"TrainerControl": {
|
|
|
|
| 5675 |
"attributes": {}
|
| 5676 |
}
|
| 5677 |
},
|
| 5678 |
+
"total_flos": 7.391104716421202e+17,
|
| 5679 |
"train_batch_size": 4,
|
| 5680 |
"trial_name": null,
|
| 5681 |
"trial_params": null
|