step 10400
Browse files- README.md +0 -2
- model.safetensors +1 -1
- trainer_state.json +1756 -4
README.md
CHANGED
|
@@ -12,8 +12,6 @@ license: apache-2.0
|
|
| 12 |
|
| 13 |
# BEE-spoke-data/bert-plus-L8-v1.0-allNLI_matryoshka
|
| 14 |
|
| 15 |
-
> if this message is here, this model card is a copy of the one from the _base model_ and not the actual model card.
|
| 16 |
-
|
| 17 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
| 18 |
|
| 19 |
- this was finetuned at 512 ctx (allNLI is all short-ctx examples) but the base model supports 4096
|
|
|
|
| 12 |
|
| 13 |
# BEE-spoke-data/bert-plus-L8-v1.0-allNLI_matryoshka
|
| 14 |
|
|
|
|
|
|
|
| 15 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
| 16 |
|
| 17 |
- this was finetuned at 512 ctx (allNLI is all short-ctx examples) but the base model supports 4096
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 352324400
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1395c5c36e20564cc37941ccb71e32f9fbf1039cc0aa1f5550d04df07dcab4f0
|
| 3 |
size 352324400
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "checkpoints/BEE-spoke-data-bert-plus-L8-v1.0-allNLI_matryoshka-synthetic-text-similarity-Mar-07_22-56/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 300,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -13427,6 +13427,1758 @@
|
|
| 13427 |
"learning_rate": 5.7839721254355405e-06,
|
| 13428 |
"loss": 0.0143,
|
| 13429 |
"step": 9200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13430 |
}
|
| 13431 |
],
|
| 13432 |
"logging_steps": 5,
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.9605806103224412,
|
| 3 |
+
"best_model_checkpoint": "checkpoints/BEE-spoke-data-bert-plus-L8-v1.0-allNLI_matryoshka-synthetic-text-similarity-Mar-07_22-56/checkpoint-9600",
|
| 4 |
+
"epoch": 0.8361809045226131,
|
| 5 |
"eval_steps": 300,
|
| 6 |
+
"global_step": 10400,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 13427 |
"learning_rate": 5.7839721254355405e-06,
|
| 13428 |
"loss": 0.0143,
|
| 13429 |
"step": 9200
|
| 13430 |
+
},
|
| 13431 |
+
{
|
| 13432 |
+
"epoch": 0.74,
|
| 13433 |
+
"grad_norm": 0.21646694839000702,
|
| 13434 |
+
"learning_rate": 5.7750379701599224e-06,
|
| 13435 |
+
"loss": 0.0117,
|
| 13436 |
+
"step": 9205
|
| 13437 |
+
},
|
| 13438 |
+
{
|
| 13439 |
+
"epoch": 0.74,
|
| 13440 |
+
"grad_norm": 0.2780154347419739,
|
| 13441 |
+
"learning_rate": 5.766103814884303e-06,
|
| 13442 |
+
"loss": 0.0069,
|
| 13443 |
+
"step": 9210
|
| 13444 |
+
},
|
| 13445 |
+
{
|
| 13446 |
+
"epoch": 0.74,
|
| 13447 |
+
"grad_norm": 0.38007307052612305,
|
| 13448 |
+
"learning_rate": 5.7571696596086845e-06,
|
| 13449 |
+
"loss": 0.0071,
|
| 13450 |
+
"step": 9215
|
| 13451 |
+
},
|
| 13452 |
+
{
|
| 13453 |
+
"epoch": 0.74,
|
| 13454 |
+
"grad_norm": 0.23504765331745148,
|
| 13455 |
+
"learning_rate": 5.748235504333066e-06,
|
| 13456 |
+
"loss": 0.0073,
|
| 13457 |
+
"step": 9220
|
| 13458 |
+
},
|
| 13459 |
+
{
|
| 13460 |
+
"epoch": 0.74,
|
| 13461 |
+
"grad_norm": 0.3849126100540161,
|
| 13462 |
+
"learning_rate": 5.739301349057447e-06,
|
| 13463 |
+
"loss": 0.0062,
|
| 13464 |
+
"step": 9225
|
| 13465 |
+
},
|
| 13466 |
+
{
|
| 13467 |
+
"epoch": 0.74,
|
| 13468 |
+
"grad_norm": 0.39096078276634216,
|
| 13469 |
+
"learning_rate": 5.7303671937818285e-06,
|
| 13470 |
+
"loss": 0.0081,
|
| 13471 |
+
"step": 9230
|
| 13472 |
+
},
|
| 13473 |
+
{
|
| 13474 |
+
"epoch": 0.74,
|
| 13475 |
+
"grad_norm": 0.35881391167640686,
|
| 13476 |
+
"learning_rate": 5.72143303850621e-06,
|
| 13477 |
+
"loss": 0.0073,
|
| 13478 |
+
"step": 9235
|
| 13479 |
+
},
|
| 13480 |
+
{
|
| 13481 |
+
"epoch": 0.74,
|
| 13482 |
+
"grad_norm": 0.2720240652561188,
|
| 13483 |
+
"learning_rate": 5.7124988832305905e-06,
|
| 13484 |
+
"loss": 0.006,
|
| 13485 |
+
"step": 9240
|
| 13486 |
+
},
|
| 13487 |
+
{
|
| 13488 |
+
"epoch": 0.74,
|
| 13489 |
+
"grad_norm": 0.17227132618427277,
|
| 13490 |
+
"learning_rate": 5.7035647279549724e-06,
|
| 13491 |
+
"loss": 0.0073,
|
| 13492 |
+
"step": 9245
|
| 13493 |
+
},
|
| 13494 |
+
{
|
| 13495 |
+
"epoch": 0.74,
|
| 13496 |
+
"grad_norm": 0.4603612422943115,
|
| 13497 |
+
"learning_rate": 5.6946305726793535e-06,
|
| 13498 |
+
"loss": 0.0116,
|
| 13499 |
+
"step": 9250
|
| 13500 |
+
},
|
| 13501 |
+
{
|
| 13502 |
+
"epoch": 0.74,
|
| 13503 |
+
"grad_norm": 0.6847078204154968,
|
| 13504 |
+
"learning_rate": 5.685696417403735e-06,
|
| 13505 |
+
"loss": 0.0075,
|
| 13506 |
+
"step": 9255
|
| 13507 |
+
},
|
| 13508 |
+
{
|
| 13509 |
+
"epoch": 0.74,
|
| 13510 |
+
"grad_norm": 0.41257572174072266,
|
| 13511 |
+
"learning_rate": 5.676762262128116e-06,
|
| 13512 |
+
"loss": 0.0103,
|
| 13513 |
+
"step": 9260
|
| 13514 |
+
},
|
| 13515 |
+
{
|
| 13516 |
+
"epoch": 0.74,
|
| 13517 |
+
"grad_norm": 0.402539998292923,
|
| 13518 |
+
"learning_rate": 5.6678281068524974e-06,
|
| 13519 |
+
"loss": 0.0063,
|
| 13520 |
+
"step": 9265
|
| 13521 |
+
},
|
| 13522 |
+
{
|
| 13523 |
+
"epoch": 0.75,
|
| 13524 |
+
"grad_norm": 0.4992349147796631,
|
| 13525 |
+
"learning_rate": 5.658893951576879e-06,
|
| 13526 |
+
"loss": 0.0103,
|
| 13527 |
+
"step": 9270
|
| 13528 |
+
},
|
| 13529 |
+
{
|
| 13530 |
+
"epoch": 0.75,
|
| 13531 |
+
"grad_norm": 0.47921329736709595,
|
| 13532 |
+
"learning_rate": 5.6499597963012595e-06,
|
| 13533 |
+
"loss": 0.0082,
|
| 13534 |
+
"step": 9275
|
| 13535 |
+
},
|
| 13536 |
+
{
|
| 13537 |
+
"epoch": 0.75,
|
| 13538 |
+
"grad_norm": 0.2835437059402466,
|
| 13539 |
+
"learning_rate": 5.641025641025641e-06,
|
| 13540 |
+
"loss": 0.0085,
|
| 13541 |
+
"step": 9280
|
| 13542 |
+
},
|
| 13543 |
+
{
|
| 13544 |
+
"epoch": 0.75,
|
| 13545 |
+
"grad_norm": 0.23391133546829224,
|
| 13546 |
+
"learning_rate": 5.632091485750023e-06,
|
| 13547 |
+
"loss": 0.0078,
|
| 13548 |
+
"step": 9285
|
| 13549 |
+
},
|
| 13550 |
+
{
|
| 13551 |
+
"epoch": 0.75,
|
| 13552 |
+
"grad_norm": 0.3029947876930237,
|
| 13553 |
+
"learning_rate": 5.6231573304744035e-06,
|
| 13554 |
+
"loss": 0.0145,
|
| 13555 |
+
"step": 9290
|
| 13556 |
+
},
|
| 13557 |
+
{
|
| 13558 |
+
"epoch": 0.75,
|
| 13559 |
+
"grad_norm": 0.26060959696769714,
|
| 13560 |
+
"learning_rate": 5.614223175198785e-06,
|
| 13561 |
+
"loss": 0.0061,
|
| 13562 |
+
"step": 9295
|
| 13563 |
+
},
|
| 13564 |
+
{
|
| 13565 |
+
"epoch": 0.75,
|
| 13566 |
+
"grad_norm": 0.2901848554611206,
|
| 13567 |
+
"learning_rate": 5.605289019923167e-06,
|
| 13568 |
+
"loss": 0.0083,
|
| 13569 |
+
"step": 9300
|
| 13570 |
+
},
|
| 13571 |
+
{
|
| 13572 |
+
"epoch": 0.75,
|
| 13573 |
+
"eval_loss": 0.009584016166627407,
|
| 13574 |
+
"eval_pearson_cosine": 0.9615300255520981,
|
| 13575 |
+
"eval_pearson_dot": 0.9604189728759469,
|
| 13576 |
+
"eval_pearson_euclidean": 0.9517997399844298,
|
| 13577 |
+
"eval_pearson_manhattan": 0.9500464009872014,
|
| 13578 |
+
"eval_pearson_max": 0.9615300255520981,
|
| 13579 |
+
"eval_runtime": 425.5009,
|
| 13580 |
+
"eval_samples_per_second": 1.175,
|
| 13581 |
+
"eval_spearman_cosine": 0.9606140184560736,
|
| 13582 |
+
"eval_spearman_dot": 0.9573194772779091,
|
| 13583 |
+
"eval_spearman_euclidean": 0.9593890535562142,
|
| 13584 |
+
"eval_spearman_manhattan": 0.9579064236256946,
|
| 13585 |
+
"eval_spearman_max": 0.9606140184560736,
|
| 13586 |
+
"eval_steps_per_second": 1.175,
|
| 13587 |
+
"step": 9300
|
| 13588 |
+
},
|
| 13589 |
+
{
|
| 13590 |
+
"epoch": 0.75,
|
| 13591 |
+
"grad_norm": 0.4470210373401642,
|
| 13592 |
+
"learning_rate": 5.596354864647548e-06,
|
| 13593 |
+
"loss": 0.0129,
|
| 13594 |
+
"step": 9305
|
| 13595 |
+
},
|
| 13596 |
+
{
|
| 13597 |
+
"epoch": 0.75,
|
| 13598 |
+
"grad_norm": 0.4185398519039154,
|
| 13599 |
+
"learning_rate": 5.587420709371929e-06,
|
| 13600 |
+
"loss": 0.0081,
|
| 13601 |
+
"step": 9310
|
| 13602 |
+
},
|
| 13603 |
+
{
|
| 13604 |
+
"epoch": 0.75,
|
| 13605 |
+
"grad_norm": 0.6620985269546509,
|
| 13606 |
+
"learning_rate": 5.57848655409631e-06,
|
| 13607 |
+
"loss": 0.0092,
|
| 13608 |
+
"step": 9315
|
| 13609 |
+
},
|
| 13610 |
+
{
|
| 13611 |
+
"epoch": 0.75,
|
| 13612 |
+
"grad_norm": 0.5238575339317322,
|
| 13613 |
+
"learning_rate": 5.569552398820692e-06,
|
| 13614 |
+
"loss": 0.0106,
|
| 13615 |
+
"step": 9320
|
| 13616 |
+
},
|
| 13617 |
+
{
|
| 13618 |
+
"epoch": 0.75,
|
| 13619 |
+
"grad_norm": 0.8626998662948608,
|
| 13620 |
+
"learning_rate": 5.5606182435450724e-06,
|
| 13621 |
+
"loss": 0.0149,
|
| 13622 |
+
"step": 9325
|
| 13623 |
+
},
|
| 13624 |
+
{
|
| 13625 |
+
"epoch": 0.75,
|
| 13626 |
+
"grad_norm": 0.312152624130249,
|
| 13627 |
+
"learning_rate": 5.551684088269454e-06,
|
| 13628 |
+
"loss": 0.0085,
|
| 13629 |
+
"step": 9330
|
| 13630 |
+
},
|
| 13631 |
+
{
|
| 13632 |
+
"epoch": 0.75,
|
| 13633 |
+
"grad_norm": 0.2844025194644928,
|
| 13634 |
+
"learning_rate": 5.542749932993836e-06,
|
| 13635 |
+
"loss": 0.01,
|
| 13636 |
+
"step": 9335
|
| 13637 |
+
},
|
| 13638 |
+
{
|
| 13639 |
+
"epoch": 0.75,
|
| 13640 |
+
"grad_norm": 0.4623136520385742,
|
| 13641 |
+
"learning_rate": 5.533815777718216e-06,
|
| 13642 |
+
"loss": 0.0082,
|
| 13643 |
+
"step": 9340
|
| 13644 |
+
},
|
| 13645 |
+
{
|
| 13646 |
+
"epoch": 0.75,
|
| 13647 |
+
"grad_norm": 0.34604260325431824,
|
| 13648 |
+
"learning_rate": 5.524881622442598e-06,
|
| 13649 |
+
"loss": 0.0105,
|
| 13650 |
+
"step": 9345
|
| 13651 |
+
},
|
| 13652 |
+
{
|
| 13653 |
+
"epoch": 0.75,
|
| 13654 |
+
"grad_norm": 0.5630137920379639,
|
| 13655 |
+
"learning_rate": 5.51594746716698e-06,
|
| 13656 |
+
"loss": 0.008,
|
| 13657 |
+
"step": 9350
|
| 13658 |
+
},
|
| 13659 |
+
{
|
| 13660 |
+
"epoch": 0.75,
|
| 13661 |
+
"grad_norm": 0.22408334910869598,
|
| 13662 |
+
"learning_rate": 5.507013311891361e-06,
|
| 13663 |
+
"loss": 0.0086,
|
| 13664 |
+
"step": 9355
|
| 13665 |
+
},
|
| 13666 |
+
{
|
| 13667 |
+
"epoch": 0.75,
|
| 13668 |
+
"grad_norm": 0.3562946915626526,
|
| 13669 |
+
"learning_rate": 5.498079156615742e-06,
|
| 13670 |
+
"loss": 0.0077,
|
| 13671 |
+
"step": 9360
|
| 13672 |
+
},
|
| 13673 |
+
{
|
| 13674 |
+
"epoch": 0.75,
|
| 13675 |
+
"grad_norm": 0.6519134640693665,
|
| 13676 |
+
"learning_rate": 5.489145001340124e-06,
|
| 13677 |
+
"loss": 0.0273,
|
| 13678 |
+
"step": 9365
|
| 13679 |
+
},
|
| 13680 |
+
{
|
| 13681 |
+
"epoch": 0.75,
|
| 13682 |
+
"grad_norm": 0.3609578609466553,
|
| 13683 |
+
"learning_rate": 5.480210846064505e-06,
|
| 13684 |
+
"loss": 0.008,
|
| 13685 |
+
"step": 9370
|
| 13686 |
+
},
|
| 13687 |
+
{
|
| 13688 |
+
"epoch": 0.75,
|
| 13689 |
+
"grad_norm": 0.4362548291683197,
|
| 13690 |
+
"learning_rate": 5.471276690788886e-06,
|
| 13691 |
+
"loss": 0.0156,
|
| 13692 |
+
"step": 9375
|
| 13693 |
+
},
|
| 13694 |
+
{
|
| 13695 |
+
"epoch": 0.75,
|
| 13696 |
+
"grad_norm": 0.51788729429245,
|
| 13697 |
+
"learning_rate": 5.462342535513267e-06,
|
| 13698 |
+
"loss": 0.0143,
|
| 13699 |
+
"step": 9380
|
| 13700 |
+
},
|
| 13701 |
+
{
|
| 13702 |
+
"epoch": 0.75,
|
| 13703 |
+
"grad_norm": 0.3049197793006897,
|
| 13704 |
+
"learning_rate": 5.453408380237649e-06,
|
| 13705 |
+
"loss": 0.0084,
|
| 13706 |
+
"step": 9385
|
| 13707 |
+
},
|
| 13708 |
+
{
|
| 13709 |
+
"epoch": 0.75,
|
| 13710 |
+
"grad_norm": 0.27513086795806885,
|
| 13711 |
+
"learning_rate": 5.444474224962031e-06,
|
| 13712 |
+
"loss": 0.0089,
|
| 13713 |
+
"step": 9390
|
| 13714 |
+
},
|
| 13715 |
+
{
|
| 13716 |
+
"epoch": 0.76,
|
| 13717 |
+
"grad_norm": 0.5995715260505676,
|
| 13718 |
+
"learning_rate": 5.435540069686411e-06,
|
| 13719 |
+
"loss": 0.0176,
|
| 13720 |
+
"step": 9395
|
| 13721 |
+
},
|
| 13722 |
+
{
|
| 13723 |
+
"epoch": 0.76,
|
| 13724 |
+
"grad_norm": 0.4572034180164337,
|
| 13725 |
+
"learning_rate": 5.426605914410793e-06,
|
| 13726 |
+
"loss": 0.0079,
|
| 13727 |
+
"step": 9400
|
| 13728 |
+
},
|
| 13729 |
+
{
|
| 13730 |
+
"epoch": 0.76,
|
| 13731 |
+
"grad_norm": 0.38861045241355896,
|
| 13732 |
+
"learning_rate": 5.417671759135175e-06,
|
| 13733 |
+
"loss": 0.0093,
|
| 13734 |
+
"step": 9405
|
| 13735 |
+
},
|
| 13736 |
+
{
|
| 13737 |
+
"epoch": 0.76,
|
| 13738 |
+
"grad_norm": 0.2802937924861908,
|
| 13739 |
+
"learning_rate": 5.408737603859555e-06,
|
| 13740 |
+
"loss": 0.0099,
|
| 13741 |
+
"step": 9410
|
| 13742 |
+
},
|
| 13743 |
+
{
|
| 13744 |
+
"epoch": 0.76,
|
| 13745 |
+
"grad_norm": 0.2714308500289917,
|
| 13746 |
+
"learning_rate": 5.399803448583937e-06,
|
| 13747 |
+
"loss": 0.0073,
|
| 13748 |
+
"step": 9415
|
| 13749 |
+
},
|
| 13750 |
+
{
|
| 13751 |
+
"epoch": 0.76,
|
| 13752 |
+
"grad_norm": 0.4080353379249573,
|
| 13753 |
+
"learning_rate": 5.390869293308318e-06,
|
| 13754 |
+
"loss": 0.0069,
|
| 13755 |
+
"step": 9420
|
| 13756 |
+
},
|
| 13757 |
+
{
|
| 13758 |
+
"epoch": 0.76,
|
| 13759 |
+
"grad_norm": 0.22111305594444275,
|
| 13760 |
+
"learning_rate": 5.381935138032699e-06,
|
| 13761 |
+
"loss": 0.0058,
|
| 13762 |
+
"step": 9425
|
| 13763 |
+
},
|
| 13764 |
+
{
|
| 13765 |
+
"epoch": 0.76,
|
| 13766 |
+
"grad_norm": 0.7781515717506409,
|
| 13767 |
+
"learning_rate": 5.37300098275708e-06,
|
| 13768 |
+
"loss": 0.0066,
|
| 13769 |
+
"step": 9430
|
| 13770 |
+
},
|
| 13771 |
+
{
|
| 13772 |
+
"epoch": 0.76,
|
| 13773 |
+
"grad_norm": 0.3610839545726776,
|
| 13774 |
+
"learning_rate": 5.364066827481462e-06,
|
| 13775 |
+
"loss": 0.0088,
|
| 13776 |
+
"step": 9435
|
| 13777 |
+
},
|
| 13778 |
+
{
|
| 13779 |
+
"epoch": 0.76,
|
| 13780 |
+
"grad_norm": 0.3884970247745514,
|
| 13781 |
+
"learning_rate": 5.355132672205844e-06,
|
| 13782 |
+
"loss": 0.0094,
|
| 13783 |
+
"step": 9440
|
| 13784 |
+
},
|
| 13785 |
+
{
|
| 13786 |
+
"epoch": 0.76,
|
| 13787 |
+
"grad_norm": 0.5556257963180542,
|
| 13788 |
+
"learning_rate": 5.346198516930224e-06,
|
| 13789 |
+
"loss": 0.0105,
|
| 13790 |
+
"step": 9445
|
| 13791 |
+
},
|
| 13792 |
+
{
|
| 13793 |
+
"epoch": 0.76,
|
| 13794 |
+
"grad_norm": 0.5632768273353577,
|
| 13795 |
+
"learning_rate": 5.337264361654606e-06,
|
| 13796 |
+
"loss": 0.0091,
|
| 13797 |
+
"step": 9450
|
| 13798 |
+
},
|
| 13799 |
+
{
|
| 13800 |
+
"epoch": 0.76,
|
| 13801 |
+
"grad_norm": 0.6423171758651733,
|
| 13802 |
+
"learning_rate": 5.328330206378988e-06,
|
| 13803 |
+
"loss": 0.0079,
|
| 13804 |
+
"step": 9455
|
| 13805 |
+
},
|
| 13806 |
+
{
|
| 13807 |
+
"epoch": 0.76,
|
| 13808 |
+
"grad_norm": 0.5519477725028992,
|
| 13809 |
+
"learning_rate": 5.319396051103368e-06,
|
| 13810 |
+
"loss": 0.0192,
|
| 13811 |
+
"step": 9460
|
| 13812 |
+
},
|
| 13813 |
+
{
|
| 13814 |
+
"epoch": 0.76,
|
| 13815 |
+
"grad_norm": 0.5822392106056213,
|
| 13816 |
+
"learning_rate": 5.31046189582775e-06,
|
| 13817 |
+
"loss": 0.0123,
|
| 13818 |
+
"step": 9465
|
| 13819 |
+
},
|
| 13820 |
+
{
|
| 13821 |
+
"epoch": 0.76,
|
| 13822 |
+
"grad_norm": 0.49340376257896423,
|
| 13823 |
+
"learning_rate": 5.301527740552132e-06,
|
| 13824 |
+
"loss": 0.0083,
|
| 13825 |
+
"step": 9470
|
| 13826 |
+
},
|
| 13827 |
+
{
|
| 13828 |
+
"epoch": 0.76,
|
| 13829 |
+
"grad_norm": 0.3505455553531647,
|
| 13830 |
+
"learning_rate": 5.292593585276512e-06,
|
| 13831 |
+
"loss": 0.011,
|
| 13832 |
+
"step": 9475
|
| 13833 |
+
},
|
| 13834 |
+
{
|
| 13835 |
+
"epoch": 0.76,
|
| 13836 |
+
"grad_norm": 0.7089307308197021,
|
| 13837 |
+
"learning_rate": 5.283659430000894e-06,
|
| 13838 |
+
"loss": 0.0079,
|
| 13839 |
+
"step": 9480
|
| 13840 |
+
},
|
| 13841 |
+
{
|
| 13842 |
+
"epoch": 0.76,
|
| 13843 |
+
"grad_norm": 0.791226863861084,
|
| 13844 |
+
"learning_rate": 5.274725274725275e-06,
|
| 13845 |
+
"loss": 0.0124,
|
| 13846 |
+
"step": 9485
|
| 13847 |
+
},
|
| 13848 |
+
{
|
| 13849 |
+
"epoch": 0.76,
|
| 13850 |
+
"grad_norm": 0.38624152541160583,
|
| 13851 |
+
"learning_rate": 5.265791119449657e-06,
|
| 13852 |
+
"loss": 0.0076,
|
| 13853 |
+
"step": 9490
|
| 13854 |
+
},
|
| 13855 |
+
{
|
| 13856 |
+
"epoch": 0.76,
|
| 13857 |
+
"grad_norm": 0.2802993953227997,
|
| 13858 |
+
"learning_rate": 5.256856964174037e-06,
|
| 13859 |
+
"loss": 0.0183,
|
| 13860 |
+
"step": 9495
|
| 13861 |
+
},
|
| 13862 |
+
{
|
| 13863 |
+
"epoch": 0.76,
|
| 13864 |
+
"grad_norm": 0.4529344439506531,
|
| 13865 |
+
"learning_rate": 5.247922808898419e-06,
|
| 13866 |
+
"loss": 0.0112,
|
| 13867 |
+
"step": 9500
|
| 13868 |
+
},
|
| 13869 |
+
{
|
| 13870 |
+
"epoch": 0.76,
|
| 13871 |
+
"grad_norm": 1.167479395866394,
|
| 13872 |
+
"learning_rate": 5.238988653622801e-06,
|
| 13873 |
+
"loss": 0.0128,
|
| 13874 |
+
"step": 9505
|
| 13875 |
+
},
|
| 13876 |
+
{
|
| 13877 |
+
"epoch": 0.76,
|
| 13878 |
+
"grad_norm": 0.2423386126756668,
|
| 13879 |
+
"learning_rate": 5.230054498347181e-06,
|
| 13880 |
+
"loss": 0.0093,
|
| 13881 |
+
"step": 9510
|
| 13882 |
+
},
|
| 13883 |
+
{
|
| 13884 |
+
"epoch": 0.77,
|
| 13885 |
+
"grad_norm": 0.3792467713356018,
|
| 13886 |
+
"learning_rate": 5.221120343071563e-06,
|
| 13887 |
+
"loss": 0.0078,
|
| 13888 |
+
"step": 9515
|
| 13889 |
+
},
|
| 13890 |
+
{
|
| 13891 |
+
"epoch": 0.77,
|
| 13892 |
+
"grad_norm": 0.47463178634643555,
|
| 13893 |
+
"learning_rate": 5.212186187795945e-06,
|
| 13894 |
+
"loss": 0.0116,
|
| 13895 |
+
"step": 9520
|
| 13896 |
+
},
|
| 13897 |
+
{
|
| 13898 |
+
"epoch": 0.77,
|
| 13899 |
+
"grad_norm": 0.9293251037597656,
|
| 13900 |
+
"learning_rate": 5.203252032520326e-06,
|
| 13901 |
+
"loss": 0.0107,
|
| 13902 |
+
"step": 9525
|
| 13903 |
+
},
|
| 13904 |
+
{
|
| 13905 |
+
"epoch": 0.77,
|
| 13906 |
+
"grad_norm": 41.848262786865234,
|
| 13907 |
+
"learning_rate": 5.194317877244707e-06,
|
| 13908 |
+
"loss": 0.0182,
|
| 13909 |
+
"step": 9530
|
| 13910 |
+
},
|
| 13911 |
+
{
|
| 13912 |
+
"epoch": 0.77,
|
| 13913 |
+
"grad_norm": 0.34961816668510437,
|
| 13914 |
+
"learning_rate": 5.185383721969088e-06,
|
| 13915 |
+
"loss": 0.007,
|
| 13916 |
+
"step": 9535
|
| 13917 |
+
},
|
| 13918 |
+
{
|
| 13919 |
+
"epoch": 0.77,
|
| 13920 |
+
"grad_norm": 1.822770595550537,
|
| 13921 |
+
"learning_rate": 5.17644956669347e-06,
|
| 13922 |
+
"loss": 0.0151,
|
| 13923 |
+
"step": 9540
|
| 13924 |
+
},
|
| 13925 |
+
{
|
| 13926 |
+
"epoch": 0.77,
|
| 13927 |
+
"grad_norm": 0.3188020884990692,
|
| 13928 |
+
"learning_rate": 5.167515411417851e-06,
|
| 13929 |
+
"loss": 0.0067,
|
| 13930 |
+
"step": 9545
|
| 13931 |
+
},
|
| 13932 |
+
{
|
| 13933 |
+
"epoch": 0.77,
|
| 13934 |
+
"grad_norm": 0.43374982476234436,
|
| 13935 |
+
"learning_rate": 5.158581256142232e-06,
|
| 13936 |
+
"loss": 0.0084,
|
| 13937 |
+
"step": 9550
|
| 13938 |
+
},
|
| 13939 |
+
{
|
| 13940 |
+
"epoch": 0.77,
|
| 13941 |
+
"grad_norm": 0.4171868562698364,
|
| 13942 |
+
"learning_rate": 5.149647100866614e-06,
|
| 13943 |
+
"loss": 0.0073,
|
| 13944 |
+
"step": 9555
|
| 13945 |
+
},
|
| 13946 |
+
{
|
| 13947 |
+
"epoch": 0.77,
|
| 13948 |
+
"grad_norm": 0.5583686232566833,
|
| 13949 |
+
"learning_rate": 5.140712945590994e-06,
|
| 13950 |
+
"loss": 0.0092,
|
| 13951 |
+
"step": 9560
|
| 13952 |
+
},
|
| 13953 |
+
{
|
| 13954 |
+
"epoch": 0.77,
|
| 13955 |
+
"grad_norm": 0.24175512790679932,
|
| 13956 |
+
"learning_rate": 5.131778790315376e-06,
|
| 13957 |
+
"loss": 0.0081,
|
| 13958 |
+
"step": 9565
|
| 13959 |
+
},
|
| 13960 |
+
{
|
| 13961 |
+
"epoch": 0.77,
|
| 13962 |
+
"grad_norm": 0.3415161371231079,
|
| 13963 |
+
"learning_rate": 5.122844635039758e-06,
|
| 13964 |
+
"loss": 0.0109,
|
| 13965 |
+
"step": 9570
|
| 13966 |
+
},
|
| 13967 |
+
{
|
| 13968 |
+
"epoch": 0.77,
|
| 13969 |
+
"grad_norm": 0.5684280395507812,
|
| 13970 |
+
"learning_rate": 5.11391047976414e-06,
|
| 13971 |
+
"loss": 0.0072,
|
| 13972 |
+
"step": 9575
|
| 13973 |
+
},
|
| 13974 |
+
{
|
| 13975 |
+
"epoch": 0.77,
|
| 13976 |
+
"grad_norm": 0.9785246849060059,
|
| 13977 |
+
"learning_rate": 5.10497632448852e-06,
|
| 13978 |
+
"loss": 0.0098,
|
| 13979 |
+
"step": 9580
|
| 13980 |
+
},
|
| 13981 |
+
{
|
| 13982 |
+
"epoch": 0.77,
|
| 13983 |
+
"grad_norm": 0.2952157258987427,
|
| 13984 |
+
"learning_rate": 5.096042169212902e-06,
|
| 13985 |
+
"loss": 0.008,
|
| 13986 |
+
"step": 9585
|
| 13987 |
+
},
|
| 13988 |
+
{
|
| 13989 |
+
"epoch": 0.77,
|
| 13990 |
+
"grad_norm": 0.20075848698616028,
|
| 13991 |
+
"learning_rate": 5.087108013937283e-06,
|
| 13992 |
+
"loss": 0.0074,
|
| 13993 |
+
"step": 9590
|
| 13994 |
+
},
|
| 13995 |
+
{
|
| 13996 |
+
"epoch": 0.77,
|
| 13997 |
+
"grad_norm": 0.3542312681674957,
|
| 13998 |
+
"learning_rate": 5.078173858661664e-06,
|
| 13999 |
+
"loss": 0.0109,
|
| 14000 |
+
"step": 9595
|
| 14001 |
+
},
|
| 14002 |
+
{
|
| 14003 |
+
"epoch": 0.77,
|
| 14004 |
+
"grad_norm": 0.38026684522628784,
|
| 14005 |
+
"learning_rate": 5.069239703386045e-06,
|
| 14006 |
+
"loss": 0.0069,
|
| 14007 |
+
"step": 9600
|
| 14008 |
+
},
|
| 14009 |
+
{
|
| 14010 |
+
"epoch": 0.77,
|
| 14011 |
+
"eval_loss": 0.008200183510780334,
|
| 14012 |
+
"eval_pearson_cosine": 0.9642878383068713,
|
| 14013 |
+
"eval_pearson_dot": 0.9627028886298191,
|
| 14014 |
+
"eval_pearson_euclidean": 0.9523746354427236,
|
| 14015 |
+
"eval_pearson_manhattan": 0.9507404567548948,
|
| 14016 |
+
"eval_pearson_max": 0.9642878383068713,
|
| 14017 |
+
"eval_runtime": 425.6137,
|
| 14018 |
+
"eval_samples_per_second": 1.175,
|
| 14019 |
+
"eval_spearman_cosine": 0.9605806103224412,
|
| 14020 |
+
"eval_spearman_dot": 0.955501806007224,
|
| 14021 |
+
"eval_spearman_euclidean": 0.9597567350269401,
|
| 14022 |
+
"eval_spearman_manhattan": 0.9581544886179545,
|
| 14023 |
+
"eval_spearman_max": 0.9605806103224412,
|
| 14024 |
+
"eval_steps_per_second": 1.175,
|
| 14025 |
+
"step": 9600
|
| 14026 |
+
},
|
| 14027 |
+
{
|
| 14028 |
+
"epoch": 0.77,
|
| 14029 |
+
"grad_norm": 0.37074190378189087,
|
| 14030 |
+
"learning_rate": 5.060305548110427e-06,
|
| 14031 |
+
"loss": 0.0173,
|
| 14032 |
+
"step": 9605
|
| 14033 |
+
},
|
| 14034 |
+
{
|
| 14035 |
+
"epoch": 0.77,
|
| 14036 |
+
"grad_norm": 0.3430071175098419,
|
| 14037 |
+
"learning_rate": 5.051371392834808e-06,
|
| 14038 |
+
"loss": 0.0077,
|
| 14039 |
+
"step": 9610
|
| 14040 |
+
},
|
| 14041 |
+
{
|
| 14042 |
+
"epoch": 0.77,
|
| 14043 |
+
"grad_norm": 0.516249418258667,
|
| 14044 |
+
"learning_rate": 5.042437237559189e-06,
|
| 14045 |
+
"loss": 0.0059,
|
| 14046 |
+
"step": 9615
|
| 14047 |
+
},
|
| 14048 |
+
{
|
| 14049 |
+
"epoch": 0.77,
|
| 14050 |
+
"grad_norm": 0.3532971143722534,
|
| 14051 |
+
"learning_rate": 5.033503082283571e-06,
|
| 14052 |
+
"loss": 0.0113,
|
| 14053 |
+
"step": 9620
|
| 14054 |
+
},
|
| 14055 |
+
{
|
| 14056 |
+
"epoch": 0.77,
|
| 14057 |
+
"grad_norm": 0.4686454236507416,
|
| 14058 |
+
"learning_rate": 5.0245689270079526e-06,
|
| 14059 |
+
"loss": 0.0097,
|
| 14060 |
+
"step": 9625
|
| 14061 |
+
},
|
| 14062 |
+
{
|
| 14063 |
+
"epoch": 0.77,
|
| 14064 |
+
"grad_norm": 0.2624160051345825,
|
| 14065 |
+
"learning_rate": 5.015634771732333e-06,
|
| 14066 |
+
"loss": 0.0087,
|
| 14067 |
+
"step": 9630
|
| 14068 |
+
},
|
| 14069 |
+
{
|
| 14070 |
+
"epoch": 0.77,
|
| 14071 |
+
"grad_norm": 0.35835760831832886,
|
| 14072 |
+
"learning_rate": 5.006700616456715e-06,
|
| 14073 |
+
"loss": 0.0076,
|
| 14074 |
+
"step": 9635
|
| 14075 |
+
},
|
| 14076 |
+
{
|
| 14077 |
+
"epoch": 0.78,
|
| 14078 |
+
"grad_norm": 0.6053426265716553,
|
| 14079 |
+
"learning_rate": 4.997766461181096e-06,
|
| 14080 |
+
"loss": 0.0158,
|
| 14081 |
+
"step": 9640
|
| 14082 |
+
},
|
| 14083 |
+
{
|
| 14084 |
+
"epoch": 0.78,
|
| 14085 |
+
"grad_norm": 0.399104505777359,
|
| 14086 |
+
"learning_rate": 4.9888323059054776e-06,
|
| 14087 |
+
"loss": 0.0085,
|
| 14088 |
+
"step": 9645
|
| 14089 |
+
},
|
| 14090 |
+
{
|
| 14091 |
+
"epoch": 0.78,
|
| 14092 |
+
"grad_norm": 0.3674059212207794,
|
| 14093 |
+
"learning_rate": 4.979898150629859e-06,
|
| 14094 |
+
"loss": 0.0112,
|
| 14095 |
+
"step": 9650
|
| 14096 |
+
},
|
| 14097 |
+
{
|
| 14098 |
+
"epoch": 0.78,
|
| 14099 |
+
"grad_norm": 0.16110941767692566,
|
| 14100 |
+
"learning_rate": 4.97096399535424e-06,
|
| 14101 |
+
"loss": 0.0086,
|
| 14102 |
+
"step": 9655
|
| 14103 |
+
},
|
| 14104 |
+
{
|
| 14105 |
+
"epoch": 0.78,
|
| 14106 |
+
"grad_norm": 0.2616519331932068,
|
| 14107 |
+
"learning_rate": 4.962029840078621e-06,
|
| 14108 |
+
"loss": 0.009,
|
| 14109 |
+
"step": 9660
|
| 14110 |
+
},
|
| 14111 |
+
{
|
| 14112 |
+
"epoch": 0.78,
|
| 14113 |
+
"grad_norm": 0.6079025268554688,
|
| 14114 |
+
"learning_rate": 4.953095684803002e-06,
|
| 14115 |
+
"loss": 0.0082,
|
| 14116 |
+
"step": 9665
|
| 14117 |
+
},
|
| 14118 |
+
{
|
| 14119 |
+
"epoch": 0.78,
|
| 14120 |
+
"grad_norm": 0.30316200852394104,
|
| 14121 |
+
"learning_rate": 4.944161529527384e-06,
|
| 14122 |
+
"loss": 0.0063,
|
| 14123 |
+
"step": 9670
|
| 14124 |
+
},
|
| 14125 |
+
{
|
| 14126 |
+
"epoch": 0.78,
|
| 14127 |
+
"grad_norm": 0.36754128336906433,
|
| 14128 |
+
"learning_rate": 4.935227374251765e-06,
|
| 14129 |
+
"loss": 0.0073,
|
| 14130 |
+
"step": 9675
|
| 14131 |
+
},
|
| 14132 |
+
{
|
| 14133 |
+
"epoch": 0.78,
|
| 14134 |
+
"grad_norm": 0.6272875666618347,
|
| 14135 |
+
"learning_rate": 4.9262932189761465e-06,
|
| 14136 |
+
"loss": 0.0089,
|
| 14137 |
+
"step": 9680
|
| 14138 |
+
},
|
| 14139 |
+
{
|
| 14140 |
+
"epoch": 0.78,
|
| 14141 |
+
"grad_norm": 0.5529404878616333,
|
| 14142 |
+
"learning_rate": 4.9173590637005276e-06,
|
| 14143 |
+
"loss": 0.0095,
|
| 14144 |
+
"step": 9685
|
| 14145 |
+
},
|
| 14146 |
+
{
|
| 14147 |
+
"epoch": 0.78,
|
| 14148 |
+
"grad_norm": 0.2829398214817047,
|
| 14149 |
+
"learning_rate": 4.908424908424909e-06,
|
| 14150 |
+
"loss": 0.011,
|
| 14151 |
+
"step": 9690
|
| 14152 |
+
},
|
| 14153 |
+
{
|
| 14154 |
+
"epoch": 0.78,
|
| 14155 |
+
"grad_norm": 0.30931296944618225,
|
| 14156 |
+
"learning_rate": 4.8994907531492905e-06,
|
| 14157 |
+
"loss": 0.0103,
|
| 14158 |
+
"step": 9695
|
| 14159 |
+
},
|
| 14160 |
+
{
|
| 14161 |
+
"epoch": 0.78,
|
| 14162 |
+
"grad_norm": 0.6834177374839783,
|
| 14163 |
+
"learning_rate": 4.8905565978736715e-06,
|
| 14164 |
+
"loss": 0.0087,
|
| 14165 |
+
"step": 9700
|
| 14166 |
+
},
|
| 14167 |
+
{
|
| 14168 |
+
"epoch": 0.78,
|
| 14169 |
+
"grad_norm": 0.42869821190834045,
|
| 14170 |
+
"learning_rate": 4.8816224425980526e-06,
|
| 14171 |
+
"loss": 0.0088,
|
| 14172 |
+
"step": 9705
|
| 14173 |
+
},
|
| 14174 |
+
{
|
| 14175 |
+
"epoch": 0.78,
|
| 14176 |
+
"grad_norm": 0.3291241228580475,
|
| 14177 |
+
"learning_rate": 4.8726882873224344e-06,
|
| 14178 |
+
"loss": 0.0061,
|
| 14179 |
+
"step": 9710
|
| 14180 |
+
},
|
| 14181 |
+
{
|
| 14182 |
+
"epoch": 0.78,
|
| 14183 |
+
"grad_norm": 0.3477993607521057,
|
| 14184 |
+
"learning_rate": 4.8637541320468155e-06,
|
| 14185 |
+
"loss": 0.007,
|
| 14186 |
+
"step": 9715
|
| 14187 |
+
},
|
| 14188 |
+
{
|
| 14189 |
+
"epoch": 0.78,
|
| 14190 |
+
"grad_norm": 0.393031507730484,
|
| 14191 |
+
"learning_rate": 4.8548199767711965e-06,
|
| 14192 |
+
"loss": 0.0077,
|
| 14193 |
+
"step": 9720
|
| 14194 |
+
},
|
| 14195 |
+
{
|
| 14196 |
+
"epoch": 0.78,
|
| 14197 |
+
"grad_norm": 0.37427353858947754,
|
| 14198 |
+
"learning_rate": 4.8458858214955776e-06,
|
| 14199 |
+
"loss": 0.0086,
|
| 14200 |
+
"step": 9725
|
| 14201 |
+
},
|
| 14202 |
+
{
|
| 14203 |
+
"epoch": 0.78,
|
| 14204 |
+
"grad_norm": 0.4370558261871338,
|
| 14205 |
+
"learning_rate": 4.8369516662199594e-06,
|
| 14206 |
+
"loss": 0.0064,
|
| 14207 |
+
"step": 9730
|
| 14208 |
+
},
|
| 14209 |
+
{
|
| 14210 |
+
"epoch": 0.78,
|
| 14211 |
+
"grad_norm": 0.31545019149780273,
|
| 14212 |
+
"learning_rate": 4.8280175109443405e-06,
|
| 14213 |
+
"loss": 0.0059,
|
| 14214 |
+
"step": 9735
|
| 14215 |
+
},
|
| 14216 |
+
{
|
| 14217 |
+
"epoch": 0.78,
|
| 14218 |
+
"grad_norm": 0.7945960760116577,
|
| 14219 |
+
"learning_rate": 4.8190833556687215e-06,
|
| 14220 |
+
"loss": 0.021,
|
| 14221 |
+
"step": 9740
|
| 14222 |
+
},
|
| 14223 |
+
{
|
| 14224 |
+
"epoch": 0.78,
|
| 14225 |
+
"grad_norm": 0.29888418316841125,
|
| 14226 |
+
"learning_rate": 4.810149200393103e-06,
|
| 14227 |
+
"loss": 0.0062,
|
| 14228 |
+
"step": 9745
|
| 14229 |
+
},
|
| 14230 |
+
{
|
| 14231 |
+
"epoch": 0.78,
|
| 14232 |
+
"grad_norm": 3.3094396591186523,
|
| 14233 |
+
"learning_rate": 4.8012150451174844e-06,
|
| 14234 |
+
"loss": 0.0154,
|
| 14235 |
+
"step": 9750
|
| 14236 |
+
},
|
| 14237 |
+
{
|
| 14238 |
+
"epoch": 0.78,
|
| 14239 |
+
"grad_norm": 0.2018340677022934,
|
| 14240 |
+
"learning_rate": 4.792280889841866e-06,
|
| 14241 |
+
"loss": 0.0074,
|
| 14242 |
+
"step": 9755
|
| 14243 |
+
},
|
| 14244 |
+
{
|
| 14245 |
+
"epoch": 0.78,
|
| 14246 |
+
"grad_norm": 0.7358143329620361,
|
| 14247 |
+
"learning_rate": 4.783346734566247e-06,
|
| 14248 |
+
"loss": 0.0154,
|
| 14249 |
+
"step": 9760
|
| 14250 |
+
},
|
| 14251 |
+
{
|
| 14252 |
+
"epoch": 0.79,
|
| 14253 |
+
"grad_norm": 0.580872654914856,
|
| 14254 |
+
"learning_rate": 4.774412579290628e-06,
|
| 14255 |
+
"loss": 0.0111,
|
| 14256 |
+
"step": 9765
|
| 14257 |
+
},
|
| 14258 |
+
{
|
| 14259 |
+
"epoch": 0.79,
|
| 14260 |
+
"grad_norm": 0.3042278289794922,
|
| 14261 |
+
"learning_rate": 4.7654784240150095e-06,
|
| 14262 |
+
"loss": 0.0078,
|
| 14263 |
+
"step": 9770
|
| 14264 |
+
},
|
| 14265 |
+
{
|
| 14266 |
+
"epoch": 0.79,
|
| 14267 |
+
"grad_norm": 0.42176923155784607,
|
| 14268 |
+
"learning_rate": 4.7565442687393905e-06,
|
| 14269 |
+
"loss": 0.0062,
|
| 14270 |
+
"step": 9775
|
| 14271 |
+
},
|
| 14272 |
+
{
|
| 14273 |
+
"epoch": 0.79,
|
| 14274 |
+
"grad_norm": 0.30140987038612366,
|
| 14275 |
+
"learning_rate": 4.747610113463772e-06,
|
| 14276 |
+
"loss": 0.0063,
|
| 14277 |
+
"step": 9780
|
| 14278 |
+
},
|
| 14279 |
+
{
|
| 14280 |
+
"epoch": 0.79,
|
| 14281 |
+
"grad_norm": 0.5304137468338013,
|
| 14282 |
+
"learning_rate": 4.738675958188153e-06,
|
| 14283 |
+
"loss": 0.0097,
|
| 14284 |
+
"step": 9785
|
| 14285 |
+
},
|
| 14286 |
+
{
|
| 14287 |
+
"epoch": 0.79,
|
| 14288 |
+
"grad_norm": 0.36996015906333923,
|
| 14289 |
+
"learning_rate": 4.729741802912535e-06,
|
| 14290 |
+
"loss": 0.0097,
|
| 14291 |
+
"step": 9790
|
| 14292 |
+
},
|
| 14293 |
+
{
|
| 14294 |
+
"epoch": 0.79,
|
| 14295 |
+
"grad_norm": 0.6667109131813049,
|
| 14296 |
+
"learning_rate": 4.720807647636916e-06,
|
| 14297 |
+
"loss": 0.0075,
|
| 14298 |
+
"step": 9795
|
| 14299 |
+
},
|
| 14300 |
+
{
|
| 14301 |
+
"epoch": 0.79,
|
| 14302 |
+
"grad_norm": 0.2762182950973511,
|
| 14303 |
+
"learning_rate": 4.711873492361297e-06,
|
| 14304 |
+
"loss": 0.0064,
|
| 14305 |
+
"step": 9800
|
| 14306 |
+
},
|
| 14307 |
+
{
|
| 14308 |
+
"epoch": 0.79,
|
| 14309 |
+
"grad_norm": 0.3733229339122772,
|
| 14310 |
+
"learning_rate": 4.702939337085679e-06,
|
| 14311 |
+
"loss": 0.0054,
|
| 14312 |
+
"step": 9805
|
| 14313 |
+
},
|
| 14314 |
+
{
|
| 14315 |
+
"epoch": 0.79,
|
| 14316 |
+
"grad_norm": 0.3811498284339905,
|
| 14317 |
+
"learning_rate": 4.69400518181006e-06,
|
| 14318 |
+
"loss": 0.0062,
|
| 14319 |
+
"step": 9810
|
| 14320 |
+
},
|
| 14321 |
+
{
|
| 14322 |
+
"epoch": 0.79,
|
| 14323 |
+
"grad_norm": 0.5551919937133789,
|
| 14324 |
+
"learning_rate": 4.685071026534442e-06,
|
| 14325 |
+
"loss": 0.0089,
|
| 14326 |
+
"step": 9815
|
| 14327 |
+
},
|
| 14328 |
+
{
|
| 14329 |
+
"epoch": 0.79,
|
| 14330 |
+
"grad_norm": 0.5745194554328918,
|
| 14331 |
+
"learning_rate": 4.676136871258823e-06,
|
| 14332 |
+
"loss": 0.0097,
|
| 14333 |
+
"step": 9820
|
| 14334 |
+
},
|
| 14335 |
+
{
|
| 14336 |
+
"epoch": 0.79,
|
| 14337 |
+
"grad_norm": 0.2763228118419647,
|
| 14338 |
+
"learning_rate": 4.667202715983204e-06,
|
| 14339 |
+
"loss": 0.01,
|
| 14340 |
+
"step": 9825
|
| 14341 |
+
},
|
| 14342 |
+
{
|
| 14343 |
+
"epoch": 0.79,
|
| 14344 |
+
"grad_norm": 0.5403454899787903,
|
| 14345 |
+
"learning_rate": 4.658268560707585e-06,
|
| 14346 |
+
"loss": 0.0134,
|
| 14347 |
+
"step": 9830
|
| 14348 |
+
},
|
| 14349 |
+
{
|
| 14350 |
+
"epoch": 0.79,
|
| 14351 |
+
"grad_norm": 0.37194764614105225,
|
| 14352 |
+
"learning_rate": 4.649334405431966e-06,
|
| 14353 |
+
"loss": 0.0051,
|
| 14354 |
+
"step": 9835
|
| 14355 |
+
},
|
| 14356 |
+
{
|
| 14357 |
+
"epoch": 0.79,
|
| 14358 |
+
"grad_norm": 0.21742063760757446,
|
| 14359 |
+
"learning_rate": 4.640400250156348e-06,
|
| 14360 |
+
"loss": 0.0123,
|
| 14361 |
+
"step": 9840
|
| 14362 |
+
},
|
| 14363 |
+
{
|
| 14364 |
+
"epoch": 0.79,
|
| 14365 |
+
"grad_norm": 0.3551539182662964,
|
| 14366 |
+
"learning_rate": 4.631466094880729e-06,
|
| 14367 |
+
"loss": 0.0104,
|
| 14368 |
+
"step": 9845
|
| 14369 |
+
},
|
| 14370 |
+
{
|
| 14371 |
+
"epoch": 0.79,
|
| 14372 |
+
"grad_norm": 0.248150035738945,
|
| 14373 |
+
"learning_rate": 4.62253193960511e-06,
|
| 14374 |
+
"loss": 0.0054,
|
| 14375 |
+
"step": 9850
|
| 14376 |
+
},
|
| 14377 |
+
{
|
| 14378 |
+
"epoch": 0.79,
|
| 14379 |
+
"grad_norm": 0.6017441153526306,
|
| 14380 |
+
"learning_rate": 4.613597784329492e-06,
|
| 14381 |
+
"loss": 0.0115,
|
| 14382 |
+
"step": 9855
|
| 14383 |
+
},
|
| 14384 |
+
{
|
| 14385 |
+
"epoch": 0.79,
|
| 14386 |
+
"grad_norm": 0.46963444352149963,
|
| 14387 |
+
"learning_rate": 4.604663629053873e-06,
|
| 14388 |
+
"loss": 0.0087,
|
| 14389 |
+
"step": 9860
|
| 14390 |
+
},
|
| 14391 |
+
{
|
| 14392 |
+
"epoch": 0.79,
|
| 14393 |
+
"grad_norm": 0.26475605368614197,
|
| 14394 |
+
"learning_rate": 4.595729473778255e-06,
|
| 14395 |
+
"loss": 0.0064,
|
| 14396 |
+
"step": 9865
|
| 14397 |
+
},
|
| 14398 |
+
{
|
| 14399 |
+
"epoch": 0.79,
|
| 14400 |
+
"grad_norm": 0.3036366403102875,
|
| 14401 |
+
"learning_rate": 4.586795318502636e-06,
|
| 14402 |
+
"loss": 0.0094,
|
| 14403 |
+
"step": 9870
|
| 14404 |
+
},
|
| 14405 |
+
{
|
| 14406 |
+
"epoch": 0.79,
|
| 14407 |
+
"grad_norm": 0.4167456328868866,
|
| 14408 |
+
"learning_rate": 4.577861163227017e-06,
|
| 14409 |
+
"loss": 0.008,
|
| 14410 |
+
"step": 9875
|
| 14411 |
+
},
|
| 14412 |
+
{
|
| 14413 |
+
"epoch": 0.79,
|
| 14414 |
+
"grad_norm": 0.27215877175331116,
|
| 14415 |
+
"learning_rate": 4.568927007951398e-06,
|
| 14416 |
+
"loss": 0.0087,
|
| 14417 |
+
"step": 9880
|
| 14418 |
+
},
|
| 14419 |
+
{
|
| 14420 |
+
"epoch": 0.79,
|
| 14421 |
+
"grad_norm": 0.3947705924510956,
|
| 14422 |
+
"learning_rate": 4.55999285267578e-06,
|
| 14423 |
+
"loss": 0.0099,
|
| 14424 |
+
"step": 9885
|
| 14425 |
+
},
|
| 14426 |
+
{
|
| 14427 |
+
"epoch": 0.8,
|
| 14428 |
+
"grad_norm": 0.261850506067276,
|
| 14429 |
+
"learning_rate": 4.551058697400161e-06,
|
| 14430 |
+
"loss": 0.012,
|
| 14431 |
+
"step": 9890
|
| 14432 |
+
},
|
| 14433 |
+
{
|
| 14434 |
+
"epoch": 0.8,
|
| 14435 |
+
"grad_norm": 0.27852803468704224,
|
| 14436 |
+
"learning_rate": 4.542124542124542e-06,
|
| 14437 |
+
"loss": 0.0063,
|
| 14438 |
+
"step": 9895
|
| 14439 |
+
},
|
| 14440 |
+
{
|
| 14441 |
+
"epoch": 0.8,
|
| 14442 |
+
"grad_norm": 0.3846147060394287,
|
| 14443 |
+
"learning_rate": 4.533190386848923e-06,
|
| 14444 |
+
"loss": 0.0072,
|
| 14445 |
+
"step": 9900
|
| 14446 |
+
},
|
| 14447 |
+
{
|
| 14448 |
+
"epoch": 0.8,
|
| 14449 |
+
"eval_loss": 0.007968730293214321,
|
| 14450 |
+
"eval_pearson_cosine": 0.9658888665742654,
|
| 14451 |
+
"eval_pearson_dot": 0.9627392628207877,
|
| 14452 |
+
"eval_pearson_euclidean": 0.9531911193892233,
|
| 14453 |
+
"eval_pearson_manhattan": 0.9513587882321657,
|
| 14454 |
+
"eval_pearson_max": 0.9658888665742654,
|
| 14455 |
+
"eval_runtime": 425.901,
|
| 14456 |
+
"eval_samples_per_second": 1.174,
|
| 14457 |
+
"eval_spearman_cosine": 0.9624986019944078,
|
| 14458 |
+
"eval_spearman_dot": 0.955504398017592,
|
| 14459 |
+
"eval_spearman_euclidean": 0.9630360121440484,
|
| 14460 |
+
"eval_spearman_manhattan": 0.9610320041280165,
|
| 14461 |
+
"eval_spearman_max": 0.9630360121440484,
|
| 14462 |
+
"eval_steps_per_second": 1.174,
|
| 14463 |
+
"step": 9900
|
| 14464 |
+
},
|
| 14465 |
+
{
|
| 14466 |
+
"epoch": 0.8,
|
| 14467 |
+
"grad_norm": 0.45619475841522217,
|
| 14468 |
+
"learning_rate": 4.524256231573305e-06,
|
| 14469 |
+
"loss": 0.009,
|
| 14470 |
+
"step": 9905
|
| 14471 |
+
},
|
| 14472 |
+
{
|
| 14473 |
+
"epoch": 0.8,
|
| 14474 |
+
"grad_norm": 0.39344677329063416,
|
| 14475 |
+
"learning_rate": 4.515322076297686e-06,
|
| 14476 |
+
"loss": 0.0088,
|
| 14477 |
+
"step": 9910
|
| 14478 |
+
},
|
| 14479 |
+
{
|
| 14480 |
+
"epoch": 0.8,
|
| 14481 |
+
"grad_norm": 0.18160255253314972,
|
| 14482 |
+
"learning_rate": 4.506387921022068e-06,
|
| 14483 |
+
"loss": 0.0059,
|
| 14484 |
+
"step": 9915
|
| 14485 |
+
},
|
| 14486 |
+
{
|
| 14487 |
+
"epoch": 0.8,
|
| 14488 |
+
"grad_norm": 0.19619829952716827,
|
| 14489 |
+
"learning_rate": 4.497453765746449e-06,
|
| 14490 |
+
"loss": 0.0081,
|
| 14491 |
+
"step": 9920
|
| 14492 |
+
},
|
| 14493 |
+
{
|
| 14494 |
+
"epoch": 0.8,
|
| 14495 |
+
"grad_norm": 0.2846349775791168,
|
| 14496 |
+
"learning_rate": 4.488519610470831e-06,
|
| 14497 |
+
"loss": 0.0049,
|
| 14498 |
+
"step": 9925
|
| 14499 |
+
},
|
| 14500 |
+
{
|
| 14501 |
+
"epoch": 0.8,
|
| 14502 |
+
"grad_norm": 0.3724232316017151,
|
| 14503 |
+
"learning_rate": 4.479585455195212e-06,
|
| 14504 |
+
"loss": 0.0118,
|
| 14505 |
+
"step": 9930
|
| 14506 |
+
},
|
| 14507 |
+
{
|
| 14508 |
+
"epoch": 0.8,
|
| 14509 |
+
"grad_norm": 0.3385705351829529,
|
| 14510 |
+
"learning_rate": 4.470651299919593e-06,
|
| 14511 |
+
"loss": 0.0069,
|
| 14512 |
+
"step": 9935
|
| 14513 |
+
},
|
| 14514 |
+
{
|
| 14515 |
+
"epoch": 0.8,
|
| 14516 |
+
"grad_norm": 0.2874641418457031,
|
| 14517 |
+
"learning_rate": 4.461717144643974e-06,
|
| 14518 |
+
"loss": 0.006,
|
| 14519 |
+
"step": 9940
|
| 14520 |
+
},
|
| 14521 |
+
{
|
| 14522 |
+
"epoch": 0.8,
|
| 14523 |
+
"grad_norm": 0.3358646035194397,
|
| 14524 |
+
"learning_rate": 4.452782989368355e-06,
|
| 14525 |
+
"loss": 0.0062,
|
| 14526 |
+
"step": 9945
|
| 14527 |
+
},
|
| 14528 |
+
{
|
| 14529 |
+
"epoch": 0.8,
|
| 14530 |
+
"grad_norm": 0.7037550806999207,
|
| 14531 |
+
"learning_rate": 4.443848834092737e-06,
|
| 14532 |
+
"loss": 0.009,
|
| 14533 |
+
"step": 9950
|
| 14534 |
+
},
|
| 14535 |
+
{
|
| 14536 |
+
"epoch": 0.8,
|
| 14537 |
+
"grad_norm": 0.20332568883895874,
|
| 14538 |
+
"learning_rate": 4.434914678817118e-06,
|
| 14539 |
+
"loss": 0.0084,
|
| 14540 |
+
"step": 9955
|
| 14541 |
+
},
|
| 14542 |
+
{
|
| 14543 |
+
"epoch": 0.8,
|
| 14544 |
+
"grad_norm": 0.36842986941337585,
|
| 14545 |
+
"learning_rate": 4.425980523541499e-06,
|
| 14546 |
+
"loss": 0.0087,
|
| 14547 |
+
"step": 9960
|
| 14548 |
+
},
|
| 14549 |
+
{
|
| 14550 |
+
"epoch": 0.8,
|
| 14551 |
+
"grad_norm": 0.283997505903244,
|
| 14552 |
+
"learning_rate": 4.417046368265881e-06,
|
| 14553 |
+
"loss": 0.0086,
|
| 14554 |
+
"step": 9965
|
| 14555 |
+
},
|
| 14556 |
+
{
|
| 14557 |
+
"epoch": 0.8,
|
| 14558 |
+
"grad_norm": 0.42346441745758057,
|
| 14559 |
+
"learning_rate": 4.408112212990262e-06,
|
| 14560 |
+
"loss": 0.0078,
|
| 14561 |
+
"step": 9970
|
| 14562 |
+
},
|
| 14563 |
+
{
|
| 14564 |
+
"epoch": 0.8,
|
| 14565 |
+
"grad_norm": 0.18222256004810333,
|
| 14566 |
+
"learning_rate": 4.399178057714644e-06,
|
| 14567 |
+
"loss": 0.0074,
|
| 14568 |
+
"step": 9975
|
| 14569 |
+
},
|
| 14570 |
+
{
|
| 14571 |
+
"epoch": 0.8,
|
| 14572 |
+
"grad_norm": 0.6875673532485962,
|
| 14573 |
+
"learning_rate": 4.390243902439025e-06,
|
| 14574 |
+
"loss": 0.0102,
|
| 14575 |
+
"step": 9980
|
| 14576 |
+
},
|
| 14577 |
+
{
|
| 14578 |
+
"epoch": 0.8,
|
| 14579 |
+
"grad_norm": 0.24472391605377197,
|
| 14580 |
+
"learning_rate": 4.381309747163406e-06,
|
| 14581 |
+
"loss": 0.0065,
|
| 14582 |
+
"step": 9985
|
| 14583 |
+
},
|
| 14584 |
+
{
|
| 14585 |
+
"epoch": 0.8,
|
| 14586 |
+
"grad_norm": 0.26704928278923035,
|
| 14587 |
+
"learning_rate": 4.372375591887788e-06,
|
| 14588 |
+
"loss": 0.0083,
|
| 14589 |
+
"step": 9990
|
| 14590 |
+
},
|
| 14591 |
+
{
|
| 14592 |
+
"epoch": 0.8,
|
| 14593 |
+
"grad_norm": 0.5033184289932251,
|
| 14594 |
+
"learning_rate": 4.363441436612169e-06,
|
| 14595 |
+
"loss": 0.0067,
|
| 14596 |
+
"step": 9995
|
| 14597 |
+
},
|
| 14598 |
+
{
|
| 14599 |
+
"epoch": 0.8,
|
| 14600 |
+
"grad_norm": 0.781326413154602,
|
| 14601 |
+
"learning_rate": 4.35450728133655e-06,
|
| 14602 |
+
"loss": 0.006,
|
| 14603 |
+
"step": 10000
|
| 14604 |
+
},
|
| 14605 |
+
{
|
| 14606 |
+
"epoch": 0.8,
|
| 14607 |
+
"grad_norm": 0.46775344014167786,
|
| 14608 |
+
"learning_rate": 4.345573126060931e-06,
|
| 14609 |
+
"loss": 0.0087,
|
| 14610 |
+
"step": 10005
|
| 14611 |
+
},
|
| 14612 |
+
{
|
| 14613 |
+
"epoch": 0.8,
|
| 14614 |
+
"grad_norm": 0.3803477883338928,
|
| 14615 |
+
"learning_rate": 4.336638970785312e-06,
|
| 14616 |
+
"loss": 0.0098,
|
| 14617 |
+
"step": 10010
|
| 14618 |
+
},
|
| 14619 |
+
{
|
| 14620 |
+
"epoch": 0.81,
|
| 14621 |
+
"grad_norm": 0.23086823523044586,
|
| 14622 |
+
"learning_rate": 4.327704815509694e-06,
|
| 14623 |
+
"loss": 0.0089,
|
| 14624 |
+
"step": 10015
|
| 14625 |
+
},
|
| 14626 |
+
{
|
| 14627 |
+
"epoch": 0.81,
|
| 14628 |
+
"grad_norm": 0.18261872231960297,
|
| 14629 |
+
"learning_rate": 4.318770660234075e-06,
|
| 14630 |
+
"loss": 0.01,
|
| 14631 |
+
"step": 10020
|
| 14632 |
+
},
|
| 14633 |
+
{
|
| 14634 |
+
"epoch": 0.81,
|
| 14635 |
+
"grad_norm": 0.33674633502960205,
|
| 14636 |
+
"learning_rate": 4.309836504958457e-06,
|
| 14637 |
+
"loss": 0.0086,
|
| 14638 |
+
"step": 10025
|
| 14639 |
+
},
|
| 14640 |
+
{
|
| 14641 |
+
"epoch": 0.81,
|
| 14642 |
+
"grad_norm": 0.29859867691993713,
|
| 14643 |
+
"learning_rate": 4.300902349682838e-06,
|
| 14644 |
+
"loss": 0.009,
|
| 14645 |
+
"step": 10030
|
| 14646 |
+
},
|
| 14647 |
+
{
|
| 14648 |
+
"epoch": 0.81,
|
| 14649 |
+
"grad_norm": 0.40897712111473083,
|
| 14650 |
+
"learning_rate": 4.291968194407219e-06,
|
| 14651 |
+
"loss": 0.0092,
|
| 14652 |
+
"step": 10035
|
| 14653 |
+
},
|
| 14654 |
+
{
|
| 14655 |
+
"epoch": 0.81,
|
| 14656 |
+
"grad_norm": 0.7651856541633606,
|
| 14657 |
+
"learning_rate": 4.283034039131601e-06,
|
| 14658 |
+
"loss": 0.0108,
|
| 14659 |
+
"step": 10040
|
| 14660 |
+
},
|
| 14661 |
+
{
|
| 14662 |
+
"epoch": 0.81,
|
| 14663 |
+
"grad_norm": 0.7065618634223938,
|
| 14664 |
+
"learning_rate": 4.274099883855982e-06,
|
| 14665 |
+
"loss": 0.0132,
|
| 14666 |
+
"step": 10045
|
| 14667 |
+
},
|
| 14668 |
+
{
|
| 14669 |
+
"epoch": 0.81,
|
| 14670 |
+
"grad_norm": 0.4190121293067932,
|
| 14671 |
+
"learning_rate": 4.265165728580363e-06,
|
| 14672 |
+
"loss": 0.0063,
|
| 14673 |
+
"step": 10050
|
| 14674 |
+
},
|
| 14675 |
+
{
|
| 14676 |
+
"epoch": 0.81,
|
| 14677 |
+
"grad_norm": 0.47081393003463745,
|
| 14678 |
+
"learning_rate": 4.256231573304745e-06,
|
| 14679 |
+
"loss": 0.0065,
|
| 14680 |
+
"step": 10055
|
| 14681 |
+
},
|
| 14682 |
+
{
|
| 14683 |
+
"epoch": 0.81,
|
| 14684 |
+
"grad_norm": 0.3828545808792114,
|
| 14685 |
+
"learning_rate": 4.247297418029126e-06,
|
| 14686 |
+
"loss": 0.0087,
|
| 14687 |
+
"step": 10060
|
| 14688 |
+
},
|
| 14689 |
+
{
|
| 14690 |
+
"epoch": 0.81,
|
| 14691 |
+
"grad_norm": 0.7525375485420227,
|
| 14692 |
+
"learning_rate": 4.238363262753507e-06,
|
| 14693 |
+
"loss": 0.0113,
|
| 14694 |
+
"step": 10065
|
| 14695 |
+
},
|
| 14696 |
+
{
|
| 14697 |
+
"epoch": 0.81,
|
| 14698 |
+
"grad_norm": 0.34589239954948425,
|
| 14699 |
+
"learning_rate": 4.229429107477888e-06,
|
| 14700 |
+
"loss": 0.0078,
|
| 14701 |
+
"step": 10070
|
| 14702 |
+
},
|
| 14703 |
+
{
|
| 14704 |
+
"epoch": 0.81,
|
| 14705 |
+
"grad_norm": 0.5909443497657776,
|
| 14706 |
+
"learning_rate": 4.22049495220227e-06,
|
| 14707 |
+
"loss": 0.007,
|
| 14708 |
+
"step": 10075
|
| 14709 |
+
},
|
| 14710 |
+
{
|
| 14711 |
+
"epoch": 0.81,
|
| 14712 |
+
"grad_norm": 0.3668850362300873,
|
| 14713 |
+
"learning_rate": 4.211560796926651e-06,
|
| 14714 |
+
"loss": 0.0061,
|
| 14715 |
+
"step": 10080
|
| 14716 |
+
},
|
| 14717 |
+
{
|
| 14718 |
+
"epoch": 0.81,
|
| 14719 |
+
"grad_norm": 0.4989503026008606,
|
| 14720 |
+
"learning_rate": 4.202626641651033e-06,
|
| 14721 |
+
"loss": 0.0071,
|
| 14722 |
+
"step": 10085
|
| 14723 |
+
},
|
| 14724 |
+
{
|
| 14725 |
+
"epoch": 0.81,
|
| 14726 |
+
"grad_norm": 0.44953587651252747,
|
| 14727 |
+
"learning_rate": 4.193692486375414e-06,
|
| 14728 |
+
"loss": 0.0079,
|
| 14729 |
+
"step": 10090
|
| 14730 |
+
},
|
| 14731 |
+
{
|
| 14732 |
+
"epoch": 0.81,
|
| 14733 |
+
"grad_norm": 0.36224547028541565,
|
| 14734 |
+
"learning_rate": 4.184758331099795e-06,
|
| 14735 |
+
"loss": 0.0101,
|
| 14736 |
+
"step": 10095
|
| 14737 |
+
},
|
| 14738 |
+
{
|
| 14739 |
+
"epoch": 0.81,
|
| 14740 |
+
"grad_norm": 0.21505020558834076,
|
| 14741 |
+
"learning_rate": 4.175824175824177e-06,
|
| 14742 |
+
"loss": 0.0116,
|
| 14743 |
+
"step": 10100
|
| 14744 |
+
},
|
| 14745 |
+
{
|
| 14746 |
+
"epoch": 0.81,
|
| 14747 |
+
"grad_norm": 0.5628384351730347,
|
| 14748 |
+
"learning_rate": 4.166890020548558e-06,
|
| 14749 |
+
"loss": 0.0106,
|
| 14750 |
+
"step": 10105
|
| 14751 |
+
},
|
| 14752 |
+
{
|
| 14753 |
+
"epoch": 0.81,
|
| 14754 |
+
"grad_norm": 0.6173145174980164,
|
| 14755 |
+
"learning_rate": 4.157955865272939e-06,
|
| 14756 |
+
"loss": 0.0074,
|
| 14757 |
+
"step": 10110
|
| 14758 |
+
},
|
| 14759 |
+
{
|
| 14760 |
+
"epoch": 0.81,
|
| 14761 |
+
"grad_norm": 0.20209026336669922,
|
| 14762 |
+
"learning_rate": 4.14902170999732e-06,
|
| 14763 |
+
"loss": 0.0054,
|
| 14764 |
+
"step": 10115
|
| 14765 |
+
},
|
| 14766 |
+
{
|
| 14767 |
+
"epoch": 0.81,
|
| 14768 |
+
"grad_norm": 0.26845335960388184,
|
| 14769 |
+
"learning_rate": 4.140087554721701e-06,
|
| 14770 |
+
"loss": 0.0084,
|
| 14771 |
+
"step": 10120
|
| 14772 |
+
},
|
| 14773 |
+
{
|
| 14774 |
+
"epoch": 0.81,
|
| 14775 |
+
"grad_norm": 0.19294553995132446,
|
| 14776 |
+
"learning_rate": 4.131153399446083e-06,
|
| 14777 |
+
"loss": 0.0069,
|
| 14778 |
+
"step": 10125
|
| 14779 |
+
},
|
| 14780 |
+
{
|
| 14781 |
+
"epoch": 0.81,
|
| 14782 |
+
"grad_norm": 0.2686295211315155,
|
| 14783 |
+
"learning_rate": 4.122219244170464e-06,
|
| 14784 |
+
"loss": 0.0062,
|
| 14785 |
+
"step": 10130
|
| 14786 |
+
},
|
| 14787 |
+
{
|
| 14788 |
+
"epoch": 0.81,
|
| 14789 |
+
"grad_norm": 0.34871765971183777,
|
| 14790 |
+
"learning_rate": 4.113285088894846e-06,
|
| 14791 |
+
"loss": 0.0087,
|
| 14792 |
+
"step": 10135
|
| 14793 |
+
},
|
| 14794 |
+
{
|
| 14795 |
+
"epoch": 0.82,
|
| 14796 |
+
"grad_norm": 0.3453786373138428,
|
| 14797 |
+
"learning_rate": 4.104350933619227e-06,
|
| 14798 |
+
"loss": 0.008,
|
| 14799 |
+
"step": 10140
|
| 14800 |
+
},
|
| 14801 |
+
{
|
| 14802 |
+
"epoch": 0.82,
|
| 14803 |
+
"grad_norm": 0.4701385200023651,
|
| 14804 |
+
"learning_rate": 4.095416778343608e-06,
|
| 14805 |
+
"loss": 0.0077,
|
| 14806 |
+
"step": 10145
|
| 14807 |
+
},
|
| 14808 |
+
{
|
| 14809 |
+
"epoch": 0.82,
|
| 14810 |
+
"grad_norm": 0.13584518432617188,
|
| 14811 |
+
"learning_rate": 4.0864826230679896e-06,
|
| 14812 |
+
"loss": 0.0063,
|
| 14813 |
+
"step": 10150
|
| 14814 |
+
},
|
| 14815 |
+
{
|
| 14816 |
+
"epoch": 0.82,
|
| 14817 |
+
"grad_norm": 0.5030553936958313,
|
| 14818 |
+
"learning_rate": 4.077548467792371e-06,
|
| 14819 |
+
"loss": 0.0076,
|
| 14820 |
+
"step": 10155
|
| 14821 |
+
},
|
| 14822 |
+
{
|
| 14823 |
+
"epoch": 0.82,
|
| 14824 |
+
"grad_norm": 0.44598788022994995,
|
| 14825 |
+
"learning_rate": 4.0686143125167525e-06,
|
| 14826 |
+
"loss": 0.0067,
|
| 14827 |
+
"step": 10160
|
| 14828 |
+
},
|
| 14829 |
+
{
|
| 14830 |
+
"epoch": 0.82,
|
| 14831 |
+
"grad_norm": 0.2886448800563812,
|
| 14832 |
+
"learning_rate": 4.0596801572411335e-06,
|
| 14833 |
+
"loss": 0.0077,
|
| 14834 |
+
"step": 10165
|
| 14835 |
+
},
|
| 14836 |
+
{
|
| 14837 |
+
"epoch": 0.82,
|
| 14838 |
+
"grad_norm": 0.2822360098361969,
|
| 14839 |
+
"learning_rate": 4.0507460019655146e-06,
|
| 14840 |
+
"loss": 0.0078,
|
| 14841 |
+
"step": 10170
|
| 14842 |
+
},
|
| 14843 |
+
{
|
| 14844 |
+
"epoch": 0.82,
|
| 14845 |
+
"grad_norm": 0.5592710375785828,
|
| 14846 |
+
"learning_rate": 4.041811846689896e-06,
|
| 14847 |
+
"loss": 0.0076,
|
| 14848 |
+
"step": 10175
|
| 14849 |
+
},
|
| 14850 |
+
{
|
| 14851 |
+
"epoch": 0.82,
|
| 14852 |
+
"grad_norm": 0.310092031955719,
|
| 14853 |
+
"learning_rate": 4.032877691414277e-06,
|
| 14854 |
+
"loss": 0.0107,
|
| 14855 |
+
"step": 10180
|
| 14856 |
+
},
|
| 14857 |
+
{
|
| 14858 |
+
"epoch": 0.82,
|
| 14859 |
+
"grad_norm": 0.3115810751914978,
|
| 14860 |
+
"learning_rate": 4.0239435361386585e-06,
|
| 14861 |
+
"loss": 0.0087,
|
| 14862 |
+
"step": 10185
|
| 14863 |
+
},
|
| 14864 |
+
{
|
| 14865 |
+
"epoch": 0.82,
|
| 14866 |
+
"grad_norm": 0.9056434035301208,
|
| 14867 |
+
"learning_rate": 4.0150093808630396e-06,
|
| 14868 |
+
"loss": 0.0099,
|
| 14869 |
+
"step": 10190
|
| 14870 |
+
},
|
| 14871 |
+
{
|
| 14872 |
+
"epoch": 0.82,
|
| 14873 |
+
"grad_norm": 0.31844180822372437,
|
| 14874 |
+
"learning_rate": 4.006075225587421e-06,
|
| 14875 |
+
"loss": 0.0111,
|
| 14876 |
+
"step": 10195
|
| 14877 |
+
},
|
| 14878 |
+
{
|
| 14879 |
+
"epoch": 0.82,
|
| 14880 |
+
"grad_norm": 0.5329269766807556,
|
| 14881 |
+
"learning_rate": 3.9971410703118025e-06,
|
| 14882 |
+
"loss": 0.01,
|
| 14883 |
+
"step": 10200
|
| 14884 |
+
},
|
| 14885 |
+
{
|
| 14886 |
+
"epoch": 0.82,
|
| 14887 |
+
"eval_loss": 0.007998097687959671,
|
| 14888 |
+
"eval_pearson_cosine": 0.9672417410329249,
|
| 14889 |
+
"eval_pearson_dot": 0.965013249775285,
|
| 14890 |
+
"eval_pearson_euclidean": 0.9542839853098279,
|
| 14891 |
+
"eval_pearson_manhattan": 0.9528819643748915,
|
| 14892 |
+
"eval_pearson_max": 0.9672417410329249,
|
| 14893 |
+
"eval_runtime": 425.8541,
|
| 14894 |
+
"eval_samples_per_second": 1.174,
|
| 14895 |
+
"eval_spearman_cosine": 0.9626618986475944,
|
| 14896 |
+
"eval_spearman_dot": 0.9577223908895635,
|
| 14897 |
+
"eval_spearman_euclidean": 0.962258313033252,
|
| 14898 |
+
"eval_spearman_manhattan": 0.9619932559730238,
|
| 14899 |
+
"eval_spearman_max": 0.9626618986475944,
|
| 14900 |
+
"eval_steps_per_second": 1.174,
|
| 14901 |
+
"step": 10200
|
| 14902 |
+
},
|
| 14903 |
+
{
|
| 14904 |
+
"epoch": 0.82,
|
| 14905 |
+
"grad_norm": 0.3950428068637848,
|
| 14906 |
+
"learning_rate": 3.9882069150361835e-06,
|
| 14907 |
+
"loss": 0.0074,
|
| 14908 |
+
"step": 10205
|
| 14909 |
+
},
|
| 14910 |
+
{
|
| 14911 |
+
"epoch": 0.82,
|
| 14912 |
+
"grad_norm": 0.3376205563545227,
|
| 14913 |
+
"learning_rate": 3.979272759760565e-06,
|
| 14914 |
+
"loss": 0.0079,
|
| 14915 |
+
"step": 10210
|
| 14916 |
+
},
|
| 14917 |
+
{
|
| 14918 |
+
"epoch": 0.82,
|
| 14919 |
+
"grad_norm": 0.39471399784088135,
|
| 14920 |
+
"learning_rate": 3.9703386044849465e-06,
|
| 14921 |
+
"loss": 0.0087,
|
| 14922 |
+
"step": 10215
|
| 14923 |
+
},
|
| 14924 |
+
{
|
| 14925 |
+
"epoch": 0.82,
|
| 14926 |
+
"grad_norm": 0.42924797534942627,
|
| 14927 |
+
"learning_rate": 3.9614044492093275e-06,
|
| 14928 |
+
"loss": 0.0086,
|
| 14929 |
+
"step": 10220
|
| 14930 |
+
},
|
| 14931 |
+
{
|
| 14932 |
+
"epoch": 0.82,
|
| 14933 |
+
"grad_norm": 0.26988697052001953,
|
| 14934 |
+
"learning_rate": 3.9524702939337085e-06,
|
| 14935 |
+
"loss": 0.0076,
|
| 14936 |
+
"step": 10225
|
| 14937 |
+
},
|
| 14938 |
+
{
|
| 14939 |
+
"epoch": 0.82,
|
| 14940 |
+
"grad_norm": 0.5494524240493774,
|
| 14941 |
+
"learning_rate": 3.94353613865809e-06,
|
| 14942 |
+
"loss": 0.0096,
|
| 14943 |
+
"step": 10230
|
| 14944 |
+
},
|
| 14945 |
+
{
|
| 14946 |
+
"epoch": 0.82,
|
| 14947 |
+
"grad_norm": 0.39503470063209534,
|
| 14948 |
+
"learning_rate": 3.9346019833824715e-06,
|
| 14949 |
+
"loss": 0.0091,
|
| 14950 |
+
"step": 10235
|
| 14951 |
+
},
|
| 14952 |
+
{
|
| 14953 |
+
"epoch": 0.82,
|
| 14954 |
+
"grad_norm": 0.2849455177783966,
|
| 14955 |
+
"learning_rate": 3.9256678281068525e-06,
|
| 14956 |
+
"loss": 0.0066,
|
| 14957 |
+
"step": 10240
|
| 14958 |
+
},
|
| 14959 |
+
{
|
| 14960 |
+
"epoch": 0.82,
|
| 14961 |
+
"grad_norm": 0.2523050904273987,
|
| 14962 |
+
"learning_rate": 3.916733672831234e-06,
|
| 14963 |
+
"loss": 0.0076,
|
| 14964 |
+
"step": 10245
|
| 14965 |
+
},
|
| 14966 |
+
{
|
| 14967 |
+
"epoch": 0.82,
|
| 14968 |
+
"grad_norm": 0.4256332218647003,
|
| 14969 |
+
"learning_rate": 3.9077995175556154e-06,
|
| 14970 |
+
"loss": 0.0061,
|
| 14971 |
+
"step": 10250
|
| 14972 |
+
},
|
| 14973 |
+
{
|
| 14974 |
+
"epoch": 0.82,
|
| 14975 |
+
"grad_norm": 0.2837385833263397,
|
| 14976 |
+
"learning_rate": 3.8988653622799965e-06,
|
| 14977 |
+
"loss": 0.0069,
|
| 14978 |
+
"step": 10255
|
| 14979 |
+
},
|
| 14980 |
+
{
|
| 14981 |
+
"epoch": 0.82,
|
| 14982 |
+
"grad_norm": 0.44190511107444763,
|
| 14983 |
+
"learning_rate": 3.889931207004378e-06,
|
| 14984 |
+
"loss": 0.0059,
|
| 14985 |
+
"step": 10260
|
| 14986 |
+
},
|
| 14987 |
+
{
|
| 14988 |
+
"epoch": 0.83,
|
| 14989 |
+
"grad_norm": 0.3654380440711975,
|
| 14990 |
+
"learning_rate": 3.880997051728759e-06,
|
| 14991 |
+
"loss": 0.0071,
|
| 14992 |
+
"step": 10265
|
| 14993 |
+
},
|
| 14994 |
+
{
|
| 14995 |
+
"epoch": 0.83,
|
| 14996 |
+
"grad_norm": 0.356179416179657,
|
| 14997 |
+
"learning_rate": 3.872062896453141e-06,
|
| 14998 |
+
"loss": 0.005,
|
| 14999 |
+
"step": 10270
|
| 15000 |
+
},
|
| 15001 |
+
{
|
| 15002 |
+
"epoch": 0.83,
|
| 15003 |
+
"grad_norm": 0.4459650218486786,
|
| 15004 |
+
"learning_rate": 3.863128741177522e-06,
|
| 15005 |
+
"loss": 0.0074,
|
| 15006 |
+
"step": 10275
|
| 15007 |
+
},
|
| 15008 |
+
{
|
| 15009 |
+
"epoch": 0.83,
|
| 15010 |
+
"grad_norm": 0.6780635118484497,
|
| 15011 |
+
"learning_rate": 3.854194585901903e-06,
|
| 15012 |
+
"loss": 0.0101,
|
| 15013 |
+
"step": 10280
|
| 15014 |
+
},
|
| 15015 |
+
{
|
| 15016 |
+
"epoch": 0.83,
|
| 15017 |
+
"grad_norm": 0.42910513281822205,
|
| 15018 |
+
"learning_rate": 3.845260430626284e-06,
|
| 15019 |
+
"loss": 0.0087,
|
| 15020 |
+
"step": 10285
|
| 15021 |
+
},
|
| 15022 |
+
{
|
| 15023 |
+
"epoch": 0.83,
|
| 15024 |
+
"grad_norm": 0.38253623247146606,
|
| 15025 |
+
"learning_rate": 3.8363262753506654e-06,
|
| 15026 |
+
"loss": 0.0054,
|
| 15027 |
+
"step": 10290
|
| 15028 |
+
},
|
| 15029 |
+
{
|
| 15030 |
+
"epoch": 0.83,
|
| 15031 |
+
"grad_norm": 0.3616214990615845,
|
| 15032 |
+
"learning_rate": 3.827392120075047e-06,
|
| 15033 |
+
"loss": 0.0066,
|
| 15034 |
+
"step": 10295
|
| 15035 |
+
},
|
| 15036 |
+
{
|
| 15037 |
+
"epoch": 0.83,
|
| 15038 |
+
"grad_norm": 0.5730588436126709,
|
| 15039 |
+
"learning_rate": 3.818457964799428e-06,
|
| 15040 |
+
"loss": 0.0075,
|
| 15041 |
+
"step": 10300
|
| 15042 |
+
},
|
| 15043 |
+
{
|
| 15044 |
+
"epoch": 0.83,
|
| 15045 |
+
"grad_norm": 0.34027931094169617,
|
| 15046 |
+
"learning_rate": 3.80952380952381e-06,
|
| 15047 |
+
"loss": 0.0115,
|
| 15048 |
+
"step": 10305
|
| 15049 |
+
},
|
| 15050 |
+
{
|
| 15051 |
+
"epoch": 0.83,
|
| 15052 |
+
"grad_norm": 0.34853848814964294,
|
| 15053 |
+
"learning_rate": 3.8005896542481913e-06,
|
| 15054 |
+
"loss": 0.0098,
|
| 15055 |
+
"step": 10310
|
| 15056 |
+
},
|
| 15057 |
+
{
|
| 15058 |
+
"epoch": 0.83,
|
| 15059 |
+
"grad_norm": 0.3500560522079468,
|
| 15060 |
+
"learning_rate": 3.7916554989725723e-06,
|
| 15061 |
+
"loss": 0.0099,
|
| 15062 |
+
"step": 10315
|
| 15063 |
+
},
|
| 15064 |
+
{
|
| 15065 |
+
"epoch": 0.83,
|
| 15066 |
+
"grad_norm": 0.4545835256576538,
|
| 15067 |
+
"learning_rate": 3.7827213436969538e-06,
|
| 15068 |
+
"loss": 0.0075,
|
| 15069 |
+
"step": 10320
|
| 15070 |
+
},
|
| 15071 |
+
{
|
| 15072 |
+
"epoch": 0.83,
|
| 15073 |
+
"grad_norm": 0.3421791195869446,
|
| 15074 |
+
"learning_rate": 3.7737871884213352e-06,
|
| 15075 |
+
"loss": 0.0105,
|
| 15076 |
+
"step": 10325
|
| 15077 |
+
},
|
| 15078 |
+
{
|
| 15079 |
+
"epoch": 0.83,
|
| 15080 |
+
"grad_norm": 0.28388506174087524,
|
| 15081 |
+
"learning_rate": 3.7648530331457163e-06,
|
| 15082 |
+
"loss": 0.0135,
|
| 15083 |
+
"step": 10330
|
| 15084 |
+
},
|
| 15085 |
+
{
|
| 15086 |
+
"epoch": 0.83,
|
| 15087 |
+
"grad_norm": 0.6649767160415649,
|
| 15088 |
+
"learning_rate": 3.7559188778700977e-06,
|
| 15089 |
+
"loss": 0.0105,
|
| 15090 |
+
"step": 10335
|
| 15091 |
+
},
|
| 15092 |
+
{
|
| 15093 |
+
"epoch": 0.83,
|
| 15094 |
+
"grad_norm": 0.41207408905029297,
|
| 15095 |
+
"learning_rate": 3.7469847225944788e-06,
|
| 15096 |
+
"loss": 0.0065,
|
| 15097 |
+
"step": 10340
|
| 15098 |
+
},
|
| 15099 |
+
{
|
| 15100 |
+
"epoch": 0.83,
|
| 15101 |
+
"grad_norm": 0.3896176815032959,
|
| 15102 |
+
"learning_rate": 3.7380505673188607e-06,
|
| 15103 |
+
"loss": 0.0098,
|
| 15104 |
+
"step": 10345
|
| 15105 |
+
},
|
| 15106 |
+
{
|
| 15107 |
+
"epoch": 0.83,
|
| 15108 |
+
"grad_norm": 0.26551979780197144,
|
| 15109 |
+
"learning_rate": 3.7291164120432417e-06,
|
| 15110 |
+
"loss": 0.0088,
|
| 15111 |
+
"step": 10350
|
| 15112 |
+
},
|
| 15113 |
+
{
|
| 15114 |
+
"epoch": 0.83,
|
| 15115 |
+
"grad_norm": 0.22879204154014587,
|
| 15116 |
+
"learning_rate": 3.7201822567676227e-06,
|
| 15117 |
+
"loss": 0.0094,
|
| 15118 |
+
"step": 10355
|
| 15119 |
+
},
|
| 15120 |
+
{
|
| 15121 |
+
"epoch": 0.83,
|
| 15122 |
+
"grad_norm": 0.6052958369255066,
|
| 15123 |
+
"learning_rate": 3.711248101492004e-06,
|
| 15124 |
+
"loss": 0.0092,
|
| 15125 |
+
"step": 10360
|
| 15126 |
+
},
|
| 15127 |
+
{
|
| 15128 |
+
"epoch": 0.83,
|
| 15129 |
+
"grad_norm": 0.2694813311100006,
|
| 15130 |
+
"learning_rate": 3.7023139462163852e-06,
|
| 15131 |
+
"loss": 0.0054,
|
| 15132 |
+
"step": 10365
|
| 15133 |
+
},
|
| 15134 |
+
{
|
| 15135 |
+
"epoch": 0.83,
|
| 15136 |
+
"grad_norm": 0.3190039098262787,
|
| 15137 |
+
"learning_rate": 3.693379790940767e-06,
|
| 15138 |
+
"loss": 0.009,
|
| 15139 |
+
"step": 10370
|
| 15140 |
+
},
|
| 15141 |
+
{
|
| 15142 |
+
"epoch": 0.83,
|
| 15143 |
+
"grad_norm": 0.2353006899356842,
|
| 15144 |
+
"learning_rate": 3.684445635665148e-06,
|
| 15145 |
+
"loss": 0.0088,
|
| 15146 |
+
"step": 10375
|
| 15147 |
+
},
|
| 15148 |
+
{
|
| 15149 |
+
"epoch": 0.83,
|
| 15150 |
+
"grad_norm": 0.9957902431488037,
|
| 15151 |
+
"learning_rate": 3.675511480389529e-06,
|
| 15152 |
+
"loss": 0.0116,
|
| 15153 |
+
"step": 10380
|
| 15154 |
+
},
|
| 15155 |
+
{
|
| 15156 |
+
"epoch": 0.83,
|
| 15157 |
+
"grad_norm": 0.4154163897037506,
|
| 15158 |
+
"learning_rate": 3.6665773251139107e-06,
|
| 15159 |
+
"loss": 0.0055,
|
| 15160 |
+
"step": 10385
|
| 15161 |
+
},
|
| 15162 |
+
{
|
| 15163 |
+
"epoch": 0.84,
|
| 15164 |
+
"grad_norm": 0.3367329239845276,
|
| 15165 |
+
"learning_rate": 3.657643169838292e-06,
|
| 15166 |
+
"loss": 0.0061,
|
| 15167 |
+
"step": 10390
|
| 15168 |
+
},
|
| 15169 |
+
{
|
| 15170 |
+
"epoch": 0.84,
|
| 15171 |
+
"grad_norm": 0.3034825325012207,
|
| 15172 |
+
"learning_rate": 3.6487090145626736e-06,
|
| 15173 |
+
"loss": 0.0081,
|
| 15174 |
+
"step": 10395
|
| 15175 |
+
},
|
| 15176 |
+
{
|
| 15177 |
+
"epoch": 0.84,
|
| 15178 |
+
"grad_norm": 0.17845579981803894,
|
| 15179 |
+
"learning_rate": 3.6397748592870546e-06,
|
| 15180 |
+
"loss": 0.0043,
|
| 15181 |
+
"step": 10400
|
| 15182 |
}
|
| 15183 |
],
|
| 15184 |
"logging_steps": 5,
|