Upload 10 files
Browse files- config.json +28 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +15 -0
- tokenizer.json +0 -0
- tokenizer_config.json +15 -0
- trainer_state.json +2682 -0
- training_args.bin +3 -0
- unigram.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "sdadas/polish-roberta-base-v2",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"RobertaForCL"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"gradient_checkpointing": false,
|
| 11 |
+
"hidden_act": "gelu",
|
| 12 |
+
"hidden_dropout_prob": 0.1,
|
| 13 |
+
"hidden_size": 768,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-05,
|
| 17 |
+
"max_position_embeddings": 514,
|
| 18 |
+
"model_type": "roberta",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_layers": 12,
|
| 21 |
+
"pad_token_id": 1,
|
| 22 |
+
"position_embedding_type": "absolute",
|
| 23 |
+
"torch_dtype": "float32",
|
| 24 |
+
"transformers_version": "4.24.0",
|
| 25 |
+
"type_vocab_size": 1,
|
| 26 |
+
"use_cache": true,
|
| 27 |
+
"vocab_size": 50001
|
| 28 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:30efe561e4e4ce9af2b17214a5de25d8b9ee62f52cdb29517f2a0e0a19211385
|
| 3 |
+
size 995660602
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:536443671a6937f47d062168d45d7e5cd48b72cfe9323063e1b2595ac064c68b
|
| 3 |
+
size 497844714
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b64066252def1b7d97724b866f8cfda8eb5cca0df933eed500854e14a546234
|
| 3 |
+
size 1064
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "<mask>",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "<pad>",
|
| 13 |
+
"sep_token": "</s>",
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"bos_token": "<s>",
|
| 4 |
+
"cls_token": "<s>",
|
| 5 |
+
"eos_token": "</s>",
|
| 6 |
+
"errors": "replace",
|
| 7 |
+
"mask_token": "<mask>",
|
| 8 |
+
"name_or_path": "sdadas/polish-roberta-base-v2",
|
| 9 |
+
"pad_token": "<pad>",
|
| 10 |
+
"sep_token": "</s>",
|
| 11 |
+
"special_tokens_map_file": null,
|
| 12 |
+
"tokenizer_class": "RobertaTokenizer",
|
| 13 |
+
"trim_offsets": true,
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2682 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7694667033321299,
|
| 3 |
+
"best_model_checkpoint": "/praid/simcse",
|
| 4 |
+
"epoch": 0.40088254442863686,
|
| 5 |
+
"global_step": 66500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 9.999969858455306e-09,
|
| 13 |
+
"loss": 5.4599,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.0,
|
| 18 |
+
"eval_avg_sts": 0.5995849851747349,
|
| 19 |
+
"eval_sickr_spearman": 0.5697828282953606,
|
| 20 |
+
"eval_stsb_spearman": 0.6293871420541092,
|
| 21 |
+
"step": 250
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.0,
|
| 25 |
+
"learning_rate": 9.984929227653058e-09,
|
| 26 |
+
"loss": 5.4858,
|
| 27 |
+
"step": 500
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.0,
|
| 31 |
+
"eval_avg_sts": 0.603632450991215,
|
| 32 |
+
"eval_sickr_spearman": 0.5695678305286125,
|
| 33 |
+
"eval_stsb_spearman": 0.6376970714538175,
|
| 34 |
+
"step": 500
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.0,
|
| 38 |
+
"eval_avg_sts": 0.6086904770175188,
|
| 39 |
+
"eval_sickr_spearman": 0.5718612396457673,
|
| 40 |
+
"eval_stsb_spearman": 0.6455197143892701,
|
| 41 |
+
"step": 750
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.01,
|
| 45 |
+
"learning_rate": 9.969858455306118e-09,
|
| 46 |
+
"loss": 5.2729,
|
| 47 |
+
"step": 1000
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.01,
|
| 51 |
+
"eval_avg_sts": 0.612799381419272,
|
| 52 |
+
"eval_sickr_spearman": 0.5742047501651341,
|
| 53 |
+
"eval_stsb_spearman": 0.6513940126734101,
|
| 54 |
+
"step": 1000
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.01,
|
| 58 |
+
"eval_avg_sts": 0.6162570263582352,
|
| 59 |
+
"eval_sickr_spearman": 0.5758439027009196,
|
| 60 |
+
"eval_stsb_spearman": 0.656670150015551,
|
| 61 |
+
"step": 1250
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.01,
|
| 65 |
+
"learning_rate": 9.954787682959177e-09,
|
| 66 |
+
"loss": 4.9236,
|
| 67 |
+
"step": 1500
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.01,
|
| 71 |
+
"eval_avg_sts": 0.6186805435983472,
|
| 72 |
+
"eval_sickr_spearman": 0.5769587834066011,
|
| 73 |
+
"eval_stsb_spearman": 0.6604023037900933,
|
| 74 |
+
"step": 1500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.01,
|
| 78 |
+
"eval_avg_sts": 0.6207924657118333,
|
| 79 |
+
"eval_sickr_spearman": 0.5785645184067163,
|
| 80 |
+
"eval_stsb_spearman": 0.6630204130169505,
|
| 81 |
+
"step": 1750
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.01,
|
| 85 |
+
"learning_rate": 9.939716910612235e-09,
|
| 86 |
+
"loss": 4.3524,
|
| 87 |
+
"step": 2000
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.01,
|
| 91 |
+
"eval_avg_sts": 0.6228190394905305,
|
| 92 |
+
"eval_sickr_spearman": 0.5800011736250131,
|
| 93 |
+
"eval_stsb_spearman": 0.6656369053560479,
|
| 94 |
+
"step": 2000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.01,
|
| 98 |
+
"eval_avg_sts": 0.6242788079574524,
|
| 99 |
+
"eval_sickr_spearman": 0.5802247872392591,
|
| 100 |
+
"eval_stsb_spearman": 0.6683328286756457,
|
| 101 |
+
"step": 2250
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.02,
|
| 105 |
+
"learning_rate": 9.924646138265295e-09,
|
| 106 |
+
"loss": 3.8783,
|
| 107 |
+
"step": 2500
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.02,
|
| 111 |
+
"eval_avg_sts": 0.6263156249834569,
|
| 112 |
+
"eval_sickr_spearman": 0.5810882644873206,
|
| 113 |
+
"eval_stsb_spearman": 0.6715429854795933,
|
| 114 |
+
"step": 2500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.02,
|
| 118 |
+
"eval_avg_sts": 0.6282653788172143,
|
| 119 |
+
"eval_sickr_spearman": 0.5811133151884266,
|
| 120 |
+
"eval_stsb_spearman": 0.6754174424460021,
|
| 121 |
+
"step": 2750
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.02,
|
| 125 |
+
"learning_rate": 9.909575365918353e-09,
|
| 126 |
+
"loss": 3.3578,
|
| 127 |
+
"step": 3000
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.02,
|
| 131 |
+
"eval_avg_sts": 0.6304970270247872,
|
| 132 |
+
"eval_sickr_spearman": 0.5811867242012102,
|
| 133 |
+
"eval_stsb_spearman": 0.6798073298483642,
|
| 134 |
+
"step": 3000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.02,
|
| 138 |
+
"eval_avg_sts": 0.6327063870946847,
|
| 139 |
+
"eval_sickr_spearman": 0.5807497065029901,
|
| 140 |
+
"eval_stsb_spearman": 0.6846630676863793,
|
| 141 |
+
"step": 3250
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.02,
|
| 145 |
+
"learning_rate": 9.894504593571412e-09,
|
| 146 |
+
"loss": 2.9718,
|
| 147 |
+
"step": 3500
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.02,
|
| 151 |
+
"eval_avg_sts": 0.635125450375027,
|
| 152 |
+
"eval_sickr_spearman": 0.5803658779673568,
|
| 153 |
+
"eval_stsb_spearman": 0.6898850227826971,
|
| 154 |
+
"step": 3500
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.02,
|
| 158 |
+
"eval_avg_sts": 0.6375971333458483,
|
| 159 |
+
"eval_sickr_spearman": 0.5801840487233453,
|
| 160 |
+
"eval_stsb_spearman": 0.6950102179683513,
|
| 161 |
+
"step": 3750
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.02,
|
| 165 |
+
"learning_rate": 9.87943382122447e-09,
|
| 166 |
+
"loss": 2.4584,
|
| 167 |
+
"step": 4000
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.02,
|
| 171 |
+
"eval_avg_sts": 0.6397488256977145,
|
| 172 |
+
"eval_sickr_spearman": 0.5797635156813201,
|
| 173 |
+
"eval_stsb_spearman": 0.6997341357141089,
|
| 174 |
+
"step": 4000
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.03,
|
| 178 |
+
"eval_avg_sts": 0.6416441729368727,
|
| 179 |
+
"eval_sickr_spearman": 0.5791708648996884,
|
| 180 |
+
"eval_stsb_spearman": 0.7041174809740571,
|
| 181 |
+
"step": 4250
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 0.03,
|
| 185 |
+
"learning_rate": 9.86436304887753e-09,
|
| 186 |
+
"loss": 1.6636,
|
| 187 |
+
"step": 4500
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.03,
|
| 191 |
+
"eval_avg_sts": 0.6439610982515633,
|
| 192 |
+
"eval_sickr_spearman": 0.5797300483430633,
|
| 193 |
+
"eval_stsb_spearman": 0.7081921481600633,
|
| 194 |
+
"step": 4500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.03,
|
| 198 |
+
"eval_avg_sts": 0.6455335224686861,
|
| 199 |
+
"eval_sickr_spearman": 0.5791015895016001,
|
| 200 |
+
"eval_stsb_spearman": 0.7119654554357722,
|
| 201 |
+
"step": 4750
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.03,
|
| 205 |
+
"learning_rate": 9.849292276530588e-09,
|
| 206 |
+
"loss": 1.2653,
|
| 207 |
+
"step": 5000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.03,
|
| 211 |
+
"eval_avg_sts": 0.6473309813927987,
|
| 212 |
+
"eval_sickr_spearman": 0.5790988005567453,
|
| 213 |
+
"eval_stsb_spearman": 0.715563162228852,
|
| 214 |
+
"step": 5000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.03,
|
| 218 |
+
"eval_avg_sts": 0.6491464635996412,
|
| 219 |
+
"eval_sickr_spearman": 0.5793477636876175,
|
| 220 |
+
"eval_stsb_spearman": 0.7189451635116649,
|
| 221 |
+
"step": 5250
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.03,
|
| 225 |
+
"learning_rate": 9.834221504183646e-09,
|
| 226 |
+
"loss": 1.0416,
|
| 227 |
+
"step": 5500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.03,
|
| 231 |
+
"eval_avg_sts": 0.6504410144276549,
|
| 232 |
+
"eval_sickr_spearman": 0.5787376820006436,
|
| 233 |
+
"eval_stsb_spearman": 0.7221443468546662,
|
| 234 |
+
"step": 5500
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.03,
|
| 238 |
+
"eval_avg_sts": 0.6517348476964206,
|
| 239 |
+
"eval_sickr_spearman": 0.578490262749959,
|
| 240 |
+
"eval_stsb_spearman": 0.7249794326428824,
|
| 241 |
+
"step": 5750
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.04,
|
| 245 |
+
"learning_rate": 9.819150731836706e-09,
|
| 246 |
+
"loss": 0.8965,
|
| 247 |
+
"step": 6000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.04,
|
| 251 |
+
"eval_avg_sts": 0.6534361836672715,
|
| 252 |
+
"eval_sickr_spearman": 0.5791566213598945,
|
| 253 |
+
"eval_stsb_spearman": 0.7277157459746485,
|
| 254 |
+
"step": 6000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.04,
|
| 258 |
+
"eval_avg_sts": 0.654605629234608,
|
| 259 |
+
"eval_sickr_spearman": 0.5794011022579643,
|
| 260 |
+
"eval_stsb_spearman": 0.7298101562112517,
|
| 261 |
+
"step": 6250
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.04,
|
| 265 |
+
"learning_rate": 9.804079959489765e-09,
|
| 266 |
+
"loss": 0.7796,
|
| 267 |
+
"step": 6500
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.04,
|
| 271 |
+
"eval_avg_sts": 0.6556384016928898,
|
| 272 |
+
"eval_sickr_spearman": 0.5794165410598388,
|
| 273 |
+
"eval_stsb_spearman": 0.7318602623259409,
|
| 274 |
+
"step": 6500
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.04,
|
| 278 |
+
"eval_avg_sts": 0.6565914169527195,
|
| 279 |
+
"eval_sickr_spearman": 0.5794604669413009,
|
| 280 |
+
"eval_stsb_spearman": 0.7337223669641381,
|
| 281 |
+
"step": 6750
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.04,
|
| 285 |
+
"learning_rate": 9.789009187142823e-09,
|
| 286 |
+
"loss": 0.6669,
|
| 287 |
+
"step": 7000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.04,
|
| 291 |
+
"eval_avg_sts": 0.6580426081739461,
|
| 292 |
+
"eval_sickr_spearman": 0.5804354023783784,
|
| 293 |
+
"eval_stsb_spearman": 0.7356498139695139,
|
| 294 |
+
"step": 7000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.04,
|
| 298 |
+
"eval_avg_sts": 0.659003327547673,
|
| 299 |
+
"eval_sickr_spearman": 0.5809793462302257,
|
| 300 |
+
"eval_stsb_spearman": 0.7370273088651204,
|
| 301 |
+
"step": 7250
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.05,
|
| 305 |
+
"learning_rate": 9.773938414795883e-09,
|
| 306 |
+
"loss": 0.5639,
|
| 307 |
+
"step": 7500
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.05,
|
| 311 |
+
"eval_avg_sts": 0.6599478368878535,
|
| 312 |
+
"eval_sickr_spearman": 0.5816024263923263,
|
| 313 |
+
"eval_stsb_spearman": 0.7382932473833808,
|
| 314 |
+
"step": 7500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.05,
|
| 318 |
+
"eval_avg_sts": 0.6606422753240502,
|
| 319 |
+
"eval_sickr_spearman": 0.581862097079337,
|
| 320 |
+
"eval_stsb_spearman": 0.7394224535687633,
|
| 321 |
+
"step": 7750
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.05,
|
| 325 |
+
"learning_rate": 9.758867642448941e-09,
|
| 326 |
+
"loss": 0.4826,
|
| 327 |
+
"step": 8000
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.05,
|
| 331 |
+
"eval_avg_sts": 0.6615698341083216,
|
| 332 |
+
"eval_sickr_spearman": 0.5826618768190197,
|
| 333 |
+
"eval_stsb_spearman": 0.7404777913976235,
|
| 334 |
+
"step": 8000
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.05,
|
| 338 |
+
"eval_avg_sts": 0.6619288475610956,
|
| 339 |
+
"eval_sickr_spearman": 0.582522877799563,
|
| 340 |
+
"eval_stsb_spearman": 0.7413348173226283,
|
| 341 |
+
"step": 8250
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.05,
|
| 345 |
+
"learning_rate": 9.743796870102e-09,
|
| 346 |
+
"loss": 0.4145,
|
| 347 |
+
"step": 8500
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.05,
|
| 351 |
+
"eval_avg_sts": 0.6623974970827452,
|
| 352 |
+
"eval_sickr_spearman": 0.5825905595148771,
|
| 353 |
+
"eval_stsb_spearman": 0.7422044346506133,
|
| 354 |
+
"step": 8500
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.05,
|
| 358 |
+
"eval_avg_sts": 0.662889528502383,
|
| 359 |
+
"eval_sickr_spearman": 0.5827362320809506,
|
| 360 |
+
"eval_stsb_spearman": 0.7430428249238155,
|
| 361 |
+
"step": 8750
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.05,
|
| 365 |
+
"learning_rate": 9.72872609775506e-09,
|
| 366 |
+
"loss": 0.3629,
|
| 367 |
+
"step": 9000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.05,
|
| 371 |
+
"eval_avg_sts": 0.6636302923347899,
|
| 372 |
+
"eval_sickr_spearman": 0.5834812289752723,
|
| 373 |
+
"eval_stsb_spearman": 0.7437793556943073,
|
| 374 |
+
"step": 9000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.06,
|
| 378 |
+
"eval_avg_sts": 0.664159926773715,
|
| 379 |
+
"eval_sickr_spearman": 0.5837915488929486,
|
| 380 |
+
"eval_stsb_spearman": 0.7445283046544816,
|
| 381 |
+
"step": 9250
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.06,
|
| 385 |
+
"learning_rate": 9.713655325408118e-09,
|
| 386 |
+
"loss": 0.3202,
|
| 387 |
+
"step": 9500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.06,
|
| 391 |
+
"eval_avg_sts": 0.6648401388341261,
|
| 392 |
+
"eval_sickr_spearman": 0.5845315157260146,
|
| 393 |
+
"eval_stsb_spearman": 0.7451487619422374,
|
| 394 |
+
"step": 9500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.06,
|
| 398 |
+
"eval_avg_sts": 0.6654341412381362,
|
| 399 |
+
"eval_sickr_spearman": 0.5849954766236348,
|
| 400 |
+
"eval_stsb_spearman": 0.7458728058526375,
|
| 401 |
+
"step": 9750
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.06,
|
| 405 |
+
"learning_rate": 9.698584553061174e-09,
|
| 406 |
+
"loss": 0.2807,
|
| 407 |
+
"step": 10000
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.06,
|
| 411 |
+
"eval_avg_sts": 0.665940208652456,
|
| 412 |
+
"eval_sickr_spearman": 0.5854395559566877,
|
| 413 |
+
"eval_stsb_spearman": 0.7464408613482243,
|
| 414 |
+
"step": 10000
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.06,
|
| 418 |
+
"eval_avg_sts": 0.6665948829175545,
|
| 419 |
+
"eval_sickr_spearman": 0.5861564745245929,
|
| 420 |
+
"eval_stsb_spearman": 0.7470332913105162,
|
| 421 |
+
"step": 10250
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 0.06,
|
| 425 |
+
"learning_rate": 9.683513780714234e-09,
|
| 426 |
+
"loss": 0.2452,
|
| 427 |
+
"step": 10500
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.06,
|
| 431 |
+
"eval_avg_sts": 0.6671936027854921,
|
| 432 |
+
"eval_sickr_spearman": 0.5865762605278176,
|
| 433 |
+
"eval_stsb_spearman": 0.7478109450431667,
|
| 434 |
+
"step": 10500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.06,
|
| 438 |
+
"eval_avg_sts": 0.6677066615222154,
|
| 439 |
+
"eval_sickr_spearman": 0.5870991378854943,
|
| 440 |
+
"eval_stsb_spearman": 0.7483141851589363,
|
| 441 |
+
"step": 10750
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.07,
|
| 445 |
+
"learning_rate": 9.668443008367292e-09,
|
| 446 |
+
"loss": 0.2158,
|
| 447 |
+
"step": 11000
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.07,
|
| 451 |
+
"eval_avg_sts": 0.668111446446719,
|
| 452 |
+
"eval_sickr_spearman": 0.5873266361015022,
|
| 453 |
+
"eval_stsb_spearman": 0.748896256791936,
|
| 454 |
+
"step": 11000
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 0.07,
|
| 458 |
+
"eval_avg_sts": 0.6684118774641665,
|
| 459 |
+
"eval_sickr_spearman": 0.5873295246515303,
|
| 460 |
+
"eval_stsb_spearman": 0.7494942302768027,
|
| 461 |
+
"step": 11250
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.07,
|
| 465 |
+
"learning_rate": 9.65337223602035e-09,
|
| 466 |
+
"loss": 0.191,
|
| 467 |
+
"step": 11500
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.07,
|
| 471 |
+
"eval_avg_sts": 0.6685451430874475,
|
| 472 |
+
"eval_sickr_spearman": 0.5870249320313236,
|
| 473 |
+
"eval_stsb_spearman": 0.7500653541435714,
|
| 474 |
+
"step": 11500
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.07,
|
| 478 |
+
"eval_avg_sts": 0.6688289293567933,
|
| 479 |
+
"eval_sickr_spearman": 0.5870888785526357,
|
| 480 |
+
"eval_stsb_spearman": 0.7505689801609509,
|
| 481 |
+
"step": 11750
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 0.07,
|
| 485 |
+
"learning_rate": 9.63830146367341e-09,
|
| 486 |
+
"loss": 0.1703,
|
| 487 |
+
"step": 12000
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.07,
|
| 491 |
+
"eval_avg_sts": 0.6689011931208393,
|
| 492 |
+
"eval_sickr_spearman": 0.5867304991388019,
|
| 493 |
+
"eval_stsb_spearman": 0.7510718871028766,
|
| 494 |
+
"step": 12000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.07,
|
| 498 |
+
"eval_avg_sts": 0.6691967252591424,
|
| 499 |
+
"eval_sickr_spearman": 0.5868564498805454,
|
| 500 |
+
"eval_stsb_spearman": 0.7515370006377393,
|
| 501 |
+
"step": 12250
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.08,
|
| 505 |
+
"learning_rate": 9.623230691326469e-09,
|
| 506 |
+
"loss": 0.1528,
|
| 507 |
+
"step": 12500
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.08,
|
| 511 |
+
"eval_avg_sts": 0.6694502557179467,
|
| 512 |
+
"eval_sickr_spearman": 0.5868579937607329,
|
| 513 |
+
"eval_stsb_spearman": 0.7520425176751604,
|
| 514 |
+
"step": 12500
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.08,
|
| 518 |
+
"eval_avg_sts": 0.6699632588811146,
|
| 519 |
+
"eval_sickr_spearman": 0.587389835584014,
|
| 520 |
+
"eval_stsb_spearman": 0.7525366821782152,
|
| 521 |
+
"step": 12750
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.08,
|
| 525 |
+
"learning_rate": 9.608159918979527e-09,
|
| 526 |
+
"loss": 0.1371,
|
| 527 |
+
"step": 13000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.08,
|
| 531 |
+
"eval_avg_sts": 0.6703483303011183,
|
| 532 |
+
"eval_sickr_spearman": 0.5876732123022901,
|
| 533 |
+
"eval_stsb_spearman": 0.7530234482999465,
|
| 534 |
+
"step": 13000
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.08,
|
| 538 |
+
"eval_avg_sts": 0.6707168479183498,
|
| 539 |
+
"eval_sickr_spearman": 0.5879633123697694,
|
| 540 |
+
"eval_stsb_spearman": 0.7534703834669302,
|
| 541 |
+
"step": 13250
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.08,
|
| 545 |
+
"learning_rate": 9.593089146632587e-09,
|
| 546 |
+
"loss": 0.1252,
|
| 547 |
+
"step": 13500
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.08,
|
| 551 |
+
"eval_avg_sts": 0.6712428310314182,
|
| 552 |
+
"eval_sickr_spearman": 0.5885554651255342,
|
| 553 |
+
"eval_stsb_spearman": 0.7539301969373022,
|
| 554 |
+
"step": 13500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.08,
|
| 558 |
+
"eval_avg_sts": 0.6715811760730475,
|
| 559 |
+
"eval_sickr_spearman": 0.5886914759897894,
|
| 560 |
+
"eval_stsb_spearman": 0.7544708761563054,
|
| 561 |
+
"step": 13750
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.08,
|
| 565 |
+
"learning_rate": 9.578018374285645e-09,
|
| 566 |
+
"loss": 0.1124,
|
| 567 |
+
"step": 14000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.08,
|
| 571 |
+
"eval_avg_sts": 0.6720556074865778,
|
| 572 |
+
"eval_sickr_spearman": 0.5892022513188999,
|
| 573 |
+
"eval_stsb_spearman": 0.7549089636542557,
|
| 574 |
+
"step": 14000
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 0.09,
|
| 578 |
+
"eval_avg_sts": 0.6723349901448149,
|
| 579 |
+
"eval_sickr_spearman": 0.5893755145180006,
|
| 580 |
+
"eval_stsb_spearman": 0.7552944657716292,
|
| 581 |
+
"step": 14250
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 0.09,
|
| 585 |
+
"learning_rate": 9.562947601938704e-09,
|
| 586 |
+
"loss": 0.1028,
|
| 587 |
+
"step": 14500
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.09,
|
| 591 |
+
"eval_avg_sts": 0.6726527822394905,
|
| 592 |
+
"eval_sickr_spearman": 0.589669150569135,
|
| 593 |
+
"eval_stsb_spearman": 0.755636413909846,
|
| 594 |
+
"step": 14500
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.09,
|
| 598 |
+
"eval_avg_sts": 0.6730171959193298,
|
| 599 |
+
"eval_sickr_spearman": 0.5900167228216568,
|
| 600 |
+
"eval_stsb_spearman": 0.756017669017003,
|
| 601 |
+
"step": 14750
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.09,
|
| 605 |
+
"learning_rate": 9.547876829591762e-09,
|
| 606 |
+
"loss": 0.094,
|
| 607 |
+
"step": 15000
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.09,
|
| 611 |
+
"eval_avg_sts": 0.6735421112570777,
|
| 612 |
+
"eval_sickr_spearman": 0.5906376116699428,
|
| 613 |
+
"eval_stsb_spearman": 0.7564466108442127,
|
| 614 |
+
"step": 15000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.09,
|
| 618 |
+
"eval_avg_sts": 0.6735606956141224,
|
| 619 |
+
"eval_sickr_spearman": 0.5903632990224448,
|
| 620 |
+
"eval_stsb_spearman": 0.7567580922057999,
|
| 621 |
+
"step": 15250
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.09,
|
| 625 |
+
"learning_rate": 9.532806057244822e-09,
|
| 626 |
+
"loss": 0.0868,
|
| 627 |
+
"step": 15500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.09,
|
| 631 |
+
"eval_avg_sts": 0.6737977477369647,
|
| 632 |
+
"eval_sickr_spearman": 0.5904703745838319,
|
| 633 |
+
"eval_stsb_spearman": 0.7571251208900974,
|
| 634 |
+
"step": 15500
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.09,
|
| 638 |
+
"eval_avg_sts": 0.6742547600358272,
|
| 639 |
+
"eval_sickr_spearman": 0.591002216407113,
|
| 640 |
+
"eval_stsb_spearman": 0.7575073036645413,
|
| 641 |
+
"step": 15750
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.1,
|
| 645 |
+
"learning_rate": 9.51773528489788e-09,
|
| 646 |
+
"loss": 0.08,
|
| 647 |
+
"step": 16000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.1,
|
| 651 |
+
"eval_avg_sts": 0.6744646822989027,
|
| 652 |
+
"eval_sickr_spearman": 0.5910811535070195,
|
| 653 |
+
"eval_stsb_spearman": 0.7578482110907857,
|
| 654 |
+
"step": 16000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.1,
|
| 658 |
+
"eval_avg_sts": 0.6748799475614093,
|
| 659 |
+
"eval_sickr_spearman": 0.5915160794965986,
|
| 660 |
+
"eval_stsb_spearman": 0.7582438156262201,
|
| 661 |
+
"step": 16250
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.1,
|
| 665 |
+
"learning_rate": 9.502664512550939e-09,
|
| 666 |
+
"loss": 0.0728,
|
| 667 |
+
"step": 16500
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.1,
|
| 671 |
+
"eval_avg_sts": 0.6751155830172513,
|
| 672 |
+
"eval_sickr_spearman": 0.5916266412390543,
|
| 673 |
+
"eval_stsb_spearman": 0.7586045247954482,
|
| 674 |
+
"step": 16500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.1,
|
| 678 |
+
"eval_avg_sts": 0.6754758089125048,
|
| 679 |
+
"eval_sickr_spearman": 0.5920287971265901,
|
| 680 |
+
"eval_stsb_spearman": 0.7589228206984197,
|
| 681 |
+
"step": 16750
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.1,
|
| 685 |
+
"learning_rate": 9.487593740203998e-09,
|
| 686 |
+
"loss": 0.0679,
|
| 687 |
+
"step": 17000
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.1,
|
| 691 |
+
"eval_avg_sts": 0.6760131658651831,
|
| 692 |
+
"eval_sickr_spearman": 0.5928007372203121,
|
| 693 |
+
"eval_stsb_spearman": 0.7592255945100542,
|
| 694 |
+
"step": 17000
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.1,
|
| 698 |
+
"eval_avg_sts": 0.6762108045214816,
|
| 699 |
+
"eval_sickr_spearman": 0.5929441686699845,
|
| 700 |
+
"eval_stsb_spearman": 0.7594774403729787,
|
| 701 |
+
"step": 17250
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 0.11,
|
| 705 |
+
"learning_rate": 9.472522967857057e-09,
|
| 706 |
+
"loss": 0.0627,
|
| 707 |
+
"step": 17500
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 0.11,
|
| 711 |
+
"eval_avg_sts": 0.6766050515480309,
|
| 712 |
+
"eval_sickr_spearman": 0.5934244648160396,
|
| 713 |
+
"eval_stsb_spearman": 0.7597856382800222,
|
| 714 |
+
"step": 17500
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.11,
|
| 718 |
+
"eval_avg_sts": 0.6767814381948531,
|
| 719 |
+
"eval_sickr_spearman": 0.5934582807724034,
|
| 720 |
+
"eval_stsb_spearman": 0.7601045956173028,
|
| 721 |
+
"step": 17750
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.11,
|
| 725 |
+
"learning_rate": 9.457452195510115e-09,
|
| 726 |
+
"loss": 0.0584,
|
| 727 |
+
"step": 18000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.11,
|
| 731 |
+
"eval_avg_sts": 0.6769896075076911,
|
| 732 |
+
"eval_sickr_spearman": 0.5936528096760212,
|
| 733 |
+
"eval_stsb_spearman": 0.7603264053393611,
|
| 734 |
+
"step": 18000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.11,
|
| 738 |
+
"eval_avg_sts": 0.677209761262622,
|
| 739 |
+
"eval_sickr_spearman": 0.593758341357221,
|
| 740 |
+
"eval_stsb_spearman": 0.7606611811680228,
|
| 741 |
+
"step": 18250
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.11,
|
| 745 |
+
"learning_rate": 9.442381423163175e-09,
|
| 746 |
+
"loss": 0.0534,
|
| 747 |
+
"step": 18500
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.11,
|
| 751 |
+
"eval_avg_sts": 0.6775168218365983,
|
| 752 |
+
"eval_sickr_spearman": 0.5941294702332478,
|
| 753 |
+
"eval_stsb_spearman": 0.7609041734399488,
|
| 754 |
+
"step": 18500
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.11,
|
| 758 |
+
"eval_avg_sts": 0.6777132242888597,
|
| 759 |
+
"eval_sickr_spearman": 0.5942964085038387,
|
| 760 |
+
"eval_stsb_spearman": 0.7611300400738806,
|
| 761 |
+
"step": 18750
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.11,
|
| 765 |
+
"learning_rate": 9.427310650816233e-09,
|
| 766 |
+
"loss": 0.0507,
|
| 767 |
+
"step": 19000
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.11,
|
| 771 |
+
"eval_avg_sts": 0.6778013660288258,
|
| 772 |
+
"eval_sickr_spearman": 0.594225439817803,
|
| 773 |
+
"eval_stsb_spearman": 0.7613772922398485,
|
| 774 |
+
"step": 19000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.12,
|
| 778 |
+
"eval_avg_sts": 0.6781823053409229,
|
| 779 |
+
"eval_sickr_spearman": 0.594783129183577,
|
| 780 |
+
"eval_stsb_spearman": 0.7615814814982689,
|
| 781 |
+
"step": 19250
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 0.12,
|
| 785 |
+
"learning_rate": 9.412239878469292e-09,
|
| 786 |
+
"loss": 0.0472,
|
| 787 |
+
"step": 19500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.12,
|
| 791 |
+
"eval_avg_sts": 0.6783265737471122,
|
| 792 |
+
"eval_sickr_spearman": 0.5948400037375791,
|
| 793 |
+
"eval_stsb_spearman": 0.7618131437566455,
|
| 794 |
+
"step": 19500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.12,
|
| 798 |
+
"eval_avg_sts": 0.6784263694342607,
|
| 799 |
+
"eval_sickr_spearman": 0.5948344258478697,
|
| 800 |
+
"eval_stsb_spearman": 0.7620183130206516,
|
| 801 |
+
"step": 19750
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.12,
|
| 805 |
+
"learning_rate": 9.397169106122352e-09,
|
| 806 |
+
"loss": 0.0433,
|
| 807 |
+
"step": 20000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.12,
|
| 811 |
+
"eval_avg_sts": 0.6786123514092111,
|
| 812 |
+
"eval_sickr_spearman": 0.5949314412867451,
|
| 813 |
+
"eval_stsb_spearman": 0.7622932615316771,
|
| 814 |
+
"step": 20000
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.12,
|
| 818 |
+
"eval_avg_sts": 0.6787803024178876,
|
| 819 |
+
"eval_sickr_spearman": 0.5951198444722,
|
| 820 |
+
"eval_stsb_spearman": 0.762440760363575,
|
| 821 |
+
"step": 20250
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.12,
|
| 825 |
+
"learning_rate": 9.38209833377541e-09,
|
| 826 |
+
"loss": 0.0414,
|
| 827 |
+
"step": 20500
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 0.12,
|
| 831 |
+
"eval_avg_sts": 0.679031627015686,
|
| 832 |
+
"eval_sickr_spearman": 0.5953973942878331,
|
| 833 |
+
"eval_stsb_spearman": 0.7626658597435387,
|
| 834 |
+
"step": 20500
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.13,
|
| 838 |
+
"eval_avg_sts": 0.6793044898299303,
|
| 839 |
+
"eval_sickr_spearman": 0.5957000944097455,
|
| 840 |
+
"eval_stsb_spearman": 0.7629088852501152,
|
| 841 |
+
"step": 20750
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.13,
|
| 845 |
+
"learning_rate": 9.367027561428468e-09,
|
| 846 |
+
"loss": 0.0391,
|
| 847 |
+
"step": 21000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 0.13,
|
| 851 |
+
"eval_avg_sts": 0.6796013133058254,
|
| 852 |
+
"eval_sickr_spearman": 0.5961489651235983,
|
| 853 |
+
"eval_stsb_spearman": 0.7630536614880524,
|
| 854 |
+
"step": 21000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.13,
|
| 858 |
+
"eval_avg_sts": 0.6797676755235869,
|
| 859 |
+
"eval_sickr_spearman": 0.5963239216106464,
|
| 860 |
+
"eval_stsb_spearman": 0.7632114294365272,
|
| 861 |
+
"step": 21250
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.13,
|
| 865 |
+
"learning_rate": 9.351956789081526e-09,
|
| 866 |
+
"loss": 0.0368,
|
| 867 |
+
"step": 21500
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.13,
|
| 871 |
+
"eval_avg_sts": 0.6799738806786508,
|
| 872 |
+
"eval_sickr_spearman": 0.5966217410790631,
|
| 873 |
+
"eval_stsb_spearman": 0.7633260202782385,
|
| 874 |
+
"step": 21500
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.13,
|
| 878 |
+
"eval_avg_sts": 0.6802068401411112,
|
| 879 |
+
"eval_sickr_spearman": 0.5968975976067485,
|
| 880 |
+
"eval_stsb_spearman": 0.763516082675474,
|
| 881 |
+
"step": 21750
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.13,
|
| 885 |
+
"learning_rate": 9.336886016734586e-09,
|
| 886 |
+
"loss": 0.0347,
|
| 887 |
+
"step": 22000
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.13,
|
| 891 |
+
"eval_avg_sts": 0.6802982007395906,
|
| 892 |
+
"eval_sickr_spearman": 0.5969305669191386,
|
| 893 |
+
"eval_stsb_spearman": 0.7636658345600424,
|
| 894 |
+
"step": 22000
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.13,
|
| 898 |
+
"eval_avg_sts": 0.680360206773362,
|
| 899 |
+
"eval_sickr_spearman": 0.5969050679947523,
|
| 900 |
+
"eval_stsb_spearman": 0.7638153455519716,
|
| 901 |
+
"step": 22250
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.14,
|
| 905 |
+
"learning_rate": 9.321815244387645e-09,
|
| 906 |
+
"loss": 0.0329,
|
| 907 |
+
"step": 22500
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 0.14,
|
| 911 |
+
"eval_avg_sts": 0.6804424688694628,
|
| 912 |
+
"eval_sickr_spearman": 0.596870206184068,
|
| 913 |
+
"eval_stsb_spearman": 0.7640147315548577,
|
| 914 |
+
"step": 22500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.14,
|
| 918 |
+
"eval_avg_sts": 0.6804860386341411,
|
| 919 |
+
"eval_sickr_spearman": 0.5967837488935713,
|
| 920 |
+
"eval_stsb_spearman": 0.764188328374711,
|
| 921 |
+
"step": 22750
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.14,
|
| 925 |
+
"learning_rate": 9.306744472040703e-09,
|
| 926 |
+
"loss": 0.0311,
|
| 927 |
+
"step": 23000
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.14,
|
| 931 |
+
"eval_avg_sts": 0.6805827014319539,
|
| 932 |
+
"eval_sickr_spearman": 0.5968390795673858,
|
| 933 |
+
"eval_stsb_spearman": 0.7643263232965221,
|
| 934 |
+
"step": 23000
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.14,
|
| 938 |
+
"eval_avg_sts": 0.6806972200568748,
|
| 939 |
+
"eval_sickr_spearman": 0.5969168712077983,
|
| 940 |
+
"eval_stsb_spearman": 0.7644775689059513,
|
| 941 |
+
"step": 23250
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.14,
|
| 945 |
+
"learning_rate": 9.291673699693763e-09,
|
| 946 |
+
"loss": 0.0297,
|
| 947 |
+
"step": 23500
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.14,
|
| 951 |
+
"eval_avg_sts": 0.6808736496467026,
|
| 952 |
+
"eval_sickr_spearman": 0.5970506409556523,
|
| 953 |
+
"eval_stsb_spearman": 0.7646966583377529,
|
| 954 |
+
"step": 23500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.14,
|
| 958 |
+
"eval_avg_sts": 0.6810096396199108,
|
| 959 |
+
"eval_sickr_spearman": 0.5971779363672365,
|
| 960 |
+
"eval_stsb_spearman": 0.7648413428725852,
|
| 961 |
+
"step": 23750
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.14,
|
| 965 |
+
"learning_rate": 9.276602927346821e-09,
|
| 966 |
+
"loss": 0.0284,
|
| 967 |
+
"step": 24000
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.14,
|
| 971 |
+
"eval_avg_sts": 0.6811351307194636,
|
| 972 |
+
"eval_sickr_spearman": 0.5972490693237311,
|
| 973 |
+
"eval_stsb_spearman": 0.7650211921151961,
|
| 974 |
+
"step": 24000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.15,
|
| 978 |
+
"eval_avg_sts": 0.6813719782830303,
|
| 979 |
+
"eval_sickr_spearman": 0.5975800424521857,
|
| 980 |
+
"eval_stsb_spearman": 0.7651639141138749,
|
| 981 |
+
"step": 24250
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.15,
|
| 985 |
+
"learning_rate": 9.26153215499988e-09,
|
| 986 |
+
"loss": 0.027,
|
| 987 |
+
"step": 24500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.15,
|
| 991 |
+
"eval_avg_sts": 0.6815488144179666,
|
| 992 |
+
"eval_sickr_spearman": 0.5977527578228325,
|
| 993 |
+
"eval_stsb_spearman": 0.7653448710131006,
|
| 994 |
+
"step": 24500
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.15,
|
| 998 |
+
"eval_avg_sts": 0.6816865586733576,
|
| 999 |
+
"eval_sickr_spearman": 0.5979984339829829,
|
| 1000 |
+
"eval_stsb_spearman": 0.7653746833637323,
|
| 1001 |
+
"step": 24750
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.15,
|
| 1005 |
+
"learning_rate": 9.24646138265294e-09,
|
| 1006 |
+
"loss": 0.026,
|
| 1007 |
+
"step": 25000
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 0.15,
|
| 1011 |
+
"eval_avg_sts": 0.681900742705164,
|
| 1012 |
+
"eval_sickr_spearman": 0.5982624872976225,
|
| 1013 |
+
"eval_stsb_spearman": 0.7655389981127053,
|
| 1014 |
+
"step": 25000
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.15,
|
| 1018 |
+
"eval_avg_sts": 0.6819519221717609,
|
| 1019 |
+
"eval_sickr_spearman": 0.5982854462900874,
|
| 1020 |
+
"eval_stsb_spearman": 0.7656183980534346,
|
| 1021 |
+
"step": 25250
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 0.15,
|
| 1025 |
+
"learning_rate": 9.231390610305998e-09,
|
| 1026 |
+
"loss": 0.0247,
|
| 1027 |
+
"step": 25500
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 0.15,
|
| 1031 |
+
"eval_avg_sts": 0.6819903776778287,
|
| 1032 |
+
"eval_sickr_spearman": 0.5982351456775288,
|
| 1033 |
+
"eval_stsb_spearman": 0.7657456096781285,
|
| 1034 |
+
"step": 25500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.16,
|
| 1038 |
+
"eval_avg_sts": 0.6821595590267062,
|
| 1039 |
+
"eval_sickr_spearman": 0.5985327659355987,
|
| 1040 |
+
"eval_stsb_spearman": 0.7657863521178138,
|
| 1041 |
+
"step": 25750
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.16,
|
| 1045 |
+
"learning_rate": 9.216319837959056e-09,
|
| 1046 |
+
"loss": 0.0238,
|
| 1047 |
+
"step": 26000
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.16,
|
| 1051 |
+
"eval_avg_sts": 0.6822216826747098,
|
| 1052 |
+
"eval_sickr_spearman": 0.5984852542678948,
|
| 1053 |
+
"eval_stsb_spearman": 0.7659581110815249,
|
| 1054 |
+
"step": 26000
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.16,
|
| 1058 |
+
"eval_avg_sts": 0.6822324678023577,
|
| 1059 |
+
"eval_sickr_spearman": 0.5984695664530869,
|
| 1060 |
+
"eval_stsb_spearman": 0.7659953691516284,
|
| 1061 |
+
"step": 26250
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 0.16,
|
| 1065 |
+
"learning_rate": 9.201249065612116e-09,
|
| 1066 |
+
"loss": 0.0226,
|
| 1067 |
+
"step": 26500
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.16,
|
| 1071 |
+
"eval_avg_sts": 0.6824471355452602,
|
| 1072 |
+
"eval_sickr_spearman": 0.598771668943959,
|
| 1073 |
+
"eval_stsb_spearman": 0.7661226021465615,
|
| 1074 |
+
"step": 26500
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.16,
|
| 1078 |
+
"eval_avg_sts": 0.6825452027323136,
|
| 1079 |
+
"eval_sickr_spearman": 0.5989038948116256,
|
| 1080 |
+
"eval_stsb_spearman": 0.7661865106530016,
|
| 1081 |
+
"step": 26750
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.16,
|
| 1085 |
+
"learning_rate": 9.186178293265174e-09,
|
| 1086 |
+
"loss": 0.0219,
|
| 1087 |
+
"step": 27000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.16,
|
| 1091 |
+
"eval_avg_sts": 0.682652241260723,
|
| 1092 |
+
"eval_sickr_spearman": 0.5990028027487954,
|
| 1093 |
+
"eval_stsb_spearman": 0.7663016797726506,
|
| 1094 |
+
"step": 27000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.16,
|
| 1098 |
+
"eval_avg_sts": 0.6827966424642413,
|
| 1099 |
+
"eval_sickr_spearman": 0.5991921521833974,
|
| 1100 |
+
"eval_stsb_spearman": 0.7664011327450853,
|
| 1101 |
+
"step": 27250
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.17,
|
| 1105 |
+
"learning_rate": 9.171107520918232e-09,
|
| 1106 |
+
"loss": 0.0208,
|
| 1107 |
+
"step": 27500
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.17,
|
| 1111 |
+
"eval_avg_sts": 0.6828478001704832,
|
| 1112 |
+
"eval_sickr_spearman": 0.5991956881670526,
|
| 1113 |
+
"eval_stsb_spearman": 0.7664999121739138,
|
| 1114 |
+
"step": 27500
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.17,
|
| 1118 |
+
"eval_avg_sts": 0.6830724849298578,
|
| 1119 |
+
"eval_sickr_spearman": 0.5994928103992555,
|
| 1120 |
+
"eval_stsb_spearman": 0.7666521594604602,
|
| 1121 |
+
"step": 27750
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.17,
|
| 1125 |
+
"learning_rate": 9.15603674857129e-09,
|
| 1126 |
+
"loss": 0.02,
|
| 1127 |
+
"step": 28000
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 0.17,
|
| 1131 |
+
"eval_avg_sts": 0.6830958603489119,
|
| 1132 |
+
"eval_sickr_spearman": 0.5994919637552817,
|
| 1133 |
+
"eval_stsb_spearman": 0.7666997569425422,
|
| 1134 |
+
"step": 28000
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.17,
|
| 1138 |
+
"eval_avg_sts": 0.6830795409925201,
|
| 1139 |
+
"eval_sickr_spearman": 0.5993960439733135,
|
| 1140 |
+
"eval_stsb_spearman": 0.7667630380117267,
|
| 1141 |
+
"step": 28250
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 0.17,
|
| 1145 |
+
"learning_rate": 9.14096597622435e-09,
|
| 1146 |
+
"loss": 0.0195,
|
| 1147 |
+
"step": 28500
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.17,
|
| 1151 |
+
"eval_avg_sts": 0.6831755132481567,
|
| 1152 |
+
"eval_sickr_spearman": 0.5994744332447662,
|
| 1153 |
+
"eval_stsb_spearman": 0.7668765932515471,
|
| 1154 |
+
"step": 28500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.17,
|
| 1158 |
+
"eval_avg_sts": 0.6833569004212535,
|
| 1159 |
+
"eval_sickr_spearman": 0.5997113441496589,
|
| 1160 |
+
"eval_stsb_spearman": 0.7670024566928482,
|
| 1161 |
+
"step": 28750
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.17,
|
| 1165 |
+
"learning_rate": 9.125895203877409e-09,
|
| 1166 |
+
"loss": 0.0187,
|
| 1167 |
+
"step": 29000
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.17,
|
| 1171 |
+
"eval_avg_sts": 0.6834568554940637,
|
| 1172 |
+
"eval_sickr_spearman": 0.5998296750956386,
|
| 1173 |
+
"eval_stsb_spearman": 0.7670840358924887,
|
| 1174 |
+
"step": 29000
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 0.18,
|
| 1178 |
+
"eval_avg_sts": 0.6834650699517616,
|
| 1179 |
+
"eval_sickr_spearman": 0.5997757388942514,
|
| 1180 |
+
"eval_stsb_spearman": 0.7671544010092718,
|
| 1181 |
+
"step": 29250
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 0.18,
|
| 1185 |
+
"learning_rate": 9.110824431530467e-09,
|
| 1186 |
+
"loss": 0.0179,
|
| 1187 |
+
"step": 29500
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 0.18,
|
| 1191 |
+
"eval_avg_sts": 0.6835179845459645,
|
| 1192 |
+
"eval_sickr_spearman": 0.5998240474033423,
|
| 1193 |
+
"eval_stsb_spearman": 0.7672119216885865,
|
| 1194 |
+
"step": 29500
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.18,
|
| 1198 |
+
"eval_avg_sts": 0.6835631347255009,
|
| 1199 |
+
"eval_sickr_spearman": 0.5997912773012991,
|
| 1200 |
+
"eval_stsb_spearman": 0.7673349921497027,
|
| 1201 |
+
"step": 29750
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.18,
|
| 1205 |
+
"learning_rate": 9.095753659183525e-09,
|
| 1206 |
+
"loss": 0.0176,
|
| 1207 |
+
"step": 30000
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.18,
|
| 1211 |
+
"eval_avg_sts": 0.6836041595446583,
|
| 1212 |
+
"eval_sickr_spearman": 0.5998075627471473,
|
| 1213 |
+
"eval_stsb_spearman": 0.7674007563421691,
|
| 1214 |
+
"step": 30000
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.18,
|
| 1218 |
+
"eval_avg_sts": 0.6837299737321467,
|
| 1219 |
+
"eval_sickr_spearman": 0.6000278395880849,
|
| 1220 |
+
"eval_stsb_spearman": 0.7674321078762084,
|
| 1221 |
+
"step": 30250
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.18,
|
| 1225 |
+
"learning_rate": 9.080682886836584e-09,
|
| 1226 |
+
"loss": 0.017,
|
| 1227 |
+
"step": 30500
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.18,
|
| 1231 |
+
"eval_avg_sts": 0.6837571024354734,
|
| 1232 |
+
"eval_sickr_spearman": 0.5999885951497719,
|
| 1233 |
+
"eval_stsb_spearman": 0.7675256097211749,
|
| 1234 |
+
"step": 30500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.19,
|
| 1238 |
+
"eval_avg_sts": 0.6838406104965742,
|
| 1239 |
+
"eval_sickr_spearman": 0.6001280423925087,
|
| 1240 |
+
"eval_stsb_spearman": 0.7675531786006398,
|
| 1241 |
+
"step": 30750
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.19,
|
| 1245 |
+
"learning_rate": 9.065612114489644e-09,
|
| 1246 |
+
"loss": 0.0164,
|
| 1247 |
+
"step": 31000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 0.19,
|
| 1251 |
+
"eval_avg_sts": 0.6839781957977387,
|
| 1252 |
+
"eval_sickr_spearman": 0.6003198321538586,
|
| 1253 |
+
"eval_stsb_spearman": 0.7676365594416187,
|
| 1254 |
+
"step": 31000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 0.19,
|
| 1258 |
+
"eval_avg_sts": 0.6841738441816796,
|
| 1259 |
+
"eval_sickr_spearman": 0.6006274129292668,
|
| 1260 |
+
"eval_stsb_spearman": 0.7677202754340923,
|
| 1261 |
+
"step": 31250
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.19,
|
| 1265 |
+
"learning_rate": 9.050541342142702e-09,
|
| 1266 |
+
"loss": 0.0159,
|
| 1267 |
+
"step": 31500
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 0.19,
|
| 1271 |
+
"eval_avg_sts": 0.6842156611206922,
|
| 1272 |
+
"eval_sickr_spearman": 0.600646138701863,
|
| 1273 |
+
"eval_stsb_spearman": 0.7677851835395214,
|
| 1274 |
+
"step": 31500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.19,
|
| 1278 |
+
"eval_avg_sts": 0.6841708520756637,
|
| 1279 |
+
"eval_sickr_spearman": 0.600490057395171,
|
| 1280 |
+
"eval_stsb_spearman": 0.7678516467561564,
|
| 1281 |
+
"step": 31750
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.19,
|
| 1285 |
+
"learning_rate": 9.03547056979576e-09,
|
| 1286 |
+
"loss": 0.0153,
|
| 1287 |
+
"step": 32000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.19,
|
| 1291 |
+
"eval_avg_sts": 0.6840981881424868,
|
| 1292 |
+
"eval_sickr_spearman": 0.6003198321538586,
|
| 1293 |
+
"eval_stsb_spearman": 0.7678765441311148,
|
| 1294 |
+
"step": 32000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 0.19,
|
| 1298 |
+
"eval_avg_sts": 0.6842101456597175,
|
| 1299 |
+
"eval_sickr_spearman": 0.6004448366464549,
|
| 1300 |
+
"eval_stsb_spearman": 0.7679754546729801,
|
| 1301 |
+
"step": 32250
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 0.2,
|
| 1305 |
+
"learning_rate": 9.02039979744882e-09,
|
| 1306 |
+
"loss": 0.015,
|
| 1307 |
+
"step": 32500
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 0.2,
|
| 1311 |
+
"eval_avg_sts": 0.684201669305099,
|
| 1312 |
+
"eval_sickr_spearman": 0.6004205329841494,
|
| 1313 |
+
"eval_stsb_spearman": 0.7679828056260486,
|
| 1314 |
+
"step": 32500
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 0.2,
|
| 1318 |
+
"eval_avg_sts": 0.6843570539386391,
|
| 1319 |
+
"eval_sickr_spearman": 0.600673928545237,
|
| 1320 |
+
"eval_stsb_spearman": 0.7680401793320412,
|
| 1321 |
+
"step": 32750
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.2,
|
| 1325 |
+
"learning_rate": 9.005329025101879e-09,
|
| 1326 |
+
"loss": 0.0143,
|
| 1327 |
+
"step": 33000
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.2,
|
| 1331 |
+
"eval_avg_sts": 0.6844358440678995,
|
| 1332 |
+
"eval_sickr_spearman": 0.6008083955293048,
|
| 1333 |
+
"eval_stsb_spearman": 0.7680632926064943,
|
| 1334 |
+
"step": 33000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.2,
|
| 1338 |
+
"eval_avg_sts": 0.6845585283651245,
|
| 1339 |
+
"eval_sickr_spearman": 0.6010039702872432,
|
| 1340 |
+
"eval_stsb_spearman": 0.7681130864430059,
|
| 1341 |
+
"step": 33250
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.2,
|
| 1345 |
+
"learning_rate": 8.990258252754937e-09,
|
| 1346 |
+
"loss": 0.014,
|
| 1347 |
+
"step": 33500
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.2,
|
| 1351 |
+
"eval_avg_sts": 0.6845940476732861,
|
| 1352 |
+
"eval_sickr_spearman": 0.6010599483946848,
|
| 1353 |
+
"eval_stsb_spearman": 0.7681281469518876,
|
| 1354 |
+
"step": 33500
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.2,
|
| 1358 |
+
"eval_avg_sts": 0.6846217305035208,
|
| 1359 |
+
"eval_sickr_spearman": 0.6010828575845629,
|
| 1360 |
+
"eval_stsb_spearman": 0.7681606034224787,
|
| 1361 |
+
"step": 33750
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.2,
|
| 1365 |
+
"learning_rate": 8.975187480407995e-09,
|
| 1366 |
+
"loss": 0.0135,
|
| 1367 |
+
"step": 34000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.2,
|
| 1371 |
+
"eval_avg_sts": 0.6847057396162426,
|
| 1372 |
+
"eval_sickr_spearman": 0.6011810184829323,
|
| 1373 |
+
"eval_stsb_spearman": 0.7682304607495529,
|
| 1374 |
+
"step": 34000
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.21,
|
| 1378 |
+
"eval_avg_sts": 0.684865672317778,
|
| 1379 |
+
"eval_sickr_spearman": 0.6015010001024268,
|
| 1380 |
+
"eval_stsb_spearman": 0.768230344533129,
|
| 1381 |
+
"step": 34250
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.21,
|
| 1385 |
+
"learning_rate": 8.960116708061055e-09,
|
| 1386 |
+
"loss": 0.013,
|
| 1387 |
+
"step": 34500
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 0.21,
|
| 1391 |
+
"eval_avg_sts": 0.6849459219282323,
|
| 1392 |
+
"eval_sickr_spearman": 0.6016653984410962,
|
| 1393 |
+
"eval_stsb_spearman": 0.7682264454153683,
|
| 1394 |
+
"step": 34500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.21,
|
| 1398 |
+
"eval_avg_sts": 0.684890754096638,
|
| 1399 |
+
"eval_sickr_spearman": 0.6015818297006276,
|
| 1400 |
+
"eval_stsb_spearman": 0.7681996784926486,
|
| 1401 |
+
"step": 34750
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.21,
|
| 1405 |
+
"learning_rate": 8.945045935714113e-09,
|
| 1406 |
+
"loss": 0.0127,
|
| 1407 |
+
"step": 35000
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 0.21,
|
| 1411 |
+
"eval_avg_sts": 0.6848642701791275,
|
| 1412 |
+
"eval_sickr_spearman": 0.6015226642276378,
|
| 1413 |
+
"eval_stsb_spearman": 0.7682058761306173,
|
| 1414 |
+
"step": 35000
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.21,
|
| 1418 |
+
"eval_avg_sts": 0.6848120976692862,
|
| 1419 |
+
"eval_sickr_spearman": 0.6013494010285371,
|
| 1420 |
+
"eval_stsb_spearman": 0.7682747943100354,
|
| 1421 |
+
"step": 35250
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 0.21,
|
| 1425 |
+
"learning_rate": 8.929975163367172e-09,
|
| 1426 |
+
"loss": 0.0126,
|
| 1427 |
+
"step": 35500
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.21,
|
| 1431 |
+
"eval_avg_sts": 0.6848348708799592,
|
| 1432 |
+
"eval_sickr_spearman": 0.6014064249902991,
|
| 1433 |
+
"eval_stsb_spearman": 0.7682633167696193,
|
| 1434 |
+
"step": 35500
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.22,
|
| 1438 |
+
"eval_avg_sts": 0.6849041036803378,
|
| 1439 |
+
"eval_sickr_spearman": 0.6014944261609835,
|
| 1440 |
+
"eval_stsb_spearman": 0.7683137811996923,
|
| 1441 |
+
"step": 35750
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.22,
|
| 1445 |
+
"learning_rate": 8.914904391020232e-09,
|
| 1446 |
+
"loss": 0.0123,
|
| 1447 |
+
"step": 36000
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 0.22,
|
| 1451 |
+
"eval_avg_sts": 0.6850199516781598,
|
| 1452 |
+
"eval_sickr_spearman": 0.6016458260245264,
|
| 1453 |
+
"eval_stsb_spearman": 0.7683940773317931,
|
| 1454 |
+
"step": 36000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.22,
|
| 1458 |
+
"eval_avg_sts": 0.6850955218108025,
|
| 1459 |
+
"eval_sickr_spearman": 0.6017597245402905,
|
| 1460 |
+
"eval_stsb_spearman": 0.7684313190813145,
|
| 1461 |
+
"step": 36250
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 0.22,
|
| 1465 |
+
"learning_rate": 8.89983361867329e-09,
|
| 1466 |
+
"loss": 0.0117,
|
| 1467 |
+
"step": 36500
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.22,
|
| 1471 |
+
"eval_avg_sts": 0.6850673538769948,
|
| 1472 |
+
"eval_sickr_spearman": 0.6016572308168788,
|
| 1473 |
+
"eval_stsb_spearman": 0.7684774769371108,
|
| 1474 |
+
"step": 36500
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 0.22,
|
| 1478 |
+
"eval_avg_sts": 0.6851541104973866,
|
| 1479 |
+
"eval_sickr_spearman": 0.6018085310752482,
|
| 1480 |
+
"eval_stsb_spearman": 0.768499689919525,
|
| 1481 |
+
"step": 36750
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 0.22,
|
| 1485 |
+
"learning_rate": 8.884762846326348e-09,
|
| 1486 |
+
"loss": 0.0115,
|
| 1487 |
+
"step": 37000
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.22,
|
| 1491 |
+
"eval_avg_sts": 0.685221784647517,
|
| 1492 |
+
"eval_sickr_spearman": 0.6018979267183601,
|
| 1493 |
+
"eval_stsb_spearman": 0.7685456425766739,
|
| 1494 |
+
"step": 37000
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.22,
|
| 1498 |
+
"eval_avg_sts": 0.6852791986160438,
|
| 1499 |
+
"eval_sickr_spearman": 0.6019803001967482,
|
| 1500 |
+
"eval_stsb_spearman": 0.7685780970353394,
|
| 1501 |
+
"step": 37250
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.23,
|
| 1505 |
+
"learning_rate": 8.869692073979408e-09,
|
| 1506 |
+
"loss": 0.0111,
|
| 1507 |
+
"step": 37500
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.23,
|
| 1511 |
+
"eval_avg_sts": 0.6853061325467835,
|
| 1512 |
+
"eval_sickr_spearman": 0.602020391279035,
|
| 1513 |
+
"eval_stsb_spearman": 0.7685918738145322,
|
| 1514 |
+
"step": 37500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.23,
|
| 1518 |
+
"eval_avg_sts": 0.6853268783841169,
|
| 1519 |
+
"eval_sickr_spearman": 0.6020406609318185,
|
| 1520 |
+
"eval_stsb_spearman": 0.7686130958364152,
|
| 1521 |
+
"step": 37750
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.23,
|
| 1525 |
+
"learning_rate": 8.854621301632466e-09,
|
| 1526 |
+
"loss": 0.011,
|
| 1527 |
+
"step": 38000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 0.23,
|
| 1531 |
+
"eval_avg_sts": 0.6853295441786447,
|
| 1532 |
+
"eval_sickr_spearman": 0.6019849318373105,
|
| 1533 |
+
"eval_stsb_spearman": 0.768674156519979,
|
| 1534 |
+
"step": 38000
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 0.23,
|
| 1538 |
+
"eval_avg_sts": 0.6853341365400603,
|
| 1539 |
+
"eval_sickr_spearman": 0.6020266664049582,
|
| 1540 |
+
"eval_stsb_spearman": 0.7686416066751625,
|
| 1541 |
+
"step": 38250
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 0.23,
|
| 1545 |
+
"learning_rate": 8.839550529285525e-09,
|
| 1546 |
+
"loss": 0.0109,
|
| 1547 |
+
"step": 38500
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 0.23,
|
| 1551 |
+
"eval_avg_sts": 0.6853246034683043,
|
| 1552 |
+
"eval_sickr_spearman": 0.6020385692231776,
|
| 1553 |
+
"eval_stsb_spearman": 0.7686106377134309,
|
| 1554 |
+
"step": 38500
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 0.23,
|
| 1558 |
+
"eval_avg_sts": 0.6853568834070937,
|
| 1559 |
+
"eval_sickr_spearman": 0.602098581340141,
|
| 1560 |
+
"eval_stsb_spearman": 0.7686151854740463,
|
| 1561 |
+
"step": 38750
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 0.24,
|
| 1565 |
+
"learning_rate": 8.824479756938583e-09,
|
| 1566 |
+
"loss": 0.0105,
|
| 1567 |
+
"step": 39000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 0.24,
|
| 1571 |
+
"eval_avg_sts": 0.6854495878694604,
|
| 1572 |
+
"eval_sickr_spearman": 0.6022639757305444,
|
| 1573 |
+
"eval_stsb_spearman": 0.7686352000083764,
|
| 1574 |
+
"step": 39000
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.24,
|
| 1578 |
+
"eval_avg_sts": 0.685494545030269,
|
| 1579 |
+
"eval_sickr_spearman": 0.6023270258052962,
|
| 1580 |
+
"eval_stsb_spearman": 0.7686620642552416,
|
| 1581 |
+
"step": 39250
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.24,
|
| 1585 |
+
"learning_rate": 8.809408984591643e-09,
|
| 1586 |
+
"loss": 0.0104,
|
| 1587 |
+
"step": 39500
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.24,
|
| 1591 |
+
"eval_avg_sts": 0.6855596476767631,
|
| 1592 |
+
"eval_sickr_spearman": 0.6024400776770863,
|
| 1593 |
+
"eval_stsb_spearman": 0.76867921767644,
|
| 1594 |
+
"step": 39500
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.24,
|
| 1598 |
+
"eval_avg_sts": 0.6855377092658718,
|
| 1599 |
+
"eval_sickr_spearman": 0.6024064012182297,
|
| 1600 |
+
"eval_stsb_spearman": 0.7686690173135138,
|
| 1601 |
+
"step": 39750
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.24,
|
| 1605 |
+
"learning_rate": 8.794338212244701e-09,
|
| 1606 |
+
"loss": 0.0101,
|
| 1607 |
+
"step": 40000
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 0.24,
|
| 1611 |
+
"eval_avg_sts": 0.6855094541954295,
|
| 1612 |
+
"eval_sickr_spearman": 0.6023502338106945,
|
| 1613 |
+
"eval_stsb_spearman": 0.7686686745801645,
|
| 1614 |
+
"step": 40000
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.24,
|
| 1618 |
+
"eval_avg_sts": 0.6855909496558752,
|
| 1619 |
+
"eval_sickr_spearman": 0.6024469504340499,
|
| 1620 |
+
"eval_stsb_spearman": 0.7687349488777003,
|
| 1621 |
+
"step": 40250
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 0.24,
|
| 1625 |
+
"learning_rate": 8.77926743989776e-09,
|
| 1626 |
+
"loss": 0.0098,
|
| 1627 |
+
"step": 40500
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.24,
|
| 1631 |
+
"eval_avg_sts": 0.6855538581482595,
|
| 1632 |
+
"eval_sickr_spearman": 0.6023648757711819,
|
| 1633 |
+
"eval_stsb_spearman": 0.7687428405253371,
|
| 1634 |
+
"step": 40500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.25,
|
| 1638 |
+
"eval_avg_sts": 0.6855329227615412,
|
| 1639 |
+
"eval_sickr_spearman": 0.6022896240626906,
|
| 1640 |
+
"eval_stsb_spearman": 0.7687762214603917,
|
| 1641 |
+
"step": 40750
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.25,
|
| 1645 |
+
"learning_rate": 8.76419666755082e-09,
|
| 1646 |
+
"loss": 0.0097,
|
| 1647 |
+
"step": 41000
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.25,
|
| 1651 |
+
"eval_avg_sts": 0.6856315964474966,
|
| 1652 |
+
"eval_sickr_spearman": 0.6024412729391669,
|
| 1653 |
+
"eval_stsb_spearman": 0.7688219199558264,
|
| 1654 |
+
"step": 41000
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.25,
|
| 1658 |
+
"eval_avg_sts": 0.6856635284536272,
|
| 1659 |
+
"eval_sickr_spearman": 0.6024548690453339,
|
| 1660 |
+
"eval_stsb_spearman": 0.7688721878619207,
|
| 1661 |
+
"step": 41250
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.25,
|
| 1665 |
+
"learning_rate": 8.749125895203878e-09,
|
| 1666 |
+
"loss": 0.0094,
|
| 1667 |
+
"step": 41500
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 0.25,
|
| 1671 |
+
"eval_avg_sts": 0.6857199355051676,
|
| 1672 |
+
"eval_sickr_spearman": 0.6025655303929629,
|
| 1673 |
+
"eval_stsb_spearman": 0.7688743406173724,
|
| 1674 |
+
"step": 41500
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.25,
|
| 1678 |
+
"eval_avg_sts": 0.68571207819909,
|
| 1679 |
+
"eval_sickr_spearman": 0.6025172716864586,
|
| 1680 |
+
"eval_stsb_spearman": 0.7689068847117213,
|
| 1681 |
+
"step": 41750
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 0.25,
|
| 1685 |
+
"learning_rate": 8.734055122856936e-09,
|
| 1686 |
+
"loss": 0.0092,
|
| 1687 |
+
"step": 42000
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 0.25,
|
| 1691 |
+
"eval_avg_sts": 0.685660615312776,
|
| 1692 |
+
"eval_sickr_spearman": 0.6024261329528127,
|
| 1693 |
+
"eval_stsb_spearman": 0.7688950976727392,
|
| 1694 |
+
"step": 42000
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.25,
|
| 1698 |
+
"eval_avg_sts": 0.6857670926985571,
|
| 1699 |
+
"eval_sickr_spearman": 0.602601687070901,
|
| 1700 |
+
"eval_stsb_spearman": 0.7689324983262132,
|
| 1701 |
+
"step": 42250
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 0.26,
|
| 1705 |
+
"learning_rate": 8.718984350509996e-09,
|
| 1706 |
+
"loss": 0.009,
|
| 1707 |
+
"step": 42500
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.26,
|
| 1711 |
+
"eval_avg_sts": 0.6857819219395811,
|
| 1712 |
+
"eval_sickr_spearman": 0.602576984987902,
|
| 1713 |
+
"eval_stsb_spearman": 0.7689868588912602,
|
| 1714 |
+
"step": 42500
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 0.26,
|
| 1718 |
+
"eval_avg_sts": 0.6858294973016628,
|
| 1719 |
+
"eval_sickr_spearman": 0.602676988581979,
|
| 1720 |
+
"eval_stsb_spearman": 0.7689820060213465,
|
| 1721 |
+
"step": 42750
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 0.26,
|
| 1725 |
+
"learning_rate": 8.703913578163054e-09,
|
| 1726 |
+
"loss": 0.0089,
|
| 1727 |
+
"step": 43000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 0.26,
|
| 1731 |
+
"eval_avg_sts": 0.6858174137793278,
|
| 1732 |
+
"eval_sickr_spearman": 0.6026706636534691,
|
| 1733 |
+
"eval_stsb_spearman": 0.7689641639051866,
|
| 1734 |
+
"step": 43000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 0.26,
|
| 1738 |
+
"eval_avg_sts": 0.6858729153589367,
|
| 1739 |
+
"eval_sickr_spearman": 0.6027516924620167,
|
| 1740 |
+
"eval_stsb_spearman": 0.7689941382558566,
|
| 1741 |
+
"step": 43250
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 0.26,
|
| 1745 |
+
"learning_rate": 8.688842805816112e-09,
|
| 1746 |
+
"loss": 0.0087,
|
| 1747 |
+
"step": 43500
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 0.26,
|
| 1751 |
+
"eval_avg_sts": 0.6858947608283459,
|
| 1752 |
+
"eval_sickr_spearman": 0.6027528379215106,
|
| 1753 |
+
"eval_stsb_spearman": 0.7690366837351812,
|
| 1754 |
+
"step": 43500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.26,
|
| 1758 |
+
"eval_avg_sts": 0.6858911001750951,
|
| 1759 |
+
"eval_sickr_spearman": 0.6027214622918947,
|
| 1760 |
+
"eval_stsb_spearman": 0.7690607380582957,
|
| 1761 |
+
"step": 43750
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.27,
|
| 1765 |
+
"learning_rate": 8.673772033469172e-09,
|
| 1766 |
+
"loss": 0.0085,
|
| 1767 |
+
"step": 44000
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.27,
|
| 1771 |
+
"eval_avg_sts": 0.6859079265589377,
|
| 1772 |
+
"eval_sickr_spearman": 0.602721561897068,
|
| 1773 |
+
"eval_stsb_spearman": 0.7690942912208074,
|
| 1774 |
+
"step": 44000
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.27,
|
| 1778 |
+
"eval_avg_sts": 0.6860282186997773,
|
| 1779 |
+
"eval_sickr_spearman": 0.6029633036528699,
|
| 1780 |
+
"eval_stsb_spearman": 0.7690931337466848,
|
| 1781 |
+
"step": 44250
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.27,
|
| 1785 |
+
"learning_rate": 8.65870126112223e-09,
|
| 1786 |
+
"loss": 0.0082,
|
| 1787 |
+
"step": 44500
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.27,
|
| 1791 |
+
"eval_avg_sts": 0.6861163399063477,
|
| 1792 |
+
"eval_sickr_spearman": 0.6030905492618673,
|
| 1793 |
+
"eval_stsb_spearman": 0.769142130550828,
|
| 1794 |
+
"step": 44500
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.27,
|
| 1798 |
+
"eval_avg_sts": 0.6861892434287493,
|
| 1799 |
+
"eval_sickr_spearman": 0.603183331480874,
|
| 1800 |
+
"eval_stsb_spearman": 0.7691951553766246,
|
| 1801 |
+
"step": 44750
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.27,
|
| 1805 |
+
"learning_rate": 8.643630488775289e-09,
|
| 1806 |
+
"loss": 0.0082,
|
| 1807 |
+
"step": 45000
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 0.27,
|
| 1811 |
+
"eval_avg_sts": 0.6861477662185661,
|
| 1812 |
+
"eval_sickr_spearman": 0.603116446606947,
|
| 1813 |
+
"eval_stsb_spearman": 0.7691790858301852,
|
| 1814 |
+
"step": 45000
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.27,
|
| 1818 |
+
"eval_avg_sts": 0.6862196383572703,
|
| 1819 |
+
"eval_sickr_spearman": 0.603254100956563,
|
| 1820 |
+
"eval_stsb_spearman": 0.7691851757579776,
|
| 1821 |
+
"step": 45250
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 0.27,
|
| 1825 |
+
"learning_rate": 8.628559716428347e-09,
|
| 1826 |
+
"loss": 0.0081,
|
| 1827 |
+
"step": 45500
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.27,
|
| 1831 |
+
"eval_avg_sts": 0.686198388321166,
|
| 1832 |
+
"eval_sickr_spearman": 0.6032430447823174,
|
| 1833 |
+
"eval_stsb_spearman": 0.7691537318600145,
|
| 1834 |
+
"step": 45500
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.28,
|
| 1838 |
+
"eval_avg_sts": 0.6861370624006556,
|
| 1839 |
+
"eval_sickr_spearman": 0.6031246640337512,
|
| 1840 |
+
"eval_stsb_spearman": 0.7691494607675601,
|
| 1841 |
+
"step": 45750
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 0.28,
|
| 1845 |
+
"learning_rate": 8.613488944081407e-09,
|
| 1846 |
+
"loss": 0.0079,
|
| 1847 |
+
"step": 46000
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 0.28,
|
| 1851 |
+
"eval_avg_sts": 0.6861745072632264,
|
| 1852 |
+
"eval_sickr_spearman": 0.6032205340131327,
|
| 1853 |
+
"eval_stsb_spearman": 0.7691284805133201,
|
| 1854 |
+
"step": 46000
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 0.28,
|
| 1858 |
+
"eval_avg_sts": 0.6862384540266784,
|
| 1859 |
+
"eval_sickr_spearman": 0.6033426499557009,
|
| 1860 |
+
"eval_stsb_spearman": 0.7691342580976559,
|
| 1861 |
+
"step": 46250
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 0.28,
|
| 1865 |
+
"learning_rate": 8.598418171734465e-09,
|
| 1866 |
+
"loss": 0.0078,
|
| 1867 |
+
"step": 46500
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.28,
|
| 1871 |
+
"eval_avg_sts": 0.6862749736889266,
|
| 1872 |
+
"eval_sickr_spearman": 0.6034227325151013,
|
| 1873 |
+
"eval_stsb_spearman": 0.769127214862752,
|
| 1874 |
+
"step": 46500
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.28,
|
| 1878 |
+
"eval_avg_sts": 0.6863208881531138,
|
| 1879 |
+
"eval_sickr_spearman": 0.6035210428212308,
|
| 1880 |
+
"eval_stsb_spearman": 0.7691207334849967,
|
| 1881 |
+
"step": 46750
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 0.28,
|
| 1885 |
+
"learning_rate": 8.583347399387524e-09,
|
| 1886 |
+
"loss": 0.0077,
|
| 1887 |
+
"step": 47000
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 0.28,
|
| 1891 |
+
"eval_avg_sts": 0.6863572468626877,
|
| 1892 |
+
"eval_sickr_spearman": 0.6035845411192627,
|
| 1893 |
+
"eval_stsb_spearman": 0.7691299526061127,
|
| 1894 |
+
"step": 47000
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 0.28,
|
| 1898 |
+
"eval_avg_sts": 0.6863364230688664,
|
| 1899 |
+
"eval_sickr_spearman": 0.603581104740781,
|
| 1900 |
+
"eval_stsb_spearman": 0.7690917413969519,
|
| 1901 |
+
"step": 47250
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 0.29,
|
| 1905 |
+
"learning_rate": 8.568276627040584e-09,
|
| 1906 |
+
"loss": 0.0075,
|
| 1907 |
+
"step": 47500
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 0.29,
|
| 1911 |
+
"eval_avg_sts": 0.6863403563629373,
|
| 1912 |
+
"eval_sickr_spearman": 0.6035747300096844,
|
| 1913 |
+
"eval_stsb_spearman": 0.7691059827161902,
|
| 1914 |
+
"step": 47500
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.29,
|
| 1918 |
+
"eval_avg_sts": 0.6863889077611471,
|
| 1919 |
+
"eval_sickr_spearman": 0.6036478900095347,
|
| 1920 |
+
"eval_stsb_spearman": 0.7691299255127595,
|
| 1921 |
+
"step": 47750
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.29,
|
| 1925 |
+
"learning_rate": 8.553205854693642e-09,
|
| 1926 |
+
"loss": 0.0075,
|
| 1927 |
+
"step": 48000
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.29,
|
| 1931 |
+
"eval_avg_sts": 0.6864311113667037,
|
| 1932 |
+
"eval_sickr_spearman": 0.6037579936520918,
|
| 1933 |
+
"eval_stsb_spearman": 0.7691042290813154,
|
| 1934 |
+
"step": 48000
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.29,
|
| 1938 |
+
"eval_avg_sts": 0.686500195230545,
|
| 1939 |
+
"eval_sickr_spearman": 0.6039076603017187,
|
| 1940 |
+
"eval_stsb_spearman": 0.7690927301593713,
|
| 1941 |
+
"step": 48250
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.29,
|
| 1945 |
+
"learning_rate": 8.5381350823467e-09,
|
| 1946 |
+
"loss": 0.0072,
|
| 1947 |
+
"step": 48500
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.29,
|
| 1951 |
+
"eval_avg_sts": 0.6864608843930222,
|
| 1952 |
+
"eval_sickr_spearman": 0.6038096488111093,
|
| 1953 |
+
"eval_stsb_spearman": 0.769112119974935,
|
| 1954 |
+
"step": 48500
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.29,
|
| 1958 |
+
"eval_avg_sts": 0.6865355929074337,
|
| 1959 |
+
"eval_sickr_spearman": 0.603904373330997,
|
| 1960 |
+
"eval_stsb_spearman": 0.7691668124838703,
|
| 1961 |
+
"step": 48750
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 0.3,
|
| 1965 |
+
"learning_rate": 8.52306430999976e-09,
|
| 1966 |
+
"loss": 0.0071,
|
| 1967 |
+
"step": 49000
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 0.3,
|
| 1971 |
+
"eval_avg_sts": 0.6866365816613527,
|
| 1972 |
+
"eval_sickr_spearman": 0.6040533328677922,
|
| 1973 |
+
"eval_stsb_spearman": 0.7692198304549132,
|
| 1974 |
+
"step": 49000
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 0.3,
|
| 1978 |
+
"eval_avg_sts": 0.6866187505580584,
|
| 1979 |
+
"eval_sickr_spearman": 0.6040197659243618,
|
| 1980 |
+
"eval_stsb_spearman": 0.7692177351917551,
|
| 1981 |
+
"step": 49250
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 0.3,
|
| 1985 |
+
"learning_rate": 8.507993537652819e-09,
|
| 1986 |
+
"loss": 0.0071,
|
| 1987 |
+
"step": 49500
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 0.3,
|
| 1991 |
+
"eval_avg_sts": 0.6866502129431028,
|
| 1992 |
+
"eval_sickr_spearman": 0.6040536316833123,
|
| 1993 |
+
"eval_stsb_spearman": 0.7692467942028933,
|
| 1994 |
+
"step": 49500
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.3,
|
| 1998 |
+
"eval_avg_sts": 0.686725828130758,
|
| 1999 |
+
"eval_sickr_spearman": 0.6041687752637436,
|
| 2000 |
+
"eval_stsb_spearman": 0.7692828809977725,
|
| 2001 |
+
"step": 49750
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.3,
|
| 2005 |
+
"learning_rate": 8.492922765305877e-09,
|
| 2006 |
+
"loss": 0.0069,
|
| 2007 |
+
"step": 50000
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 0.3,
|
| 2011 |
+
"eval_avg_sts": 0.6867152080403021,
|
| 2012 |
+
"eval_sickr_spearman": 0.6041749009819067,
|
| 2013 |
+
"eval_stsb_spearman": 0.7692555150986975,
|
| 2014 |
+
"step": 50000
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"epoch": 0.3,
|
| 2018 |
+
"eval_avg_sts": 0.6867052144023214,
|
| 2019 |
+
"eval_sickr_spearman": 0.6041375490418878,
|
| 2020 |
+
"eval_stsb_spearman": 0.7692728797627548,
|
| 2021 |
+
"step": 50250
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 0.3,
|
| 2025 |
+
"learning_rate": 8.477851992958935e-09,
|
| 2026 |
+
"loss": 0.0067,
|
| 2027 |
+
"step": 50500
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 0.3,
|
| 2031 |
+
"eval_avg_sts": 0.6867264422199005,
|
| 2032 |
+
"eval_sickr_spearman": 0.6041753990077735,
|
| 2033 |
+
"eval_stsb_spearman": 0.7692774854320276,
|
| 2034 |
+
"step": 50500
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 0.31,
|
| 2038 |
+
"eval_avg_sts": 0.6867642770376152,
|
| 2039 |
+
"eval_sickr_spearman": 0.6042159383133408,
|
| 2040 |
+
"eval_stsb_spearman": 0.7693126157618897,
|
| 2041 |
+
"step": 50750
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 0.31,
|
| 2045 |
+
"learning_rate": 8.462781220611993e-09,
|
| 2046 |
+
"loss": 0.0067,
|
| 2047 |
+
"step": 51000
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.31,
|
| 2051 |
+
"eval_avg_sts": 0.6867016135707726,
|
| 2052 |
+
"eval_sickr_spearman": 0.6041016911794699,
|
| 2053 |
+
"eval_stsb_spearman": 0.7693015359620754,
|
| 2054 |
+
"step": 51000
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.31,
|
| 2058 |
+
"eval_avg_sts": 0.6867455519331296,
|
| 2059 |
+
"eval_sickr_spearman": 0.6041852101173518,
|
| 2060 |
+
"eval_stsb_spearman": 0.7693058937489072,
|
| 2061 |
+
"step": 51250
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.31,
|
| 2065 |
+
"learning_rate": 8.447710448265052e-09,
|
| 2066 |
+
"loss": 0.0066,
|
| 2067 |
+
"step": 51500
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.31,
|
| 2071 |
+
"eval_avg_sts": 0.6866994936909452,
|
| 2072 |
+
"eval_sickr_spearman": 0.604103284862244,
|
| 2073 |
+
"eval_stsb_spearman": 0.7692957025196464,
|
| 2074 |
+
"step": 51500
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.31,
|
| 2078 |
+
"eval_avg_sts": 0.6866555978626698,
|
| 2079 |
+
"eval_sickr_spearman": 0.6040234513157771,
|
| 2080 |
+
"eval_stsb_spearman": 0.7692877444095627,
|
| 2081 |
+
"step": 51750
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.31,
|
| 2085 |
+
"learning_rate": 8.432639675918112e-09,
|
| 2086 |
+
"loss": 0.0064,
|
| 2087 |
+
"step": 52000
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 0.31,
|
| 2091 |
+
"eval_avg_sts": 0.6866565885255093,
|
| 2092 |
+
"eval_sickr_spearman": 0.6039624431470797,
|
| 2093 |
+
"eval_stsb_spearman": 0.7693507339039388,
|
| 2094 |
+
"step": 52000
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 0.31,
|
| 2098 |
+
"eval_avg_sts": 0.6866521898462536,
|
| 2099 |
+
"eval_sickr_spearman": 0.6039828622076233,
|
| 2100 |
+
"eval_stsb_spearman": 0.7693215174848839,
|
| 2101 |
+
"step": 52250
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 0.32,
|
| 2105 |
+
"learning_rate": 8.41756890357117e-09,
|
| 2106 |
+
"loss": 0.0064,
|
| 2107 |
+
"step": 52500
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 0.32,
|
| 2111 |
+
"eval_avg_sts": 0.6866535588513245,
|
| 2112 |
+
"eval_sickr_spearman": 0.6039745451756456,
|
| 2113 |
+
"eval_stsb_spearman": 0.7693325725270033,
|
| 2114 |
+
"step": 52500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.32,
|
| 2118 |
+
"eval_avg_sts": 0.6865157750150845,
|
| 2119 |
+
"eval_sickr_spearman": 0.6037358413776323,
|
| 2120 |
+
"eval_stsb_spearman": 0.7692957086525368,
|
| 2121 |
+
"step": 52750
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 0.32,
|
| 2125 |
+
"learning_rate": 8.402498131224228e-09,
|
| 2126 |
+
"loss": 0.0063,
|
| 2127 |
+
"step": 53000
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 0.32,
|
| 2131 |
+
"eval_avg_sts": 0.6865347142126632,
|
| 2132 |
+
"eval_sickr_spearman": 0.6037773767349331,
|
| 2133 |
+
"eval_stsb_spearman": 0.7692920516903932,
|
| 2134 |
+
"step": 53000
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.32,
|
| 2138 |
+
"eval_avg_sts": 0.6865489888709764,
|
| 2139 |
+
"eval_sickr_spearman": 0.6038000369118779,
|
| 2140 |
+
"eval_stsb_spearman": 0.7692979408300747,
|
| 2141 |
+
"step": 53250
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.32,
|
| 2145 |
+
"learning_rate": 8.387427358877288e-09,
|
| 2146 |
+
"loss": 0.0063,
|
| 2147 |
+
"step": 53500
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 0.32,
|
| 2151 |
+
"eval_avg_sts": 0.686531075517989,
|
| 2152 |
+
"eval_sickr_spearman": 0.6037771775245864,
|
| 2153 |
+
"eval_stsb_spearman": 0.7692849735113917,
|
| 2154 |
+
"step": 53500
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.32,
|
| 2158 |
+
"eval_avg_sts": 0.6865282711445495,
|
| 2159 |
+
"eval_sickr_spearman": 0.6037651751011937,
|
| 2160 |
+
"eval_stsb_spearman": 0.7692913671879054,
|
| 2161 |
+
"step": 53750
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 0.33,
|
| 2165 |
+
"learning_rate": 8.372356586530346e-09,
|
| 2166 |
+
"loss": 0.0061,
|
| 2167 |
+
"step": 54000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 0.33,
|
| 2171 |
+
"eval_avg_sts": 0.6865362702221531,
|
| 2172 |
+
"eval_sickr_spearman": 0.6037866898186445,
|
| 2173 |
+
"eval_stsb_spearman": 0.7692858506256617,
|
| 2174 |
+
"step": 54000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 0.33,
|
| 2178 |
+
"eval_avg_sts": 0.6865545509742186,
|
| 2179 |
+
"eval_sickr_spearman": 0.6038044195395067,
|
| 2180 |
+
"eval_stsb_spearman": 0.7693046824089305,
|
| 2181 |
+
"step": 54250
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 0.33,
|
| 2185 |
+
"learning_rate": 8.357285814183405e-09,
|
| 2186 |
+
"loss": 0.0061,
|
| 2187 |
+
"step": 54500
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.33,
|
| 2191 |
+
"eval_avg_sts": 0.6865219027035194,
|
| 2192 |
+
"eval_sickr_spearman": 0.6037482422217184,
|
| 2193 |
+
"eval_stsb_spearman": 0.7692955631853203,
|
| 2194 |
+
"step": 54500
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 0.33,
|
| 2198 |
+
"eval_avg_sts": 0.686487675439571,
|
| 2199 |
+
"eval_sickr_spearman": 0.6036708490019995,
|
| 2200 |
+
"eval_stsb_spearman": 0.7693045018771424,
|
| 2201 |
+
"step": 54750
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 0.33,
|
| 2205 |
+
"learning_rate": 8.342215041836465e-09,
|
| 2206 |
+
"loss": 0.0059,
|
| 2207 |
+
"step": 55000
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.33,
|
| 2211 |
+
"eval_avg_sts": 0.6864760791160216,
|
| 2212 |
+
"eval_sickr_spearman": 0.6036339452852609,
|
| 2213 |
+
"eval_stsb_spearman": 0.7693182129467823,
|
| 2214 |
+
"step": 55000
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.33,
|
| 2218 |
+
"eval_avg_sts": 0.6864571435422872,
|
| 2219 |
+
"eval_sickr_spearman": 0.6036292140395252,
|
| 2220 |
+
"eval_stsb_spearman": 0.7692850730450491,
|
| 2221 |
+
"step": 55250
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.33,
|
| 2225 |
+
"learning_rate": 8.327144269489523e-09,
|
| 2226 |
+
"loss": 0.0059,
|
| 2227 |
+
"step": 55500
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 0.33,
|
| 2231 |
+
"eval_avg_sts": 0.6863885216776884,
|
| 2232 |
+
"eval_sickr_spearman": 0.6034630726103215,
|
| 2233 |
+
"eval_stsb_spearman": 0.7693139707450554,
|
| 2234 |
+
"step": 55500
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.34,
|
| 2238 |
+
"eval_avg_sts": 0.6862557558724565,
|
| 2239 |
+
"eval_sickr_spearman": 0.6032416005073034,
|
| 2240 |
+
"eval_stsb_spearman": 0.7692699112376095,
|
| 2241 |
+
"step": 55750
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.34,
|
| 2245 |
+
"learning_rate": 8.312073497142581e-09,
|
| 2246 |
+
"loss": 0.0058,
|
| 2247 |
+
"step": 56000
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 0.34,
|
| 2251 |
+
"eval_avg_sts": 0.6862266731200726,
|
| 2252 |
+
"eval_sickr_spearman": 0.603156338478887,
|
| 2253 |
+
"eval_stsb_spearman": 0.7692970077612581,
|
| 2254 |
+
"step": 56000
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 0.34,
|
| 2258 |
+
"eval_avg_sts": 0.6861787937303312,
|
| 2259 |
+
"eval_sickr_spearman": 0.6030421411476029,
|
| 2260 |
+
"eval_stsb_spearman": 0.7693154463130595,
|
| 2261 |
+
"step": 56250
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 0.34,
|
| 2265 |
+
"learning_rate": 8.297002724795641e-09,
|
| 2266 |
+
"loss": 0.0057,
|
| 2267 |
+
"step": 56500
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 0.34,
|
| 2271 |
+
"eval_avg_sts": 0.6862001677803937,
|
| 2272 |
+
"eval_sickr_spearman": 0.6030790448643415,
|
| 2273 |
+
"eval_stsb_spearman": 0.7693212906964461,
|
| 2274 |
+
"step": 56500
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.34,
|
| 2278 |
+
"eval_avg_sts": 0.6861681437241836,
|
| 2279 |
+
"eval_sickr_spearman": 0.6030312343811175,
|
| 2280 |
+
"eval_stsb_spearman": 0.7693050530672497,
|
| 2281 |
+
"step": 56750
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 0.34,
|
| 2285 |
+
"learning_rate": 8.2819319524487e-09,
|
| 2286 |
+
"loss": 0.0056,
|
| 2287 |
+
"step": 57000
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 0.34,
|
| 2291 |
+
"eval_avg_sts": 0.6861963542675861,
|
| 2292 |
+
"eval_sickr_spearman": 0.6030919437342946,
|
| 2293 |
+
"eval_stsb_spearman": 0.7693007648008775,
|
| 2294 |
+
"step": 57000
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 0.35,
|
| 2298 |
+
"eval_avg_sts": 0.6861640128431876,
|
| 2299 |
+
"eval_sickr_spearman": 0.603011114136094,
|
| 2300 |
+
"eval_stsb_spearman": 0.7693169115502814,
|
| 2301 |
+
"step": 57250
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 0.35,
|
| 2305 |
+
"learning_rate": 8.266861180101758e-09,
|
| 2306 |
+
"loss": 0.0056,
|
| 2307 |
+
"step": 57500
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 0.35,
|
| 2311 |
+
"eval_avg_sts": 0.686116276282335,
|
| 2312 |
+
"eval_sickr_spearman": 0.6029430340000863,
|
| 2313 |
+
"eval_stsb_spearman": 0.7692895185645835,
|
| 2314 |
+
"step": 57500
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.35,
|
| 2318 |
+
"eval_avg_sts": 0.6861118072711813,
|
| 2319 |
+
"eval_sickr_spearman": 0.6029420379483524,
|
| 2320 |
+
"eval_stsb_spearman": 0.7692815765940101,
|
| 2321 |
+
"step": 57750
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 0.35,
|
| 2325 |
+
"learning_rate": 8.251790407754816e-09,
|
| 2326 |
+
"loss": 0.0055,
|
| 2327 |
+
"step": 58000
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.35,
|
| 2331 |
+
"eval_avg_sts": 0.6860992615916662,
|
| 2332 |
+
"eval_sickr_spearman": 0.6029261011206111,
|
| 2333 |
+
"eval_stsb_spearman": 0.7692724220627214,
|
| 2334 |
+
"step": 58000
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.35,
|
| 2338 |
+
"eval_avg_sts": 0.6860843714724788,
|
| 2339 |
+
"eval_sickr_spearman": 0.6028835697115763,
|
| 2340 |
+
"eval_stsb_spearman": 0.7692851732333813,
|
| 2341 |
+
"step": 58250
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 0.35,
|
| 2345 |
+
"learning_rate": 8.236719635407876e-09,
|
| 2346 |
+
"loss": 0.0054,
|
| 2347 |
+
"step": 58500
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.35,
|
| 2351 |
+
"eval_avg_sts": 0.6860407186846093,
|
| 2352 |
+
"eval_sickr_spearman": 0.602817232666103,
|
| 2353 |
+
"eval_stsb_spearman": 0.7692642047031155,
|
| 2354 |
+
"step": 58500
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.35,
|
| 2358 |
+
"eval_avg_sts": 0.6860687869250233,
|
| 2359 |
+
"eval_sickr_spearman": 0.6028696747898894,
|
| 2360 |
+
"eval_stsb_spearman": 0.7692678990601571,
|
| 2361 |
+
"step": 58750
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.36,
|
| 2365 |
+
"learning_rate": 8.221648863060934e-09,
|
| 2366 |
+
"loss": 0.0054,
|
| 2367 |
+
"step": 59000
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 0.36,
|
| 2371 |
+
"eval_avg_sts": 0.6859999949935256,
|
| 2372 |
+
"eval_sickr_spearman": 0.6027299287316323,
|
| 2373 |
+
"eval_stsb_spearman": 0.769270061255419,
|
| 2374 |
+
"step": 59000
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 0.36,
|
| 2378 |
+
"eval_avg_sts": 0.6860607016043287,
|
| 2379 |
+
"eval_sickr_spearman": 0.6028618557837788,
|
| 2380 |
+
"eval_stsb_spearman": 0.7692595474248788,
|
| 2381 |
+
"step": 59250
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 0.36,
|
| 2385 |
+
"learning_rate": 8.206578090713993e-09,
|
| 2386 |
+
"loss": 0.0053,
|
| 2387 |
+
"step": 59500
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 0.36,
|
| 2391 |
+
"eval_avg_sts": 0.6860587113413112,
|
| 2392 |
+
"eval_sickr_spearman": 0.6028526423052408,
|
| 2393 |
+
"eval_stsb_spearman": 0.7692647803773817,
|
| 2394 |
+
"step": 59500
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 0.36,
|
| 2398 |
+
"eval_avg_sts": 0.6860611200235003,
|
| 2399 |
+
"eval_sickr_spearman": 0.6028758503106392,
|
| 2400 |
+
"eval_stsb_spearman": 0.7692463897363614,
|
| 2401 |
+
"step": 59750
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 0.36,
|
| 2405 |
+
"learning_rate": 8.191507318367052e-09,
|
| 2406 |
+
"loss": 0.0053,
|
| 2407 |
+
"step": 60000
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 0.36,
|
| 2411 |
+
"eval_avg_sts": 0.6859463597837794,
|
| 2412 |
+
"eval_sickr_spearman": 0.6026531331429537,
|
| 2413 |
+
"eval_stsb_spearman": 0.7692395864246052,
|
| 2414 |
+
"step": 60000
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.36,
|
| 2418 |
+
"eval_avg_sts": 0.6859239543312032,
|
| 2419 |
+
"eval_sickr_spearman": 0.6025836585345187,
|
| 2420 |
+
"eval_stsb_spearman": 0.7692642501278879,
|
| 2421 |
+
"step": 60250
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 0.36,
|
| 2425 |
+
"learning_rate": 8.17643654602011e-09,
|
| 2426 |
+
"loss": 0.0052,
|
| 2427 |
+
"step": 60500
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 0.36,
|
| 2431 |
+
"eval_avg_sts": 0.6859466504629157,
|
| 2432 |
+
"eval_sickr_spearman": 0.6025991969415665,
|
| 2433 |
+
"eval_stsb_spearman": 0.7692941039842648,
|
| 2434 |
+
"step": 60500
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 0.37,
|
| 2438 |
+
"eval_avg_sts": 0.6858718275078098,
|
| 2439 |
+
"eval_sickr_spearman": 0.6024521797056525,
|
| 2440 |
+
"eval_stsb_spearman": 0.7692914753099672,
|
| 2441 |
+
"step": 60750
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 0.37,
|
| 2445 |
+
"learning_rate": 8.161365773673169e-09,
|
| 2446 |
+
"loss": 0.0052,
|
| 2447 |
+
"step": 61000
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 0.37,
|
| 2451 |
+
"eval_avg_sts": 0.6858825814276086,
|
| 2452 |
+
"eval_sickr_spearman": 0.6024696106109946,
|
| 2453 |
+
"eval_stsb_spearman": 0.7692955522442226,
|
| 2454 |
+
"step": 61000
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 0.37,
|
| 2458 |
+
"eval_avg_sts": 0.6859028515979515,
|
| 2459 |
+
"eval_sickr_spearman": 0.6024987451242092,
|
| 2460 |
+
"eval_stsb_spearman": 0.7693069580716939,
|
| 2461 |
+
"step": 61250
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 0.37,
|
| 2465 |
+
"learning_rate": 8.146295001326229e-09,
|
| 2466 |
+
"loss": 0.005,
|
| 2467 |
+
"step": 61500
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 0.37,
|
| 2471 |
+
"eval_avg_sts": 0.6859036333129487,
|
| 2472 |
+
"eval_sickr_spearman": 0.6025038747906385,
|
| 2473 |
+
"eval_stsb_spearman": 0.7693033918352589,
|
| 2474 |
+
"step": 61500
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.37,
|
| 2478 |
+
"eval_avg_sts": 0.6859029568341515,
|
| 2479 |
+
"eval_sickr_spearman": 0.6024901292767116,
|
| 2480 |
+
"eval_stsb_spearman": 0.7693157843915914,
|
| 2481 |
+
"step": 61750
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.37,
|
| 2485 |
+
"learning_rate": 8.131224228979287e-09,
|
| 2486 |
+
"loss": 0.0051,
|
| 2487 |
+
"step": 62000
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.37,
|
| 2491 |
+
"eval_avg_sts": 0.6858457296550131,
|
| 2492 |
+
"eval_sickr_spearman": 0.6024017296853338,
|
| 2493 |
+
"eval_stsb_spearman": 0.7692897296246924,
|
| 2494 |
+
"step": 62000
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.38,
|
| 2498 |
+
"eval_avg_sts": 0.6859309086152754,
|
| 2499 |
+
"eval_sickr_spearman": 0.6025292243072646,
|
| 2500 |
+
"eval_stsb_spearman": 0.769332592923286,
|
| 2501 |
+
"step": 62250
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.38,
|
| 2505 |
+
"learning_rate": 8.116153456632346e-09,
|
| 2506 |
+
"loss": 0.0049,
|
| 2507 |
+
"step": 62500
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 0.38,
|
| 2511 |
+
"eval_avg_sts": 0.6859456802235347,
|
| 2512 |
+
"eval_sickr_spearman": 0.6025707596645655,
|
| 2513 |
+
"eval_stsb_spearman": 0.769320600782504,
|
| 2514 |
+
"step": 62500
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 0.38,
|
| 2518 |
+
"eval_avg_sts": 0.6860315979765899,
|
| 2519 |
+
"eval_sickr_spearman": 0.6026728051646969,
|
| 2520 |
+
"eval_stsb_spearman": 0.7693903907884827,
|
| 2521 |
+
"step": 62750
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 0.38,
|
| 2525 |
+
"learning_rate": 8.101082684285404e-09,
|
| 2526 |
+
"loss": 0.0049,
|
| 2527 |
+
"step": 63000
|
| 2528 |
+
},
|
| 2529 |
+
{
|
| 2530 |
+
"epoch": 0.38,
|
| 2531 |
+
"eval_avg_sts": 0.686043555768491,
|
| 2532 |
+
"eval_sickr_spearman": 0.602662296818905,
|
| 2533 |
+
"eval_stsb_spearman": 0.7694248147180769,
|
| 2534 |
+
"step": 63000
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 0.38,
|
| 2538 |
+
"eval_avg_sts": 0.6861302651703853,
|
| 2539 |
+
"eval_sickr_spearman": 0.6028059274789239,
|
| 2540 |
+
"eval_stsb_spearman": 0.7694546028618466,
|
| 2541 |
+
"step": 63250
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 0.38,
|
| 2545 |
+
"learning_rate": 8.086011911938464e-09,
|
| 2546 |
+
"loss": 0.0049,
|
| 2547 |
+
"step": 63500
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 0.38,
|
| 2551 |
+
"eval_avg_sts": 0.6861418997023286,
|
| 2552 |
+
"eval_sickr_spearman": 0.6028308287722698,
|
| 2553 |
+
"eval_stsb_spearman": 0.7694529706323875,
|
| 2554 |
+
"step": 63500
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 0.38,
|
| 2558 |
+
"eval_avg_sts": 0.6861436270503518,
|
| 2559 |
+
"eval_sickr_spearman": 0.6028189757566372,
|
| 2560 |
+
"eval_stsb_spearman": 0.7694682783440664,
|
| 2561 |
+
"step": 63750
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 0.39,
|
| 2565 |
+
"learning_rate": 8.070941139591522e-09,
|
| 2566 |
+
"loss": 0.0048,
|
| 2567 |
+
"step": 64000
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 0.39,
|
| 2571 |
+
"eval_avg_sts": 0.6861273202059766,
|
| 2572 |
+
"eval_sickr_spearman": 0.6027953693305453,
|
| 2573 |
+
"eval_stsb_spearman": 0.7694592710814079,
|
| 2574 |
+
"step": 64000
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 0.39,
|
| 2578 |
+
"eval_avg_sts": 0.6860970441745926,
|
| 2579 |
+
"eval_sickr_spearman": 0.6027587146267401,
|
| 2580 |
+
"eval_stsb_spearman": 0.7694353737224451,
|
| 2581 |
+
"step": 64250
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 0.39,
|
| 2585 |
+
"learning_rate": 8.05587036724458e-09,
|
| 2586 |
+
"loss": 0.0048,
|
| 2587 |
+
"step": 64500
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 0.39,
|
| 2591 |
+
"eval_avg_sts": 0.6861758328776337,
|
| 2592 |
+
"eval_sickr_spearman": 0.6029076243609485,
|
| 2593 |
+
"eval_stsb_spearman": 0.7694440413943191,
|
| 2594 |
+
"step": 64500
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 0.39,
|
| 2598 |
+
"eval_avg_sts": 0.6861925925169595,
|
| 2599 |
+
"eval_sickr_spearman": 0.602933073482748,
|
| 2600 |
+
"eval_stsb_spearman": 0.769452111551171,
|
| 2601 |
+
"step": 64750
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 0.39,
|
| 2605 |
+
"learning_rate": 8.04079959489764e-09,
|
| 2606 |
+
"loss": 0.0047,
|
| 2607 |
+
"step": 65000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 0.39,
|
| 2611 |
+
"eval_avg_sts": 0.6862022830505737,
|
| 2612 |
+
"eval_sickr_spearman": 0.6029402450552316,
|
| 2613 |
+
"eval_stsb_spearman": 0.7694643210459159,
|
| 2614 |
+
"step": 65000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.39,
|
| 2618 |
+
"eval_avg_sts": 0.6862314740498379,
|
| 2619 |
+
"eval_sickr_spearman": 0.6030238635982871,
|
| 2620 |
+
"eval_stsb_spearman": 0.7694390845013885,
|
| 2621 |
+
"step": 65250
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 0.39,
|
| 2625 |
+
"learning_rate": 8.025728822550699e-09,
|
| 2626 |
+
"loss": 0.0046,
|
| 2627 |
+
"step": 65500
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.39,
|
| 2631 |
+
"eval_avg_sts": 0.6861844875947325,
|
| 2632 |
+
"eval_sickr_spearman": 0.6029598174718015,
|
| 2633 |
+
"eval_stsb_spearman": 0.7694091577176634,
|
| 2634 |
+
"step": 65500
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.4,
|
| 2638 |
+
"eval_avg_sts": 0.6862009373727546,
|
| 2639 |
+
"eval_sickr_spearman": 0.6030062832851849,
|
| 2640 |
+
"eval_stsb_spearman": 0.7693955914603242,
|
| 2641 |
+
"step": 65750
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.4,
|
| 2645 |
+
"learning_rate": 8.010658050203757e-09,
|
| 2646 |
+
"loss": 0.0046,
|
| 2647 |
+
"step": 66000
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 0.4,
|
| 2651 |
+
"eval_avg_sts": 0.6862719488998004,
|
| 2652 |
+
"eval_sickr_spearman": 0.6031286482406865,
|
| 2653 |
+
"eval_stsb_spearman": 0.7694152495589143,
|
| 2654 |
+
"step": 66000
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 0.4,
|
| 2658 |
+
"eval_avg_sts": 0.686297897917143,
|
| 2659 |
+
"eval_sickr_spearman": 0.6031807915489527,
|
| 2660 |
+
"eval_stsb_spearman": 0.7694150042853335,
|
| 2661 |
+
"step": 66250
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 0.4,
|
| 2665 |
+
"learning_rate": 7.995587277856817e-09,
|
| 2666 |
+
"loss": 0.0047,
|
| 2667 |
+
"step": 66500
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 0.4,
|
| 2671 |
+
"eval_avg_sts": 0.6862834571479077,
|
| 2672 |
+
"eval_sickr_spearman": 0.6031002109636855,
|
| 2673 |
+
"eval_stsb_spearman": 0.7694667033321299,
|
| 2674 |
+
"step": 66500
|
| 2675 |
+
}
|
| 2676 |
+
],
|
| 2677 |
+
"max_steps": 331768,
|
| 2678 |
+
"num_train_epochs": 2,
|
| 2679 |
+
"total_flos": 0,
|
| 2680 |
+
"trial_name": null,
|
| 2681 |
+
"trial_params": null
|
| 2682 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:954b7cde9f7b2cfa895273481895c63cc8bb74e7ab42637de941daa5acecf17b
|
| 3 |
+
size 3832
|
unigram.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|