root
commited on
Commit
·
0a185c9
1
Parent(s):
526ae3e
init
Browse files- config.json +26 -0
- eval_results.txt +12 -0
- merges.txt +0 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
- train_results.txt +3 -0
- trainer_state.json +1083 -0
- training_args.bin +3 -0
- vocab.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "result/my-unsup-simcse-roberta-base-d0.1-rs192-std0.1-t0.05",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"RobertaForCL"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"eos_token_id": 2,
|
| 9 |
+
"gradient_checkpointing": false,
|
| 10 |
+
"hidden_act": "gelu",
|
| 11 |
+
"hidden_dropout_prob": 0.1,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"layer_norm_eps": 1e-05,
|
| 16 |
+
"max_position_embeddings": 514,
|
| 17 |
+
"model_type": "roberta",
|
| 18 |
+
"num_attention_heads": 12,
|
| 19 |
+
"num_hidden_layers": 12,
|
| 20 |
+
"pad_token_id": 1,
|
| 21 |
+
"position_embedding_type": "absolute",
|
| 22 |
+
"transformers_version": "4.2.1",
|
| 23 |
+
"type_vocab_size": 1,
|
| 24 |
+
"use_cache": true,
|
| 25 |
+
"vocab_size": 50265
|
| 26 |
+
}
|
eval_results.txt
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
epoch = 1.0
|
| 2 |
+
eval_CR = 86.81
|
| 3 |
+
eval_MPQA = 85.37
|
| 4 |
+
eval_MR = 80.38
|
| 5 |
+
eval_MRPC = 75.61
|
| 6 |
+
eval_SST2 = 86.47
|
| 7 |
+
eval_SUBJ = 93.2
|
| 8 |
+
eval_TREC = 74.98
|
| 9 |
+
eval_avg_sts = 0.778271737124629
|
| 10 |
+
eval_avg_transfer = 83.26
|
| 11 |
+
eval_sickr_spearman = 0.7131796271103109
|
| 12 |
+
eval_stsb_spearman = 0.8433638471389473
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad80aa49ecd13a9e6b4105e334eef706462677fe8e1c10599f09c7e039b98aa8
|
| 3 |
+
size 997289965
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e3408163f721652e895c928b1f88fd9e20af3e4e70a5e8b84a6da272861a351
|
| 3 |
+
size 501035657
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d13f17241fe3ee82151cd92a07c283b6a8de10bc88ad69e2817521f1b9e3237
|
| 3 |
+
size 623
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "model_max_length": 512, "name_or_path": "roberta-base"}
|
train_results.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
epoch = 1.0
|
| 2 |
+
train_runtime = 3862.446
|
| 3 |
+
train_samples_per_second = 4.046
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1083 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8433638471389473,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-roberta-base-d0.1-rs192-std0.1-t0.05",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 15626,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"eval_avg_sts": 0.7572535655220898,
|
| 13 |
+
"eval_sickr_spearman": 0.7054352844388353,
|
| 14 |
+
"eval_stsb_spearman": 0.8090718466053444,
|
| 15 |
+
"step": 125
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.02,
|
| 19 |
+
"eval_avg_sts": 0.7614025514877952,
|
| 20 |
+
"eval_sickr_spearman": 0.7051708251943171,
|
| 21 |
+
"eval_stsb_spearman": 0.8176342777812734,
|
| 22 |
+
"step": 250
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.02,
|
| 26 |
+
"eval_avg_sts": 0.7636217659471919,
|
| 27 |
+
"eval_sickr_spearman": 0.7079373205749496,
|
| 28 |
+
"eval_stsb_spearman": 0.8193062113194342,
|
| 29 |
+
"step": 375
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.03,
|
| 33 |
+
"learning_rate": 9.680020478689364e-06,
|
| 34 |
+
"loss": 0.2125,
|
| 35 |
+
"step": 500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.03,
|
| 39 |
+
"eval_avg_sts": 0.7657425201047614,
|
| 40 |
+
"eval_sickr_spearman": 0.7097071706007412,
|
| 41 |
+
"eval_stsb_spearman": 0.8217778696087817,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.04,
|
| 46 |
+
"eval_avg_sts": 0.7674773834562932,
|
| 47 |
+
"eval_sickr_spearman": 0.7100151940542637,
|
| 48 |
+
"eval_stsb_spearman": 0.8249395728583228,
|
| 49 |
+
"step": 625
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.05,
|
| 53 |
+
"eval_avg_sts": 0.7670247636071179,
|
| 54 |
+
"eval_sickr_spearman": 0.7053980603352208,
|
| 55 |
+
"eval_stsb_spearman": 0.828651466879015,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.06,
|
| 60 |
+
"eval_avg_sts": 0.7642544693075077,
|
| 61 |
+
"eval_sickr_spearman": 0.7042453619318076,
|
| 62 |
+
"eval_stsb_spearman": 0.8242635766832077,
|
| 63 |
+
"step": 875
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.06,
|
| 67 |
+
"learning_rate": 9.36004095737873e-06,
|
| 68 |
+
"loss": 0.0005,
|
| 69 |
+
"step": 1000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.06,
|
| 73 |
+
"eval_avg_sts": 0.7706729186771359,
|
| 74 |
+
"eval_sickr_spearman": 0.7109765365495476,
|
| 75 |
+
"eval_stsb_spearman": 0.8303693008047242,
|
| 76 |
+
"step": 1000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.07,
|
| 80 |
+
"eval_avg_sts": 0.7723601548256038,
|
| 81 |
+
"eval_sickr_spearman": 0.7154485202800445,
|
| 82 |
+
"eval_stsb_spearman": 0.8292717893711632,
|
| 83 |
+
"step": 1125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.08,
|
| 87 |
+
"eval_avg_sts": 0.7715988622174201,
|
| 88 |
+
"eval_sickr_spearman": 0.7117784157880568,
|
| 89 |
+
"eval_stsb_spearman": 0.8314193086467835,
|
| 90 |
+
"step": 1250
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09,
|
| 94 |
+
"eval_avg_sts": 0.7699957963614845,
|
| 95 |
+
"eval_sickr_spearman": 0.7098231176796127,
|
| 96 |
+
"eval_stsb_spearman": 0.8301684750433563,
|
| 97 |
+
"step": 1375
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1,
|
| 101 |
+
"learning_rate": 9.040061436068092e-06,
|
| 102 |
+
"loss": 0.0003,
|
| 103 |
+
"step": 1500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.1,
|
| 107 |
+
"eval_avg_sts": 0.7732021256956165,
|
| 108 |
+
"eval_sickr_spearman": 0.7176254338905241,
|
| 109 |
+
"eval_stsb_spearman": 0.8287788175007089,
|
| 110 |
+
"step": 1500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.1,
|
| 114 |
+
"eval_avg_sts": 0.7706739860482326,
|
| 115 |
+
"eval_sickr_spearman": 0.7126164144148476,
|
| 116 |
+
"eval_stsb_spearman": 0.8287315576816174,
|
| 117 |
+
"step": 1625
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.11,
|
| 121 |
+
"eval_avg_sts": 0.7719707314787304,
|
| 122 |
+
"eval_sickr_spearman": 0.7146769966976441,
|
| 123 |
+
"eval_stsb_spearman": 0.8292644662598168,
|
| 124 |
+
"step": 1750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.12,
|
| 128 |
+
"eval_avg_sts": 0.7692985063340654,
|
| 129 |
+
"eval_sickr_spearman": 0.7109207724407779,
|
| 130 |
+
"eval_stsb_spearman": 0.827676240227353,
|
| 131 |
+
"step": 1875
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.13,
|
| 135 |
+
"learning_rate": 8.720081914757458e-06,
|
| 136 |
+
"loss": 0.0002,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.13,
|
| 141 |
+
"eval_avg_sts": 0.771889783054714,
|
| 142 |
+
"eval_sickr_spearman": 0.7135132511409,
|
| 143 |
+
"eval_stsb_spearman": 0.8302663149685279,
|
| 144 |
+
"step": 2000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.14,
|
| 148 |
+
"eval_avg_sts": 0.7615509683620076,
|
| 149 |
+
"eval_sickr_spearman": 0.6917990552038427,
|
| 150 |
+
"eval_stsb_spearman": 0.8313028815201725,
|
| 151 |
+
"step": 2125
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.14,
|
| 155 |
+
"eval_avg_sts": 0.7642814292065848,
|
| 156 |
+
"eval_sickr_spearman": 0.6928178503087533,
|
| 157 |
+
"eval_stsb_spearman": 0.8357450081044163,
|
| 158 |
+
"step": 2250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.15,
|
| 162 |
+
"eval_avg_sts": 0.7664718523224301,
|
| 163 |
+
"eval_sickr_spearman": 0.6978614041463637,
|
| 164 |
+
"eval_stsb_spearman": 0.8350823004984966,
|
| 165 |
+
"step": 2375
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.16,
|
| 169 |
+
"learning_rate": 8.400102393446819e-06,
|
| 170 |
+
"loss": 0.0002,
|
| 171 |
+
"step": 2500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.16,
|
| 175 |
+
"eval_avg_sts": 0.7669677380011899,
|
| 176 |
+
"eval_sickr_spearman": 0.6978531908280179,
|
| 177 |
+
"eval_stsb_spearman": 0.8360822851743619,
|
| 178 |
+
"step": 2500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.17,
|
| 182 |
+
"eval_avg_sts": 0.767509277729593,
|
| 183 |
+
"eval_sickr_spearman": 0.697946947538025,
|
| 184 |
+
"eval_stsb_spearman": 0.837071607921161,
|
| 185 |
+
"step": 2625
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.18,
|
| 189 |
+
"eval_avg_sts": 0.7668651085967232,
|
| 190 |
+
"eval_sickr_spearman": 0.6985169326187911,
|
| 191 |
+
"eval_stsb_spearman": 0.8352132845746553,
|
| 192 |
+
"step": 2750
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.18,
|
| 196 |
+
"eval_avg_sts": 0.7668897285698411,
|
| 197 |
+
"eval_sickr_spearman": 0.6986797580526662,
|
| 198 |
+
"eval_stsb_spearman": 0.835099699087016,
|
| 199 |
+
"step": 2875
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.19,
|
| 203 |
+
"learning_rate": 8.080122872136184e-06,
|
| 204 |
+
"loss": 0.0002,
|
| 205 |
+
"step": 3000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.19,
|
| 209 |
+
"eval_avg_sts": 0.7678085026409753,
|
| 210 |
+
"eval_sickr_spearman": 0.6984642425005135,
|
| 211 |
+
"eval_stsb_spearman": 0.8371527627814372,
|
| 212 |
+
"step": 3000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2,
|
| 216 |
+
"eval_avg_sts": 0.7692156947271456,
|
| 217 |
+
"eval_sickr_spearman": 0.7026153304423024,
|
| 218 |
+
"eval_stsb_spearman": 0.8358160590119886,
|
| 219 |
+
"step": 3125
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.21,
|
| 223 |
+
"eval_avg_sts": 0.768843154466003,
|
| 224 |
+
"eval_sickr_spearman": 0.705912809649333,
|
| 225 |
+
"eval_stsb_spearman": 0.8317734992826732,
|
| 226 |
+
"step": 3250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.22,
|
| 230 |
+
"eval_avg_sts": 0.7672560021396605,
|
| 231 |
+
"eval_sickr_spearman": 0.7075799211491486,
|
| 232 |
+
"eval_stsb_spearman": 0.8269320831301724,
|
| 233 |
+
"step": 3375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.22,
|
| 237 |
+
"learning_rate": 7.760143350825547e-06,
|
| 238 |
+
"loss": 0.0002,
|
| 239 |
+
"step": 3500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.22,
|
| 243 |
+
"eval_avg_sts": 0.7665000851835462,
|
| 244 |
+
"eval_sickr_spearman": 0.7056194356817491,
|
| 245 |
+
"eval_stsb_spearman": 0.8273807346853432,
|
| 246 |
+
"step": 3500
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.23,
|
| 250 |
+
"eval_avg_sts": 0.7675835380050031,
|
| 251 |
+
"eval_sickr_spearman": 0.706044462898375,
|
| 252 |
+
"eval_stsb_spearman": 0.8291226131116314,
|
| 253 |
+
"step": 3625
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.24,
|
| 257 |
+
"eval_avg_sts": 0.766064422853222,
|
| 258 |
+
"eval_sickr_spearman": 0.7050694795702827,
|
| 259 |
+
"eval_stsb_spearman": 0.8270593661361613,
|
| 260 |
+
"step": 3750
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.25,
|
| 264 |
+
"eval_avg_sts": 0.7680428355810465,
|
| 265 |
+
"eval_sickr_spearman": 0.7066562350373923,
|
| 266 |
+
"eval_stsb_spearman": 0.8294294361247007,
|
| 267 |
+
"step": 3875
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.26,
|
| 271 |
+
"learning_rate": 7.440163829514912e-06,
|
| 272 |
+
"loss": 0.0003,
|
| 273 |
+
"step": 4000
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.26,
|
| 277 |
+
"eval_avg_sts": 0.7706781402051571,
|
| 278 |
+
"eval_sickr_spearman": 0.7060189583835114,
|
| 279 |
+
"eval_stsb_spearman": 0.8353373220268029,
|
| 280 |
+
"step": 4000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.26,
|
| 284 |
+
"eval_avg_sts": 0.7704250512723954,
|
| 285 |
+
"eval_sickr_spearman": 0.7065240054151332,
|
| 286 |
+
"eval_stsb_spearman": 0.8343260971296576,
|
| 287 |
+
"step": 4125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.27,
|
| 291 |
+
"eval_avg_sts": 0.7676150118719631,
|
| 292 |
+
"eval_sickr_spearman": 0.7018462564460752,
|
| 293 |
+
"eval_stsb_spearman": 0.833383767297851,
|
| 294 |
+
"step": 4250
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.28,
|
| 298 |
+
"eval_avg_sts": 0.7672794331395227,
|
| 299 |
+
"eval_sickr_spearman": 0.7023547617170005,
|
| 300 |
+
"eval_stsb_spearman": 0.832204104562045,
|
| 301 |
+
"step": 4375
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.29,
|
| 305 |
+
"learning_rate": 7.120184308204276e-06,
|
| 306 |
+
"loss": 0.0003,
|
| 307 |
+
"step": 4500
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.29,
|
| 311 |
+
"eval_avg_sts": 0.7603546899710306,
|
| 312 |
+
"eval_sickr_spearman": 0.6936834187877694,
|
| 313 |
+
"eval_stsb_spearman": 0.8270259611542918,
|
| 314 |
+
"step": 4500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.3,
|
| 318 |
+
"eval_avg_sts": 0.7604700472645542,
|
| 319 |
+
"eval_sickr_spearman": 0.6935327932536596,
|
| 320 |
+
"eval_stsb_spearman": 0.8274073012754489,
|
| 321 |
+
"step": 4625
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.3,
|
| 325 |
+
"eval_avg_sts": 0.761349850549489,
|
| 326 |
+
"eval_sickr_spearman": 0.6934180469523239,
|
| 327 |
+
"eval_stsb_spearman": 0.8292816541466541,
|
| 328 |
+
"step": 4750
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.31,
|
| 332 |
+
"eval_avg_sts": 0.7624410643103536,
|
| 333 |
+
"eval_sickr_spearman": 0.6947914482668454,
|
| 334 |
+
"eval_stsb_spearman": 0.830090680353862,
|
| 335 |
+
"step": 4875
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.32,
|
| 339 |
+
"learning_rate": 6.800204786893639e-06,
|
| 340 |
+
"loss": 0.0001,
|
| 341 |
+
"step": 5000
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.32,
|
| 345 |
+
"eval_avg_sts": 0.7625184502556002,
|
| 346 |
+
"eval_sickr_spearman": 0.6949592209041686,
|
| 347 |
+
"eval_stsb_spearman": 0.8300776796070318,
|
| 348 |
+
"step": 5000
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.33,
|
| 352 |
+
"eval_avg_sts": 0.7620233259863773,
|
| 353 |
+
"eval_sickr_spearman": 0.6943664690813208,
|
| 354 |
+
"eval_stsb_spearman": 0.829680182891434,
|
| 355 |
+
"step": 5125
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.34,
|
| 359 |
+
"eval_avg_sts": 0.7688935367788103,
|
| 360 |
+
"eval_sickr_spearman": 0.7047914275240574,
|
| 361 |
+
"eval_stsb_spearman": 0.8329956460335631,
|
| 362 |
+
"step": 5250
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.34,
|
| 366 |
+
"eval_avg_sts": 0.7721471528975761,
|
| 367 |
+
"eval_sickr_spearman": 0.7084637894778126,
|
| 368 |
+
"eval_stsb_spearman": 0.8358305163173394,
|
| 369 |
+
"step": 5375
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.35,
|
| 373 |
+
"learning_rate": 6.480225265583003e-06,
|
| 374 |
+
"loss": 0.0002,
|
| 375 |
+
"step": 5500
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.35,
|
| 379 |
+
"eval_avg_sts": 0.7736568332643656,
|
| 380 |
+
"eval_sickr_spearman": 0.7087314187750259,
|
| 381 |
+
"eval_stsb_spearman": 0.8385822477537053,
|
| 382 |
+
"step": 5500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.36,
|
| 386 |
+
"eval_avg_sts": 0.7734070993186029,
|
| 387 |
+
"eval_sickr_spearman": 0.7073194965171511,
|
| 388 |
+
"eval_stsb_spearman": 0.8394947021200546,
|
| 389 |
+
"step": 5625
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.37,
|
| 393 |
+
"eval_avg_sts": 0.7727328180179123,
|
| 394 |
+
"eval_sickr_spearman": 0.7090136014959747,
|
| 395 |
+
"eval_stsb_spearman": 0.83645203453985,
|
| 396 |
+
"step": 5750
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.38,
|
| 400 |
+
"eval_avg_sts": 0.7702894469668818,
|
| 401 |
+
"eval_sickr_spearman": 0.7065904804595234,
|
| 402 |
+
"eval_stsb_spearman": 0.8339884134742402,
|
| 403 |
+
"step": 5875
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.38,
|
| 407 |
+
"learning_rate": 6.1602457442723675e-06,
|
| 408 |
+
"loss": 0.0001,
|
| 409 |
+
"step": 6000
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.38,
|
| 413 |
+
"eval_avg_sts": 0.7673639531247151,
|
| 414 |
+
"eval_sickr_spearman": 0.7063840427855425,
|
| 415 |
+
"eval_stsb_spearman": 0.8283438634638877,
|
| 416 |
+
"step": 6000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.39,
|
| 420 |
+
"eval_avg_sts": 0.7656242558893468,
|
| 421 |
+
"eval_sickr_spearman": 0.7046311957696599,
|
| 422 |
+
"eval_stsb_spearman": 0.8266173160090337,
|
| 423 |
+
"step": 6125
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.4,
|
| 427 |
+
"eval_avg_sts": 0.7678162328535889,
|
| 428 |
+
"eval_sickr_spearman": 0.7065671853753259,
|
| 429 |
+
"eval_stsb_spearman": 0.8290652803318519,
|
| 430 |
+
"step": 6250
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.41,
|
| 434 |
+
"eval_avg_sts": 0.7659380803169125,
|
| 435 |
+
"eval_sickr_spearman": 0.7034162970898843,
|
| 436 |
+
"eval_stsb_spearman": 0.8284598635439405,
|
| 437 |
+
"step": 6375
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.42,
|
| 441 |
+
"learning_rate": 5.840266222961732e-06,
|
| 442 |
+
"loss": 0.0002,
|
| 443 |
+
"step": 6500
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.42,
|
| 447 |
+
"eval_avg_sts": 0.7674929110456945,
|
| 448 |
+
"eval_sickr_spearman": 0.7043084267679958,
|
| 449 |
+
"eval_stsb_spearman": 0.8306773953233932,
|
| 450 |
+
"step": 6500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.42,
|
| 454 |
+
"eval_avg_sts": 0.7674747228197092,
|
| 455 |
+
"eval_sickr_spearman": 0.7030638928986329,
|
| 456 |
+
"eval_stsb_spearman": 0.8318855527407855,
|
| 457 |
+
"step": 6625
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.43,
|
| 461 |
+
"eval_avg_sts": 0.7669269125478364,
|
| 462 |
+
"eval_sickr_spearman": 0.7022142707452941,
|
| 463 |
+
"eval_stsb_spearman": 0.8316395543503788,
|
| 464 |
+
"step": 6750
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.44,
|
| 468 |
+
"eval_avg_sts": 0.7663176164301795,
|
| 469 |
+
"eval_sickr_spearman": 0.7013282410170653,
|
| 470 |
+
"eval_stsb_spearman": 0.8313069918432938,
|
| 471 |
+
"step": 6875
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.45,
|
| 475 |
+
"learning_rate": 5.520286701651095e-06,
|
| 476 |
+
"loss": 0.0001,
|
| 477 |
+
"step": 7000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.45,
|
| 481 |
+
"eval_avg_sts": 0.7679613274313469,
|
| 482 |
+
"eval_sickr_spearman": 0.705412661790058,
|
| 483 |
+
"eval_stsb_spearman": 0.8305099930726358,
|
| 484 |
+
"step": 7000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.46,
|
| 488 |
+
"eval_avg_sts": 0.7677041476146869,
|
| 489 |
+
"eval_sickr_spearman": 0.7044645758787711,
|
| 490 |
+
"eval_stsb_spearman": 0.8309437193506026,
|
| 491 |
+
"step": 7125
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.46,
|
| 495 |
+
"eval_avg_sts": 0.771967747158485,
|
| 496 |
+
"eval_sickr_spearman": 0.7112394107677183,
|
| 497 |
+
"eval_stsb_spearman": 0.8326960835492517,
|
| 498 |
+
"step": 7250
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.47,
|
| 502 |
+
"eval_avg_sts": 0.7727980509536904,
|
| 503 |
+
"eval_sickr_spearman": 0.7122028666314655,
|
| 504 |
+
"eval_stsb_spearman": 0.8333932352759152,
|
| 505 |
+
"step": 7375
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.48,
|
| 509 |
+
"learning_rate": 5.200307180340458e-06,
|
| 510 |
+
"loss": 0.0002,
|
| 511 |
+
"step": 7500
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 0.48,
|
| 515 |
+
"eval_avg_sts": 0.7716808664260861,
|
| 516 |
+
"eval_sickr_spearman": 0.7073300153283661,
|
| 517 |
+
"eval_stsb_spearman": 0.8360317175238062,
|
| 518 |
+
"step": 7500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.49,
|
| 522 |
+
"eval_avg_sts": 0.7749475591557311,
|
| 523 |
+
"eval_sickr_spearman": 0.7102010744168291,
|
| 524 |
+
"eval_stsb_spearman": 0.8396940438946332,
|
| 525 |
+
"step": 7625
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.5,
|
| 529 |
+
"eval_avg_sts": 0.7741400965390657,
|
| 530 |
+
"eval_sickr_spearman": 0.7103573195898074,
|
| 531 |
+
"eval_stsb_spearman": 0.8379228734883241,
|
| 532 |
+
"step": 7750
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.5,
|
| 536 |
+
"eval_avg_sts": 0.7742400629462907,
|
| 537 |
+
"eval_sickr_spearman": 0.7103860902195688,
|
| 538 |
+
"eval_stsb_spearman": 0.8380940356730125,
|
| 539 |
+
"step": 7875
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.51,
|
| 543 |
+
"learning_rate": 4.8803276590298225e-06,
|
| 544 |
+
"loss": 0.0001,
|
| 545 |
+
"step": 8000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.51,
|
| 549 |
+
"eval_avg_sts": 0.7737984133316254,
|
| 550 |
+
"eval_sickr_spearman": 0.7100542913708344,
|
| 551 |
+
"eval_stsb_spearman": 0.8375425352924163,
|
| 552 |
+
"step": 8000
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.52,
|
| 556 |
+
"eval_avg_sts": 0.7736256294772575,
|
| 557 |
+
"eval_sickr_spearman": 0.7117209225596355,
|
| 558 |
+
"eval_stsb_spearman": 0.8355303363948796,
|
| 559 |
+
"step": 8125
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.53,
|
| 563 |
+
"eval_avg_sts": 0.7739892471083305,
|
| 564 |
+
"eval_sickr_spearman": 0.7121400419507844,
|
| 565 |
+
"eval_stsb_spearman": 0.8358384522658765,
|
| 566 |
+
"step": 8250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.54,
|
| 570 |
+
"eval_avg_sts": 0.771587372188157,
|
| 571 |
+
"eval_sickr_spearman": 0.7086112929903291,
|
| 572 |
+
"eval_stsb_spearman": 0.834563451385985,
|
| 573 |
+
"step": 8375
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.54,
|
| 577 |
+
"learning_rate": 4.560348137719187e-06,
|
| 578 |
+
"loss": 0.0001,
|
| 579 |
+
"step": 8500
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.54,
|
| 583 |
+
"eval_avg_sts": 0.7717136828796769,
|
| 584 |
+
"eval_sickr_spearman": 0.7081623943162886,
|
| 585 |
+
"eval_stsb_spearman": 0.8352649714430652,
|
| 586 |
+
"step": 8500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.55,
|
| 590 |
+
"eval_avg_sts": 0.7700399485422347,
|
| 591 |
+
"eval_sickr_spearman": 0.7062681917688737,
|
| 592 |
+
"eval_stsb_spearman": 0.8338117053155956,
|
| 593 |
+
"step": 8625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.56,
|
| 597 |
+
"eval_avg_sts": 0.7655558973491845,
|
| 598 |
+
"eval_sickr_spearman": 0.7032426646581855,
|
| 599 |
+
"eval_stsb_spearman": 0.8278691300401836,
|
| 600 |
+
"step": 8750
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.57,
|
| 604 |
+
"eval_avg_sts": 0.7691329874077902,
|
| 605 |
+
"eval_sickr_spearman": 0.7023757513083291,
|
| 606 |
+
"eval_stsb_spearman": 0.8358902235072514,
|
| 607 |
+
"step": 8875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.58,
|
| 611 |
+
"learning_rate": 4.24036861640855e-06,
|
| 612 |
+
"loss": 0.0001,
|
| 613 |
+
"step": 9000
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.58,
|
| 617 |
+
"eval_avg_sts": 0.7696286255223032,
|
| 618 |
+
"eval_sickr_spearman": 0.7031883895135604,
|
| 619 |
+
"eval_stsb_spearman": 0.8360688615310459,
|
| 620 |
+
"step": 9000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.58,
|
| 624 |
+
"eval_avg_sts": 0.7696384100854612,
|
| 625 |
+
"eval_sickr_spearman": 0.7036895940570672,
|
| 626 |
+
"eval_stsb_spearman": 0.835587226113855,
|
| 627 |
+
"step": 9125
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.59,
|
| 631 |
+
"eval_avg_sts": 0.7688416464364851,
|
| 632 |
+
"eval_sickr_spearman": 0.7023802182007628,
|
| 633 |
+
"eval_stsb_spearman": 0.8353030746722075,
|
| 634 |
+
"step": 9250
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.6,
|
| 638 |
+
"eval_avg_sts": 0.7660994859121804,
|
| 639 |
+
"eval_sickr_spearman": 0.7051366750811947,
|
| 640 |
+
"eval_stsb_spearman": 0.8270622967431659,
|
| 641 |
+
"step": 9375
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.61,
|
| 645 |
+
"learning_rate": 3.920389095097914e-06,
|
| 646 |
+
"loss": 0.0002,
|
| 647 |
+
"step": 9500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.61,
|
| 651 |
+
"eval_avg_sts": 0.7673638146997882,
|
| 652 |
+
"eval_sickr_spearman": 0.7064614138295504,
|
| 653 |
+
"eval_stsb_spearman": 0.828266215570026,
|
| 654 |
+
"step": 9500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.62,
|
| 658 |
+
"eval_avg_sts": 0.7688152733727235,
|
| 659 |
+
"eval_sickr_spearman": 0.707200523478889,
|
| 660 |
+
"eval_stsb_spearman": 0.830430023266558,
|
| 661 |
+
"step": 9625
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.62,
|
| 665 |
+
"eval_avg_sts": 0.7697361419751771,
|
| 666 |
+
"eval_sickr_spearman": 0.7075413521746937,
|
| 667 |
+
"eval_stsb_spearman": 0.8319309317756607,
|
| 668 |
+
"step": 9750
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.63,
|
| 672 |
+
"eval_avg_sts": 0.7718885121529019,
|
| 673 |
+
"eval_sickr_spearman": 0.7071343846522086,
|
| 674 |
+
"eval_stsb_spearman": 0.8366426396535952,
|
| 675 |
+
"step": 9875
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.64,
|
| 679 |
+
"learning_rate": 3.600409573787278e-06,
|
| 680 |
+
"loss": 0.0001,
|
| 681 |
+
"step": 10000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.64,
|
| 685 |
+
"eval_avg_sts": 0.7727693021998165,
|
| 686 |
+
"eval_sickr_spearman": 0.7095777267823654,
|
| 687 |
+
"eval_stsb_spearman": 0.8359608776172676,
|
| 688 |
+
"step": 10000
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.65,
|
| 692 |
+
"eval_avg_sts": 0.7717316927837671,
|
| 693 |
+
"eval_sickr_spearman": 0.7084112434528395,
|
| 694 |
+
"eval_stsb_spearman": 0.8350521421146947,
|
| 695 |
+
"step": 10125
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.66,
|
| 699 |
+
"eval_avg_sts": 0.7720632550487434,
|
| 700 |
+
"eval_sickr_spearman": 0.7069583986965393,
|
| 701 |
+
"eval_stsb_spearman": 0.8371681114009475,
|
| 702 |
+
"step": 10250
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.66,
|
| 706 |
+
"eval_avg_sts": 0.7716685440161564,
|
| 707 |
+
"eval_sickr_spearman": 0.7068186762224559,
|
| 708 |
+
"eval_stsb_spearman": 0.836518411809857,
|
| 709 |
+
"step": 10375
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.67,
|
| 713 |
+
"learning_rate": 3.280430052476642e-06,
|
| 714 |
+
"loss": 0.0001,
|
| 715 |
+
"step": 10500
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.67,
|
| 719 |
+
"eval_avg_sts": 0.7706994083764094,
|
| 720 |
+
"eval_sickr_spearman": 0.7044442106917614,
|
| 721 |
+
"eval_stsb_spearman": 0.8369546060610572,
|
| 722 |
+
"step": 10500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.68,
|
| 726 |
+
"eval_avg_sts": 0.770376770388492,
|
| 727 |
+
"eval_sickr_spearman": 0.704766259226904,
|
| 728 |
+
"eval_stsb_spearman": 0.83598728155008,
|
| 729 |
+
"step": 10625
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.69,
|
| 733 |
+
"eval_avg_sts": 0.7700945307869602,
|
| 734 |
+
"eval_sickr_spearman": 0.7045319154829873,
|
| 735 |
+
"eval_stsb_spearman": 0.8356571460909332,
|
| 736 |
+
"step": 10750
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 0.7,
|
| 740 |
+
"eval_avg_sts": 0.7701867471043669,
|
| 741 |
+
"eval_sickr_spearman": 0.7049941187721263,
|
| 742 |
+
"eval_stsb_spearman": 0.8353793754366073,
|
| 743 |
+
"step": 10875
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 0.7,
|
| 747 |
+
"learning_rate": 2.960450531166006e-06,
|
| 748 |
+
"loss": 0.0001,
|
| 749 |
+
"step": 11000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 0.7,
|
| 753 |
+
"eval_avg_sts": 0.7700153990116653,
|
| 754 |
+
"eval_sickr_spearman": 0.7048114084622558,
|
| 755 |
+
"eval_stsb_spearman": 0.8352193895610748,
|
| 756 |
+
"step": 11000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.71,
|
| 760 |
+
"eval_avg_sts": 0.7693725150897858,
|
| 761 |
+
"eval_sickr_spearman": 0.7057958058862298,
|
| 762 |
+
"eval_stsb_spearman": 0.8329492242933417,
|
| 763 |
+
"step": 11125
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.72,
|
| 767 |
+
"eval_avg_sts": 0.7696360829638863,
|
| 768 |
+
"eval_sickr_spearman": 0.7062946088746646,
|
| 769 |
+
"eval_stsb_spearman": 0.832977557053108,
|
| 770 |
+
"step": 11250
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.73,
|
| 774 |
+
"eval_avg_sts": 0.7709027781052127,
|
| 775 |
+
"eval_sickr_spearman": 0.7053959469667576,
|
| 776 |
+
"eval_stsb_spearman": 0.8364096092436679,
|
| 777 |
+
"step": 11375
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.74,
|
| 781 |
+
"learning_rate": 2.640471009855369e-06,
|
| 782 |
+
"loss": 0.0002,
|
| 783 |
+
"step": 11500
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.74,
|
| 787 |
+
"eval_avg_sts": 0.7705889705773711,
|
| 788 |
+
"eval_sickr_spearman": 0.7048775472889359,
|
| 789 |
+
"eval_stsb_spearman": 0.8363003938658062,
|
| 790 |
+
"step": 11500
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.74,
|
| 794 |
+
"eval_avg_sts": 0.7756785758354756,
|
| 795 |
+
"eval_sickr_spearman": 0.7125353379156201,
|
| 796 |
+
"eval_stsb_spearman": 0.8388218137553312,
|
| 797 |
+
"step": 11625
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.75,
|
| 801 |
+
"eval_avg_sts": 0.775711044555684,
|
| 802 |
+
"eval_sickr_spearman": 0.7125440315449804,
|
| 803 |
+
"eval_stsb_spearman": 0.8388780575663876,
|
| 804 |
+
"step": 11750
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.76,
|
| 808 |
+
"eval_avg_sts": 0.7756594965556575,
|
| 809 |
+
"eval_sickr_spearman": 0.710795363234923,
|
| 810 |
+
"eval_stsb_spearman": 0.840523629876392,
|
| 811 |
+
"step": 11875
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.77,
|
| 815 |
+
"learning_rate": 2.3204914885447333e-06,
|
| 816 |
+
"loss": 0.0002,
|
| 817 |
+
"step": 12000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.77,
|
| 821 |
+
"eval_avg_sts": 0.778271737124629,
|
| 822 |
+
"eval_sickr_spearman": 0.7131796271103109,
|
| 823 |
+
"eval_stsb_spearman": 0.8433638471389473,
|
| 824 |
+
"step": 12000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.78,
|
| 828 |
+
"eval_avg_sts": 0.7778824941586701,
|
| 829 |
+
"eval_sickr_spearman": 0.7129357251772082,
|
| 830 |
+
"eval_stsb_spearman": 0.842829263140132,
|
| 831 |
+
"step": 12125
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.78,
|
| 835 |
+
"eval_avg_sts": 0.7791778033988318,
|
| 836 |
+
"eval_sickr_spearman": 0.7154212866455291,
|
| 837 |
+
"eval_stsb_spearman": 0.8429343201521345,
|
| 838 |
+
"step": 12250
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.79,
|
| 842 |
+
"eval_avg_sts": 0.7784491802919915,
|
| 843 |
+
"eval_sickr_spearman": 0.7169989642344666,
|
| 844 |
+
"eval_stsb_spearman": 0.8398993963495163,
|
| 845 |
+
"step": 12375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.8,
|
| 849 |
+
"learning_rate": 2.000511967234097e-06,
|
| 850 |
+
"loss": 0.0003,
|
| 851 |
+
"step": 12500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 0.8,
|
| 855 |
+
"eval_avg_sts": 0.7782960644263726,
|
| 856 |
+
"eval_sickr_spearman": 0.7171246616269303,
|
| 857 |
+
"eval_stsb_spearman": 0.8394674672258147,
|
| 858 |
+
"step": 12500
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.81,
|
| 862 |
+
"eval_avg_sts": 0.7777553880232895,
|
| 863 |
+
"eval_sickr_spearman": 0.7163744638535685,
|
| 864 |
+
"eval_stsb_spearman": 0.8391363121930105,
|
| 865 |
+
"step": 12625
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.82,
|
| 869 |
+
"eval_avg_sts": 0.7768079269313797,
|
| 870 |
+
"eval_sickr_spearman": 0.7156439108006947,
|
| 871 |
+
"eval_stsb_spearman": 0.8379719430620648,
|
| 872 |
+
"step": 12750
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.82,
|
| 876 |
+
"eval_avg_sts": 0.7769940996454496,
|
| 877 |
+
"eval_sickr_spearman": 0.7154824302376598,
|
| 878 |
+
"eval_stsb_spearman": 0.8385057690532396,
|
| 879 |
+
"step": 12875
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.83,
|
| 883 |
+
"learning_rate": 1.6805324459234608e-06,
|
| 884 |
+
"loss": 0.0002,
|
| 885 |
+
"step": 13000
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.83,
|
| 889 |
+
"eval_avg_sts": 0.7773811376862301,
|
| 890 |
+
"eval_sickr_spearman": 0.7153101426768013,
|
| 891 |
+
"eval_stsb_spearman": 0.839452132695659,
|
| 892 |
+
"step": 13000
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.84,
|
| 896 |
+
"eval_avg_sts": 0.7765844510293946,
|
| 897 |
+
"eval_sickr_spearman": 0.713703022022682,
|
| 898 |
+
"eval_stsb_spearman": 0.8394658800361071,
|
| 899 |
+
"step": 13125
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.85,
|
| 903 |
+
"eval_avg_sts": 0.7754935189329493,
|
| 904 |
+
"eval_sickr_spearman": 0.7128497975367355,
|
| 905 |
+
"eval_stsb_spearman": 0.838137240329163,
|
| 906 |
+
"step": 13250
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.86,
|
| 910 |
+
"eval_avg_sts": 0.7753794499527557,
|
| 911 |
+
"eval_sickr_spearman": 0.7128604124101533,
|
| 912 |
+
"eval_stsb_spearman": 0.837898487495358,
|
| 913 |
+
"step": 13375
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.86,
|
| 917 |
+
"learning_rate": 1.3605529246128248e-06,
|
| 918 |
+
"loss": 0.0001,
|
| 919 |
+
"step": 13500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.86,
|
| 923 |
+
"eval_avg_sts": 0.7753019292848522,
|
| 924 |
+
"eval_sickr_spearman": 0.7128690099773107,
|
| 925 |
+
"eval_stsb_spearman": 0.8377348485923936,
|
| 926 |
+
"step": 13500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.87,
|
| 930 |
+
"eval_avg_sts": 0.7751785379635656,
|
| 931 |
+
"eval_sickr_spearman": 0.7127431615818965,
|
| 932 |
+
"eval_stsb_spearman": 0.8376139143452346,
|
| 933 |
+
"step": 13625
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.88,
|
| 937 |
+
"eval_avg_sts": 0.7751092404866357,
|
| 938 |
+
"eval_sickr_spearman": 0.712538940248228,
|
| 939 |
+
"eval_stsb_spearman": 0.8376795407250434,
|
| 940 |
+
"step": 13750
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.89,
|
| 944 |
+
"eval_avg_sts": 0.7748652873780293,
|
| 945 |
+
"eval_sickr_spearman": 0.7121167948976883,
|
| 946 |
+
"eval_stsb_spearman": 0.8376137798583702,
|
| 947 |
+
"step": 13875
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.9,
|
| 951 |
+
"learning_rate": 1.0405734033021888e-06,
|
| 952 |
+
"loss": 0.0,
|
| 953 |
+
"step": 14000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.9,
|
| 957 |
+
"eval_avg_sts": 0.7749284965089291,
|
| 958 |
+
"eval_sickr_spearman": 0.7121870643990923,
|
| 959 |
+
"eval_stsb_spearman": 0.8376699286187659,
|
| 960 |
+
"step": 14000
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.9,
|
| 964 |
+
"eval_avg_sts": 0.7736810149784585,
|
| 965 |
+
"eval_sickr_spearman": 0.7101105357906183,
|
| 966 |
+
"eval_stsb_spearman": 0.8372514941662987,
|
| 967 |
+
"step": 14125
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.91,
|
| 971 |
+
"eval_avg_sts": 0.7736952566877616,
|
| 972 |
+
"eval_sickr_spearman": 0.7101496811382905,
|
| 973 |
+
"eval_stsb_spearman": 0.8372408322372328,
|
| 974 |
+
"step": 14250
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.92,
|
| 978 |
+
"eval_avg_sts": 0.7737241698204669,
|
| 979 |
+
"eval_sickr_spearman": 0.7102500181091946,
|
| 980 |
+
"eval_stsb_spearman": 0.837198321531739,
|
| 981 |
+
"step": 14375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.93,
|
| 985 |
+
"learning_rate": 7.205938819915525e-07,
|
| 986 |
+
"loss": 0.0001,
|
| 987 |
+
"step": 14500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.93,
|
| 991 |
+
"eval_avg_sts": 0.7737795026917761,
|
| 992 |
+
"eval_sickr_spearman": 0.7103373866827106,
|
| 993 |
+
"eval_stsb_spearman": 0.8372216187008416,
|
| 994 |
+
"step": 14500
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.94,
|
| 998 |
+
"eval_avg_sts": 0.7736836444786954,
|
| 999 |
+
"eval_sickr_spearman": 0.7101872414596151,
|
| 1000 |
+
"eval_stsb_spearman": 0.8371800474977756,
|
| 1001 |
+
"step": 14625
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.94,
|
| 1005 |
+
"eval_avg_sts": 0.7734449184616194,
|
| 1006 |
+
"eval_sickr_spearman": 0.7102963681220825,
|
| 1007 |
+
"eval_stsb_spearman": 0.8365934688011564,
|
| 1008 |
+
"step": 14750
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.95,
|
| 1012 |
+
"eval_avg_sts": 0.7733655720089518,
|
| 1013 |
+
"eval_sickr_spearman": 0.7101887784548611,
|
| 1014 |
+
"eval_stsb_spearman": 0.8365423655630426,
|
| 1015 |
+
"step": 14875
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.96,
|
| 1019 |
+
"learning_rate": 4.0061436068091647e-07,
|
| 1020 |
+
"loss": 0.0001,
|
| 1021 |
+
"step": 15000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 0.96,
|
| 1025 |
+
"eval_avg_sts": 0.7735006567182534,
|
| 1026 |
+
"eval_sickr_spearman": 0.7106506935573914,
|
| 1027 |
+
"eval_stsb_spearman": 0.8363506198791154,
|
| 1028 |
+
"step": 15000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.97,
|
| 1032 |
+
"eval_avg_sts": 0.7733751888661335,
|
| 1033 |
+
"eval_sickr_spearman": 0.710443343292483,
|
| 1034 |
+
"eval_stsb_spearman": 0.8363070344397839,
|
| 1035 |
+
"step": 15125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.98,
|
| 1039 |
+
"eval_avg_sts": 0.7736170866924498,
|
| 1040 |
+
"eval_sickr_spearman": 0.7105904625561881,
|
| 1041 |
+
"eval_stsb_spearman": 0.8366437108287115,
|
| 1042 |
+
"step": 15250
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 0.98,
|
| 1046 |
+
"eval_avg_sts": 0.7738120865784781,
|
| 1047 |
+
"eval_sickr_spearman": 0.7107394070017475,
|
| 1048 |
+
"eval_stsb_spearman": 0.8368847661552088,
|
| 1049 |
+
"step": 15375
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 0.99,
|
| 1053 |
+
"learning_rate": 8.06348393702803e-08,
|
| 1054 |
+
"loss": 0.0002,
|
| 1055 |
+
"step": 15500
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.99,
|
| 1059 |
+
"eval_avg_sts": 0.7737627481564555,
|
| 1060 |
+
"eval_sickr_spearman": 0.710675861854545,
|
| 1061 |
+
"eval_stsb_spearman": 0.8368496344583661,
|
| 1062 |
+
"step": 15500
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.0,
|
| 1066 |
+
"eval_avg_sts": 0.7737442273397684,
|
| 1067 |
+
"eval_sickr_spearman": 0.7106493006554496,
|
| 1068 |
+
"eval_stsb_spearman": 0.8368391540240873,
|
| 1069 |
+
"step": 15625
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 1.0,
|
| 1073 |
+
"step": 15626,
|
| 1074 |
+
"train_runtime": 3862.446,
|
| 1075 |
+
"train_samples_per_second": 4.046
|
| 1076 |
+
}
|
| 1077 |
+
],
|
| 1078 |
+
"max_steps": 15626,
|
| 1079 |
+
"num_train_epochs": 1,
|
| 1080 |
+
"total_flos": 48090729552850944,
|
| 1081 |
+
"trial_name": null,
|
| 1082 |
+
"trial_params": null
|
| 1083 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:10411d805de1d2657380d417e399fbf47a891c0681176da1ababf6ca68326c35
|
| 3 |
+
size 2095
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|