Upload 9 files
Browse files- config.json +24 -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
- trainer_state.json +1083 -0
- training_args.bin +3 -0
- vocab.txt +0 -0
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "result/noextendneg-202112301439/unsup-53",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForCL"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"gradient_checkpointing": false,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.1,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 3072,
|
| 13 |
+
"layer_norm_eps": 1e-12,
|
| 14 |
+
"max_position_embeddings": 512,
|
| 15 |
+
"model_type": "bert",
|
| 16 |
+
"num_attention_heads": 12,
|
| 17 |
+
"num_hidden_layers": 12,
|
| 18 |
+
"pad_token_id": 0,
|
| 19 |
+
"position_embedding_type": "absolute",
|
| 20 |
+
"transformers_version": "4.2.1",
|
| 21 |
+
"type_vocab_size": 2,
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 30522
|
| 24 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2c3622c6e9eb3636041a98faac5322df8b01a5459bb6f283dd1af4aeb9d892b
|
| 3 |
+
size 1751965867
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d58e844d005997d74808332c779f4f0d9007cd4ffabcd144fab5384b7b645e6
|
| 3 |
+
size 876037055
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:032b000ce609f915a2c18c31df114adb1cd8fcc27a455360e357e7fb494712b6
|
| 3 |
+
size 623
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "name_or_path": "bert-base-uncased"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1083 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.838238453177042,
|
| 3 |
+
"best_model_checkpoint": "result/noextendneg-202112301439/unsup-53",
|
| 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": 0.6824317233771873,
|
| 13 |
+
"eval_sickr_spearman": 0.6565727646348289,
|
| 14 |
+
"eval_stsb_spearman": 0.7082906821195456,
|
| 15 |
+
"step": 125
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.02,
|
| 19 |
+
"eval_avg_sts": 0.7088615925204196,
|
| 20 |
+
"eval_sickr_spearman": 0.6830176324024222,
|
| 21 |
+
"eval_stsb_spearman": 0.7347055526384171,
|
| 22 |
+
"step": 250
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.02,
|
| 26 |
+
"eval_avg_sts": 0.748889792906392,
|
| 27 |
+
"eval_sickr_spearman": 0.7234262941045092,
|
| 28 |
+
"eval_stsb_spearman": 0.7743532917082747,
|
| 29 |
+
"step": 375
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.03,
|
| 33 |
+
"learning_rate": 2.9040061436068092e-05,
|
| 34 |
+
"loss": 1.4942,
|
| 35 |
+
"step": 500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.03,
|
| 39 |
+
"eval_avg_sts": 0.7647833717747232,
|
| 40 |
+
"eval_sickr_spearman": 0.736218225226116,
|
| 41 |
+
"eval_stsb_spearman": 0.7933485183233304,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.04,
|
| 46 |
+
"eval_avg_sts": 0.7671547263733642,
|
| 47 |
+
"eval_sickr_spearman": 0.7419857518557031,
|
| 48 |
+
"eval_stsb_spearman": 0.7923237008910253,
|
| 49 |
+
"step": 625
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.05,
|
| 53 |
+
"eval_avg_sts": 0.7743718035568281,
|
| 54 |
+
"eval_sickr_spearman": 0.7418156256944094,
|
| 55 |
+
"eval_stsb_spearman": 0.8069279814192469,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.06,
|
| 60 |
+
"eval_avg_sts": 0.778288874388263,
|
| 61 |
+
"eval_sickr_spearman": 0.7432141953060842,
|
| 62 |
+
"eval_stsb_spearman": 0.8133635534704418,
|
| 63 |
+
"step": 875
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.06,
|
| 67 |
+
"learning_rate": 2.8080122872136184e-05,
|
| 68 |
+
"loss": 1.3761,
|
| 69 |
+
"step": 1000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.06,
|
| 73 |
+
"eval_avg_sts": 0.7820337312720824,
|
| 74 |
+
"eval_sickr_spearman": 0.7496537250758921,
|
| 75 |
+
"eval_stsb_spearman": 0.8144137374682728,
|
| 76 |
+
"step": 1000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.07,
|
| 80 |
+
"eval_avg_sts": 0.7843418934590822,
|
| 81 |
+
"eval_sickr_spearman": 0.74964997864998,
|
| 82 |
+
"eval_stsb_spearman": 0.8190338082681845,
|
| 83 |
+
"step": 1125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.08,
|
| 87 |
+
"eval_avg_sts": 0.780707868643423,
|
| 88 |
+
"eval_sickr_spearman": 0.7451549399907927,
|
| 89 |
+
"eval_stsb_spearman": 0.8162607972960534,
|
| 90 |
+
"step": 1250
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09,
|
| 94 |
+
"eval_avg_sts": 0.7822364414457996,
|
| 95 |
+
"eval_sickr_spearman": 0.7458706514333221,
|
| 96 |
+
"eval_stsb_spearman": 0.818602231458277,
|
| 97 |
+
"step": 1375
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1,
|
| 101 |
+
"learning_rate": 2.7120184308204276e-05,
|
| 102 |
+
"loss": 1.3752,
|
| 103 |
+
"step": 1500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.1,
|
| 107 |
+
"eval_avg_sts": 0.784082685057127,
|
| 108 |
+
"eval_sickr_spearman": 0.7478067851322924,
|
| 109 |
+
"eval_stsb_spearman": 0.8203585849819615,
|
| 110 |
+
"step": 1500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.1,
|
| 114 |
+
"eval_avg_sts": 0.7822808064244355,
|
| 115 |
+
"eval_sickr_spearman": 0.7476906459290149,
|
| 116 |
+
"eval_stsb_spearman": 0.8168709669198561,
|
| 117 |
+
"step": 1625
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.11,
|
| 121 |
+
"eval_avg_sts": 0.788703629999421,
|
| 122 |
+
"eval_sickr_spearman": 0.7475929026375885,
|
| 123 |
+
"eval_stsb_spearman": 0.8298143573612535,
|
| 124 |
+
"step": 1750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.12,
|
| 128 |
+
"eval_avg_sts": 0.7896517154212715,
|
| 129 |
+
"eval_sickr_spearman": 0.7467137893879666,
|
| 130 |
+
"eval_stsb_spearman": 0.8325896414545765,
|
| 131 |
+
"step": 1875
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.13,
|
| 135 |
+
"learning_rate": 2.6160245744272367e-05,
|
| 136 |
+
"loss": 1.3749,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.13,
|
| 141 |
+
"eval_avg_sts": 0.7848601188664759,
|
| 142 |
+
"eval_sickr_spearman": 0.7478141338908124,
|
| 143 |
+
"eval_stsb_spearman": 0.8219061038421395,
|
| 144 |
+
"step": 2000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.14,
|
| 148 |
+
"eval_avg_sts": 0.791993659168045,
|
| 149 |
+
"eval_sickr_spearman": 0.7493434921917033,
|
| 150 |
+
"eval_stsb_spearman": 0.8346438261443867,
|
| 151 |
+
"step": 2125
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.14,
|
| 155 |
+
"eval_avg_sts": 0.7912909275247073,
|
| 156 |
+
"eval_sickr_spearman": 0.750289032454614,
|
| 157 |
+
"eval_stsb_spearman": 0.8322928225948005,
|
| 158 |
+
"step": 2250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.15,
|
| 162 |
+
"eval_avg_sts": 0.789035346084623,
|
| 163 |
+
"eval_sickr_spearman": 0.7492525212855795,
|
| 164 |
+
"eval_stsb_spearman": 0.8288181708836665,
|
| 165 |
+
"step": 2375
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.16,
|
| 169 |
+
"learning_rate": 2.5200307180340456e-05,
|
| 170 |
+
"loss": 1.3744,
|
| 171 |
+
"step": 2500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.16,
|
| 175 |
+
"eval_avg_sts": 0.7856599152275197,
|
| 176 |
+
"eval_sickr_spearman": 0.7458348202316492,
|
| 177 |
+
"eval_stsb_spearman": 0.8254850102233903,
|
| 178 |
+
"step": 2500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.17,
|
| 182 |
+
"eval_avg_sts": 0.7873291440307411,
|
| 183 |
+
"eval_sickr_spearman": 0.7489932494000165,
|
| 184 |
+
"eval_stsb_spearman": 0.8256650386614655,
|
| 185 |
+
"step": 2625
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.18,
|
| 189 |
+
"eval_avg_sts": 0.7887021861862142,
|
| 190 |
+
"eval_sickr_spearman": 0.7506508027106459,
|
| 191 |
+
"eval_stsb_spearman": 0.8267535696617826,
|
| 192 |
+
"step": 2750
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.18,
|
| 196 |
+
"eval_avg_sts": 0.7908330013178573,
|
| 197 |
+
"eval_sickr_spearman": 0.7434275494586724,
|
| 198 |
+
"eval_stsb_spearman": 0.838238453177042,
|
| 199 |
+
"step": 2875
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.19,
|
| 203 |
+
"learning_rate": 2.424036861640855e-05,
|
| 204 |
+
"loss": 1.3743,
|
| 205 |
+
"step": 3000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.19,
|
| 209 |
+
"eval_avg_sts": 0.7864405140650716,
|
| 210 |
+
"eval_sickr_spearman": 0.7452477299492377,
|
| 211 |
+
"eval_stsb_spearman": 0.8276332981809055,
|
| 212 |
+
"step": 3000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2,
|
| 216 |
+
"eval_avg_sts": 0.7856684955860707,
|
| 217 |
+
"eval_sickr_spearman": 0.7435629971647278,
|
| 218 |
+
"eval_stsb_spearman": 0.8277739940074136,
|
| 219 |
+
"step": 3125
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.21,
|
| 223 |
+
"eval_avg_sts": 0.7859315388383803,
|
| 224 |
+
"eval_sickr_spearman": 0.7420500174694273,
|
| 225 |
+
"eval_stsb_spearman": 0.8298130602073334,
|
| 226 |
+
"step": 3250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.22,
|
| 230 |
+
"eval_avg_sts": 0.7859735969509316,
|
| 231 |
+
"eval_sickr_spearman": 0.7443584402356445,
|
| 232 |
+
"eval_stsb_spearman": 0.8275887536662188,
|
| 233 |
+
"step": 3375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.22,
|
| 237 |
+
"learning_rate": 2.3280430052476642e-05,
|
| 238 |
+
"loss": 1.3743,
|
| 239 |
+
"step": 3500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.22,
|
| 243 |
+
"eval_avg_sts": 0.7842904145859011,
|
| 244 |
+
"eval_sickr_spearman": 0.745185343678003,
|
| 245 |
+
"eval_stsb_spearman": 0.823395485493799,
|
| 246 |
+
"step": 3500
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.23,
|
| 250 |
+
"eval_avg_sts": 0.7839231555379245,
|
| 251 |
+
"eval_sickr_spearman": 0.744413964188907,
|
| 252 |
+
"eval_stsb_spearman": 0.8234323468869418,
|
| 253 |
+
"step": 3625
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.24,
|
| 257 |
+
"eval_avg_sts": 0.7854707830375165,
|
| 258 |
+
"eval_sickr_spearman": 0.745022758399635,
|
| 259 |
+
"eval_stsb_spearman": 0.8259188076753979,
|
| 260 |
+
"step": 3750
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.25,
|
| 264 |
+
"eval_avg_sts": 0.7850613325633844,
|
| 265 |
+
"eval_sickr_spearman": 0.7463711294077354,
|
| 266 |
+
"eval_stsb_spearman": 0.8237515357190333,
|
| 267 |
+
"step": 3875
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.26,
|
| 271 |
+
"learning_rate": 2.2320491488544734e-05,
|
| 272 |
+
"loss": 1.3742,
|
| 273 |
+
"step": 4000
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.26,
|
| 277 |
+
"eval_avg_sts": 0.7830541412616541,
|
| 278 |
+
"eval_sickr_spearman": 0.7444240507202089,
|
| 279 |
+
"eval_stsb_spearman": 0.8216842318030994,
|
| 280 |
+
"step": 4000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.26,
|
| 284 |
+
"eval_avg_sts": 0.7813602214602952,
|
| 285 |
+
"eval_sickr_spearman": 0.741438773672526,
|
| 286 |
+
"eval_stsb_spearman": 0.8212816692480643,
|
| 287 |
+
"step": 4125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.27,
|
| 291 |
+
"eval_avg_sts": 0.7789870537555355,
|
| 292 |
+
"eval_sickr_spearman": 0.7424057838377794,
|
| 293 |
+
"eval_stsb_spearman": 0.8155683236732915,
|
| 294 |
+
"step": 4250
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.28,
|
| 298 |
+
"eval_avg_sts": 0.7813460307361557,
|
| 299 |
+
"eval_sickr_spearman": 0.7435493563319195,
|
| 300 |
+
"eval_stsb_spearman": 0.8191427051403918,
|
| 301 |
+
"step": 4375
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.29,
|
| 305 |
+
"learning_rate": 2.1360552924612826e-05,
|
| 306 |
+
"loss": 1.3743,
|
| 307 |
+
"step": 4500
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.29,
|
| 311 |
+
"eval_avg_sts": 0.776201517183297,
|
| 312 |
+
"eval_sickr_spearman": 0.7414799843575598,
|
| 313 |
+
"eval_stsb_spearman": 0.8109230500090343,
|
| 314 |
+
"step": 4500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.3,
|
| 318 |
+
"eval_avg_sts": 0.779063792601691,
|
| 319 |
+
"eval_sickr_spearman": 0.7449220371799193,
|
| 320 |
+
"eval_stsb_spearman": 0.8132055480234627,
|
| 321 |
+
"step": 4625
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.3,
|
| 325 |
+
"eval_avg_sts": 0.7771260741272044,
|
| 326 |
+
"eval_sickr_spearman": 0.7445626204478579,
|
| 327 |
+
"eval_stsb_spearman": 0.8096895278065507,
|
| 328 |
+
"step": 4750
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.31,
|
| 332 |
+
"eval_avg_sts": 0.7792767945158564,
|
| 333 |
+
"eval_sickr_spearman": 0.7424322970057733,
|
| 334 |
+
"eval_stsb_spearman": 0.8161212920259395,
|
| 335 |
+
"step": 4875
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.32,
|
| 339 |
+
"learning_rate": 2.0400614360680917e-05,
|
| 340 |
+
"loss": 1.3741,
|
| 341 |
+
"step": 5000
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.32,
|
| 345 |
+
"eval_avg_sts": 0.7801777270254462,
|
| 346 |
+
"eval_sickr_spearman": 0.745189954663741,
|
| 347 |
+
"eval_stsb_spearman": 0.8151654993871514,
|
| 348 |
+
"step": 5000
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.33,
|
| 352 |
+
"eval_avg_sts": 0.7790507403010363,
|
| 353 |
+
"eval_sickr_spearman": 0.7436056968139063,
|
| 354 |
+
"eval_stsb_spearman": 0.8144957837881661,
|
| 355 |
+
"step": 5125
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.34,
|
| 359 |
+
"eval_avg_sts": 0.7754481116510958,
|
| 360 |
+
"eval_sickr_spearman": 0.7431048765192112,
|
| 361 |
+
"eval_stsb_spearman": 0.8077913467829804,
|
| 362 |
+
"step": 5250
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.34,
|
| 366 |
+
"eval_avg_sts": 0.7759560400168242,
|
| 367 |
+
"eval_sickr_spearman": 0.7460634002433932,
|
| 368 |
+
"eval_stsb_spearman": 0.805848679790255,
|
| 369 |
+
"step": 5375
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.35,
|
| 373 |
+
"learning_rate": 1.944067579674901e-05,
|
| 374 |
+
"loss": 1.374,
|
| 375 |
+
"step": 5500
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.35,
|
| 379 |
+
"eval_avg_sts": 0.7779457482540455,
|
| 380 |
+
"eval_sickr_spearman": 0.7447529677028571,
|
| 381 |
+
"eval_stsb_spearman": 0.8111385288052337,
|
| 382 |
+
"step": 5500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.36,
|
| 386 |
+
"eval_avg_sts": 0.7812654214930765,
|
| 387 |
+
"eval_sickr_spearman": 0.7431360487040445,
|
| 388 |
+
"eval_stsb_spearman": 0.8193947942821087,
|
| 389 |
+
"step": 5625
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.37,
|
| 393 |
+
"eval_avg_sts": 0.7791932156518167,
|
| 394 |
+
"eval_sickr_spearman": 0.7427627029525661,
|
| 395 |
+
"eval_stsb_spearman": 0.8156237283510671,
|
| 396 |
+
"step": 5750
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.38,
|
| 400 |
+
"eval_avg_sts": 0.781330663934428,
|
| 401 |
+
"eval_sickr_spearman": 0.740891939582653,
|
| 402 |
+
"eval_stsb_spearman": 0.8217693882862028,
|
| 403 |
+
"step": 5875
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.38,
|
| 407 |
+
"learning_rate": 1.84807372328171e-05,
|
| 408 |
+
"loss": 1.3743,
|
| 409 |
+
"step": 6000
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.38,
|
| 413 |
+
"eval_avg_sts": 0.7792436520419758,
|
| 414 |
+
"eval_sickr_spearman": 0.7425573219628167,
|
| 415 |
+
"eval_stsb_spearman": 0.8159299821211349,
|
| 416 |
+
"step": 6000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.39,
|
| 420 |
+
"eval_avg_sts": 0.778491260905172,
|
| 421 |
+
"eval_sickr_spearman": 0.7429476707242042,
|
| 422 |
+
"eval_stsb_spearman": 0.8140348510861396,
|
| 423 |
+
"step": 6125
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.4,
|
| 427 |
+
"eval_avg_sts": 0.7781813540457969,
|
| 428 |
+
"eval_sickr_spearman": 0.7442114650652438,
|
| 429 |
+
"eval_stsb_spearman": 0.8121512430263501,
|
| 430 |
+
"step": 6250
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.41,
|
| 434 |
+
"eval_avg_sts": 0.7746842721533463,
|
| 435 |
+
"eval_sickr_spearman": 0.7427369582821953,
|
| 436 |
+
"eval_stsb_spearman": 0.8066315860244974,
|
| 437 |
+
"step": 6375
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.42,
|
| 441 |
+
"learning_rate": 1.7520798668885192e-05,
|
| 442 |
+
"loss": 1.3741,
|
| 443 |
+
"step": 6500
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.42,
|
| 447 |
+
"eval_avg_sts": 0.774494584762071,
|
| 448 |
+
"eval_sickr_spearman": 0.7398474072196785,
|
| 449 |
+
"eval_stsb_spearman": 0.8091417623044636,
|
| 450 |
+
"step": 6500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.42,
|
| 454 |
+
"eval_avg_sts": 0.7766158847039688,
|
| 455 |
+
"eval_sickr_spearman": 0.7414669679290701,
|
| 456 |
+
"eval_stsb_spearman": 0.8117648014788674,
|
| 457 |
+
"step": 6625
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.43,
|
| 461 |
+
"eval_avg_sts": 0.7785386253067408,
|
| 462 |
+
"eval_sickr_spearman": 0.7400865540737388,
|
| 463 |
+
"eval_stsb_spearman": 0.8169906965397429,
|
| 464 |
+
"step": 6750
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.44,
|
| 468 |
+
"eval_avg_sts": 0.7773462713510697,
|
| 469 |
+
"eval_sickr_spearman": 0.7418193240892201,
|
| 470 |
+
"eval_stsb_spearman": 0.8128732186129194,
|
| 471 |
+
"step": 6875
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.45,
|
| 475 |
+
"learning_rate": 1.6560860104953284e-05,
|
| 476 |
+
"loss": 1.3739,
|
| 477 |
+
"step": 7000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.45,
|
| 481 |
+
"eval_avg_sts": 0.7731374701085579,
|
| 482 |
+
"eval_sickr_spearman": 0.7378116089852239,
|
| 483 |
+
"eval_stsb_spearman": 0.808463331231892,
|
| 484 |
+
"step": 7000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.46,
|
| 488 |
+
"eval_avg_sts": 0.7740007626157097,
|
| 489 |
+
"eval_sickr_spearman": 0.7376639133483018,
|
| 490 |
+
"eval_stsb_spearman": 0.8103376118831176,
|
| 491 |
+
"step": 7125
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.46,
|
| 495 |
+
"eval_avg_sts": 0.7765981231805895,
|
| 496 |
+
"eval_sickr_spearman": 0.7397111429848985,
|
| 497 |
+
"eval_stsb_spearman": 0.8134851033762805,
|
| 498 |
+
"step": 7250
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.47,
|
| 502 |
+
"eval_avg_sts": 0.7743899061192407,
|
| 503 |
+
"eval_sickr_spearman": 0.7414097148561558,
|
| 504 |
+
"eval_stsb_spearman": 0.8073700973823258,
|
| 505 |
+
"step": 7375
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.48,
|
| 509 |
+
"learning_rate": 1.5600921541021372e-05,
|
| 510 |
+
"loss": 1.3744,
|
| 511 |
+
"step": 7500
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 0.48,
|
| 515 |
+
"eval_avg_sts": 0.7788123677105294,
|
| 516 |
+
"eval_sickr_spearman": 0.7435954661893001,
|
| 517 |
+
"eval_stsb_spearman": 0.8140292692317587,
|
| 518 |
+
"step": 7500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.49,
|
| 522 |
+
"eval_avg_sts": 0.7772272696959476,
|
| 523 |
+
"eval_sickr_spearman": 0.7404882381750656,
|
| 524 |
+
"eval_stsb_spearman": 0.8139663012168296,
|
| 525 |
+
"step": 7625
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.5,
|
| 529 |
+
"eval_avg_sts": 0.7794965841665269,
|
| 530 |
+
"eval_sickr_spearman": 0.742402998033896,
|
| 531 |
+
"eval_stsb_spearman": 0.8165901702991577,
|
| 532 |
+
"step": 7750
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.5,
|
| 536 |
+
"eval_avg_sts": 0.7720147134204476,
|
| 537 |
+
"eval_sickr_spearman": 0.739454224623306,
|
| 538 |
+
"eval_stsb_spearman": 0.8045752022175892,
|
| 539 |
+
"step": 7875
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.51,
|
| 543 |
+
"learning_rate": 1.4640982977089467e-05,
|
| 544 |
+
"loss": 1.3744,
|
| 545 |
+
"step": 8000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.51,
|
| 549 |
+
"eval_avg_sts": 0.7772065286971521,
|
| 550 |
+
"eval_sickr_spearman": 0.7399364088506433,
|
| 551 |
+
"eval_stsb_spearman": 0.814476648543661,
|
| 552 |
+
"step": 8000
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.52,
|
| 556 |
+
"eval_avg_sts": 0.7732682543973268,
|
| 557 |
+
"eval_sickr_spearman": 0.7381198245631523,
|
| 558 |
+
"eval_stsb_spearman": 0.8084166842315015,
|
| 559 |
+
"step": 8125
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.53,
|
| 563 |
+
"eval_avg_sts": 0.7754551024695455,
|
| 564 |
+
"eval_sickr_spearman": 0.739559796984267,
|
| 565 |
+
"eval_stsb_spearman": 0.8113504079548242,
|
| 566 |
+
"step": 8250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.54,
|
| 570 |
+
"eval_avg_sts": 0.7752465045338841,
|
| 571 |
+
"eval_sickr_spearman": 0.7382731878700443,
|
| 572 |
+
"eval_stsb_spearman": 0.8122198211977241,
|
| 573 |
+
"step": 8375
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.54,
|
| 577 |
+
"learning_rate": 1.368104441315756e-05,
|
| 578 |
+
"loss": 1.374,
|
| 579 |
+
"step": 8500
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.54,
|
| 583 |
+
"eval_avg_sts": 0.7738482342971208,
|
| 584 |
+
"eval_sickr_spearman": 0.7392923117803581,
|
| 585 |
+
"eval_stsb_spearman": 0.8084041568138834,
|
| 586 |
+
"step": 8500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.55,
|
| 590 |
+
"eval_avg_sts": 0.7798108521674825,
|
| 591 |
+
"eval_sickr_spearman": 0.739234530365328,
|
| 592 |
+
"eval_stsb_spearman": 0.820387173969637,
|
| 593 |
+
"step": 8625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.56,
|
| 597 |
+
"eval_avg_sts": 0.7757746119221507,
|
| 598 |
+
"eval_sickr_spearman": 0.7358846972577298,
|
| 599 |
+
"eval_stsb_spearman": 0.8156645265865717,
|
| 600 |
+
"step": 8750
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.57,
|
| 604 |
+
"eval_avg_sts": 0.7754440529084836,
|
| 605 |
+
"eval_sickr_spearman": 0.7365315801318982,
|
| 606 |
+
"eval_stsb_spearman": 0.8143565256850689,
|
| 607 |
+
"step": 8875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.58,
|
| 611 |
+
"learning_rate": 1.2721105849225649e-05,
|
| 612 |
+
"loss": 1.3738,
|
| 613 |
+
"step": 9000
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.58,
|
| 617 |
+
"eval_avg_sts": 0.7752604709709456,
|
| 618 |
+
"eval_sickr_spearman": 0.7375815400093355,
|
| 619 |
+
"eval_stsb_spearman": 0.8129394019325558,
|
| 620 |
+
"step": 9000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.58,
|
| 624 |
+
"eval_avg_sts": 0.7719041040932904,
|
| 625 |
+
"eval_sickr_spearman": 0.7360112528588094,
|
| 626 |
+
"eval_stsb_spearman": 0.8077969553277713,
|
| 627 |
+
"step": 9125
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.59,
|
| 631 |
+
"eval_avg_sts": 0.7740838243178477,
|
| 632 |
+
"eval_sickr_spearman": 0.737044840481866,
|
| 633 |
+
"eval_stsb_spearman": 0.8111228081538294,
|
| 634 |
+
"step": 9250
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.6,
|
| 638 |
+
"eval_avg_sts": 0.7744233573614463,
|
| 639 |
+
"eval_sickr_spearman": 0.7352095720959158,
|
| 640 |
+
"eval_stsb_spearman": 0.8136371426269768,
|
| 641 |
+
"step": 9375
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.61,
|
| 645 |
+
"learning_rate": 1.176116728529374e-05,
|
| 646 |
+
"loss": 1.3739,
|
| 647 |
+
"step": 9500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.61,
|
| 651 |
+
"eval_avg_sts": 0.7728152704473774,
|
| 652 |
+
"eval_sickr_spearman": 0.7345371366757821,
|
| 653 |
+
"eval_stsb_spearman": 0.8110934042189727,
|
| 654 |
+
"step": 9500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.62,
|
| 658 |
+
"eval_avg_sts": 0.7748003786183549,
|
| 659 |
+
"eval_sickr_spearman": 0.736938499623282,
|
| 660 |
+
"eval_stsb_spearman": 0.8126622576134279,
|
| 661 |
+
"step": 9625
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.62,
|
| 665 |
+
"eval_avg_sts": 0.7736787434429957,
|
| 666 |
+
"eval_sickr_spearman": 0.736373461745964,
|
| 667 |
+
"eval_stsb_spearman": 0.8109840251400274,
|
| 668 |
+
"step": 9750
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.63,
|
| 672 |
+
"eval_avg_sts": 0.7738638985267913,
|
| 673 |
+
"eval_sickr_spearman": 0.7368789890886002,
|
| 674 |
+
"eval_stsb_spearman": 0.8108488079649824,
|
| 675 |
+
"step": 9875
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.64,
|
| 679 |
+
"learning_rate": 1.0801228721361832e-05,
|
| 680 |
+
"loss": 1.3741,
|
| 681 |
+
"step": 10000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.64,
|
| 685 |
+
"eval_avg_sts": 0.7745000944024674,
|
| 686 |
+
"eval_sickr_spearman": 0.7357973767153152,
|
| 687 |
+
"eval_stsb_spearman": 0.8132028120896195,
|
| 688 |
+
"step": 10000
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.65,
|
| 692 |
+
"eval_avg_sts": 0.7755728981675435,
|
| 693 |
+
"eval_sickr_spearman": 0.7363922419066262,
|
| 694 |
+
"eval_stsb_spearman": 0.8147535544284608,
|
| 695 |
+
"step": 10125
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.66,
|
| 699 |
+
"eval_avg_sts": 0.7759902317249895,
|
| 700 |
+
"eval_sickr_spearman": 0.7375042579671216,
|
| 701 |
+
"eval_stsb_spearman": 0.8144762054828574,
|
| 702 |
+
"step": 10250
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.66,
|
| 706 |
+
"eval_avg_sts": 0.7738194667098928,
|
| 707 |
+
"eval_sickr_spearman": 0.7379243379802992,
|
| 708 |
+
"eval_stsb_spearman": 0.8097145954394862,
|
| 709 |
+
"step": 10375
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.67,
|
| 713 |
+
"learning_rate": 9.841290157429926e-06,
|
| 714 |
+
"loss": 1.3739,
|
| 715 |
+
"step": 10500
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.67,
|
| 719 |
+
"eval_avg_sts": 0.7752023601085498,
|
| 720 |
+
"eval_sickr_spearman": 0.7386433155377263,
|
| 721 |
+
"eval_stsb_spearman": 0.8117614046793732,
|
| 722 |
+
"step": 10500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.68,
|
| 726 |
+
"eval_avg_sts": 0.7748475183160914,
|
| 727 |
+
"eval_sickr_spearman": 0.7387013371082636,
|
| 728 |
+
"eval_stsb_spearman": 0.8109936995239194,
|
| 729 |
+
"step": 10625
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.69,
|
| 733 |
+
"eval_avg_sts": 0.7728917936913275,
|
| 734 |
+
"eval_sickr_spearman": 0.736409533103144,
|
| 735 |
+
"eval_stsb_spearman": 0.8093740542795109,
|
| 736 |
+
"step": 10750
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 0.7,
|
| 740 |
+
"eval_avg_sts": 0.7725897625926305,
|
| 741 |
+
"eval_sickr_spearman": 0.7348871393119617,
|
| 742 |
+
"eval_stsb_spearman": 0.8102923858732992,
|
| 743 |
+
"step": 10875
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 0.7,
|
| 747 |
+
"learning_rate": 8.881351593498018e-06,
|
| 748 |
+
"loss": 1.3738,
|
| 749 |
+
"step": 11000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 0.7,
|
| 753 |
+
"eval_avg_sts": 0.7728014259850515,
|
| 754 |
+
"eval_sickr_spearman": 0.733524785150771,
|
| 755 |
+
"eval_stsb_spearman": 0.8120780668193319,
|
| 756 |
+
"step": 11000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.71,
|
| 760 |
+
"eval_avg_sts": 0.7713921235193044,
|
| 761 |
+
"eval_sickr_spearman": 0.7339205133955196,
|
| 762 |
+
"eval_stsb_spearman": 0.8088637336430892,
|
| 763 |
+
"step": 11125
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.72,
|
| 767 |
+
"eval_avg_sts": 0.7728581212092439,
|
| 768 |
+
"eval_sickr_spearman": 0.7340948182626384,
|
| 769 |
+
"eval_stsb_spearman": 0.8116214241558494,
|
| 770 |
+
"step": 11250
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.73,
|
| 774 |
+
"eval_avg_sts": 0.7720358887305934,
|
| 775 |
+
"eval_sickr_spearman": 0.7342980378528232,
|
| 776 |
+
"eval_stsb_spearman": 0.8097737396083635,
|
| 777 |
+
"step": 11375
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.74,
|
| 781 |
+
"learning_rate": 7.921413029566108e-06,
|
| 782 |
+
"loss": 1.3737,
|
| 783 |
+
"step": 11500
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.74,
|
| 787 |
+
"eval_avg_sts": 0.7695547454198304,
|
| 788 |
+
"eval_sickr_spearman": 0.7319195377096078,
|
| 789 |
+
"eval_stsb_spearman": 0.8071899531300528,
|
| 790 |
+
"step": 11500
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.74,
|
| 794 |
+
"eval_avg_sts": 0.7706681719236664,
|
| 795 |
+
"eval_sickr_spearman": 0.732619896782625,
|
| 796 |
+
"eval_stsb_spearman": 0.8087164470647079,
|
| 797 |
+
"step": 11625
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.75,
|
| 801 |
+
"eval_avg_sts": 0.7696699833641414,
|
| 802 |
+
"eval_sickr_spearman": 0.7322283296607535,
|
| 803 |
+
"eval_stsb_spearman": 0.8071116370675294,
|
| 804 |
+
"step": 11750
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.76,
|
| 808 |
+
"eval_avg_sts": 0.7655661802332044,
|
| 809 |
+
"eval_sickr_spearman": 0.7308695473420087,
|
| 810 |
+
"eval_stsb_spearman": 0.8002628131244002,
|
| 811 |
+
"step": 11875
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.77,
|
| 815 |
+
"learning_rate": 6.961474465634199e-06,
|
| 816 |
+
"loss": 1.3738,
|
| 817 |
+
"step": 12000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.77,
|
| 821 |
+
"eval_avg_sts": 0.7683310753929219,
|
| 822 |
+
"eval_sickr_spearman": 0.733969889367798,
|
| 823 |
+
"eval_stsb_spearman": 0.8026922614180458,
|
| 824 |
+
"step": 12000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.78,
|
| 828 |
+
"eval_avg_sts": 0.7689113009272721,
|
| 829 |
+
"eval_sickr_spearman": 0.7331248782001972,
|
| 830 |
+
"eval_stsb_spearman": 0.8046977236543469,
|
| 831 |
+
"step": 12125
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.78,
|
| 835 |
+
"eval_avg_sts": 0.7727908372056179,
|
| 836 |
+
"eval_sickr_spearman": 0.7345837748752785,
|
| 837 |
+
"eval_stsb_spearman": 0.8109978995359574,
|
| 838 |
+
"step": 12250
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.79,
|
| 842 |
+
"eval_avg_sts": 0.7712377661783789,
|
| 843 |
+
"eval_sickr_spearman": 0.7337775248065384,
|
| 844 |
+
"eval_stsb_spearman": 0.8086980075502196,
|
| 845 |
+
"step": 12375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.8,
|
| 849 |
+
"learning_rate": 6.001535901702292e-06,
|
| 850 |
+
"loss": 1.3737,
|
| 851 |
+
"step": 12500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 0.8,
|
| 855 |
+
"eval_avg_sts": 0.772200684127089,
|
| 856 |
+
"eval_sickr_spearman": 0.7341291605001669,
|
| 857 |
+
"eval_stsb_spearman": 0.810272207754011,
|
| 858 |
+
"step": 12500
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.81,
|
| 862 |
+
"eval_avg_sts": 0.774531532169478,
|
| 863 |
+
"eval_sickr_spearman": 0.7340682090324417,
|
| 864 |
+
"eval_stsb_spearman": 0.8149948553065143,
|
| 865 |
+
"step": 12625
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.82,
|
| 869 |
+
"eval_avg_sts": 0.7725061483445825,
|
| 870 |
+
"eval_sickr_spearman": 0.7340744530756287,
|
| 871 |
+
"eval_stsb_spearman": 0.8109378436135363,
|
| 872 |
+
"step": 12750
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.82,
|
| 876 |
+
"eval_avg_sts": 0.7714926649368181,
|
| 877 |
+
"eval_sickr_spearman": 0.7341153275429525,
|
| 878 |
+
"eval_stsb_spearman": 0.8088700023306838,
|
| 879 |
+
"step": 12875
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.83,
|
| 883 |
+
"learning_rate": 5.0415973377703825e-06,
|
| 884 |
+
"loss": 1.3738,
|
| 885 |
+
"step": 13000
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.83,
|
| 889 |
+
"eval_avg_sts": 0.7722886101322686,
|
| 890 |
+
"eval_sickr_spearman": 0.7347338720672726,
|
| 891 |
+
"eval_stsb_spearman": 0.8098433481972646,
|
| 892 |
+
"step": 13000
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.84,
|
| 896 |
+
"eval_avg_sts": 0.7724511969333603,
|
| 897 |
+
"eval_sickr_spearman": 0.7351566418221308,
|
| 898 |
+
"eval_stsb_spearman": 0.8097457520445899,
|
| 899 |
+
"step": 13125
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.85,
|
| 903 |
+
"eval_avg_sts": 0.771434894841144,
|
| 904 |
+
"eval_sickr_spearman": 0.7347215761053044,
|
| 905 |
+
"eval_stsb_spearman": 0.8081482135769836,
|
| 906 |
+
"step": 13250
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.86,
|
| 910 |
+
"eval_avg_sts": 0.7714665080668874,
|
| 911 |
+
"eval_sickr_spearman": 0.7342130228032777,
|
| 912 |
+
"eval_stsb_spearman": 0.8087199933304972,
|
| 913 |
+
"step": 13375
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.86,
|
| 917 |
+
"learning_rate": 4.081658773838474e-06,
|
| 918 |
+
"loss": 1.3737,
|
| 919 |
+
"step": 13500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.86,
|
| 923 |
+
"eval_avg_sts": 0.7716889452949547,
|
| 924 |
+
"eval_sickr_spearman": 0.733557350237546,
|
| 925 |
+
"eval_stsb_spearman": 0.8098205403523633,
|
| 926 |
+
"step": 13500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.87,
|
| 930 |
+
"eval_avg_sts": 0.7708524022909773,
|
| 931 |
+
"eval_sickr_spearman": 0.7333273773238603,
|
| 932 |
+
"eval_stsb_spearman": 0.8083774272580941,
|
| 933 |
+
"step": 13625
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.88,
|
| 937 |
+
"eval_avg_sts": 0.7695297711312417,
|
| 938 |
+
"eval_sickr_spearman": 0.7321257832591831,
|
| 939 |
+
"eval_stsb_spearman": 0.8069337590033002,
|
| 940 |
+
"step": 13750
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.89,
|
| 944 |
+
"eval_avg_sts": 0.7723285421418242,
|
| 945 |
+
"eval_sickr_spearman": 0.7334592226973079,
|
| 946 |
+
"eval_stsb_spearman": 0.8111978615863407,
|
| 947 |
+
"step": 13875
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.9,
|
| 951 |
+
"learning_rate": 3.121720209906566e-06,
|
| 952 |
+
"loss": 1.3739,
|
| 953 |
+
"step": 14000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.9,
|
| 957 |
+
"eval_avg_sts": 0.7712955547799614,
|
| 958 |
+
"eval_sickr_spearman": 0.7343331966190758,
|
| 959 |
+
"eval_stsb_spearman": 0.8082579129408469,
|
| 960 |
+
"step": 14000
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.9,
|
| 964 |
+
"eval_avg_sts": 0.7718610397493488,
|
| 965 |
+
"eval_sickr_spearman": 0.7346661482142449,
|
| 966 |
+
"eval_stsb_spearman": 0.8090559312844529,
|
| 967 |
+
"step": 14125
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.91,
|
| 971 |
+
"eval_avg_sts": 0.7723370584228957,
|
| 972 |
+
"eval_sickr_spearman": 0.7339858837245768,
|
| 973 |
+
"eval_stsb_spearman": 0.8106882331212146,
|
| 974 |
+
"step": 14250
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.92,
|
| 978 |
+
"eval_avg_sts": 0.7716753925020436,
|
| 979 |
+
"eval_sickr_spearman": 0.7339269975942136,
|
| 980 |
+
"eval_stsb_spearman": 0.8094237874098735,
|
| 981 |
+
"step": 14375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.93,
|
| 985 |
+
"learning_rate": 2.1617816459746575e-06,
|
| 986 |
+
"loss": 1.3738,
|
| 987 |
+
"step": 14500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.93,
|
| 991 |
+
"eval_avg_sts": 0.7714165422777011,
|
| 992 |
+
"eval_sickr_spearman": 0.7332472614466615,
|
| 993 |
+
"eval_stsb_spearman": 0.8095858231087406,
|
| 994 |
+
"step": 14500
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.94,
|
| 998 |
+
"eval_avg_sts": 0.7723213232875585,
|
| 999 |
+
"eval_sickr_spearman": 0.7332828524928272,
|
| 1000 |
+
"eval_stsb_spearman": 0.8113597940822898,
|
| 1001 |
+
"step": 14625
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.94,
|
| 1005 |
+
"eval_avg_sts": 0.7719368310828905,
|
| 1006 |
+
"eval_sickr_spearman": 0.7333558597670131,
|
| 1007 |
+
"eval_stsb_spearman": 0.8105178023987679,
|
| 1008 |
+
"step": 14750
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.95,
|
| 1012 |
+
"eval_avg_sts": 0.7712586994141075,
|
| 1013 |
+
"eval_sickr_spearman": 0.7330949067928999,
|
| 1014 |
+
"eval_stsb_spearman": 0.8094224920353151,
|
| 1015 |
+
"step": 14875
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.96,
|
| 1019 |
+
"learning_rate": 1.2018430820427492e-06,
|
| 1020 |
+
"loss": 1.3737,
|
| 1021 |
+
"step": 15000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 0.96,
|
| 1025 |
+
"eval_avg_sts": 0.7710566218066379,
|
| 1026 |
+
"eval_sickr_spearman": 0.732867767714199,
|
| 1027 |
+
"eval_stsb_spearman": 0.8092454758990769,
|
| 1028 |
+
"step": 15000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.97,
|
| 1032 |
+
"eval_avg_sts": 0.7710765276458553,
|
| 1033 |
+
"eval_sickr_spearman": 0.7330723321752239,
|
| 1034 |
+
"eval_stsb_spearman": 0.8090807231164866,
|
| 1035 |
+
"step": 15125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.98,
|
| 1039 |
+
"eval_avg_sts": 0.771355472096863,
|
| 1040 |
+
"eval_sickr_spearman": 0.7332387119106056,
|
| 1041 |
+
"eval_stsb_spearman": 0.8094722322831204,
|
| 1042 |
+
"step": 15250
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 0.98,
|
| 1046 |
+
"eval_avg_sts": 0.7716261263484865,
|
| 1047 |
+
"eval_sickr_spearman": 0.7334210860027661,
|
| 1048 |
+
"eval_stsb_spearman": 0.8098311666942067,
|
| 1049 |
+
"step": 15375
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 0.99,
|
| 1053 |
+
"learning_rate": 2.4190451811084087e-07,
|
| 1054 |
+
"loss": 1.3738,
|
| 1055 |
+
"step": 15500
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.99,
|
| 1059 |
+
"eval_avg_sts": 0.7715659350580353,
|
| 1060 |
+
"eval_sickr_spearman": 0.7333172123300392,
|
| 1061 |
+
"eval_stsb_spearman": 0.8098146577860313,
|
| 1062 |
+
"step": 15500
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.0,
|
| 1066 |
+
"eval_avg_sts": 0.7714923724726137,
|
| 1067 |
+
"eval_sickr_spearman": 0.7332129192091333,
|
| 1068 |
+
"eval_stsb_spearman": 0.8097718257360942,
|
| 1069 |
+
"step": 15625
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 1.0,
|
| 1073 |
+
"step": 15626,
|
| 1074 |
+
"train_runtime": 8616.861,
|
| 1075 |
+
"train_samples_per_second": 1.813
|
| 1076 |
+
}
|
| 1077 |
+
],
|
| 1078 |
+
"max_steps": 15626,
|
| 1079 |
+
"num_train_epochs": 1,
|
| 1080 |
+
"total_flos": 420412116908851200,
|
| 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 sha256:13b5319475d6c77ecb69affe16ab4908878509f565a335f0703a25e5c9da9dd0
|
| 3 |
+
size 2095
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|