Commit ·
9d88a96
1
Parent(s): 5abeb01
upload
Browse files- model/.DS_Store +0 -0
- model/my-unsup-simcse-bert-base-uncased-0413/config.json +24 -0
- model/my-unsup-simcse-bert-base-uncased-0413/optimizer.pt +3 -0
- model/my-unsup-simcse-bert-base-uncased-0413/pytorch_model.bin +3 -0
- model/my-unsup-simcse-bert-base-uncased-0413/scheduler.pt +3 -0
- model/my-unsup-simcse-bert-base-uncased-0413/special_tokens_map.json +1 -0
- model/my-unsup-simcse-bert-base-uncased-0413/tokenizer_config.json +1 -0
- model/my-unsup-simcse-bert-base-uncased-0413/trainer_state.json +741 -0
- model/my-unsup-simcse-bert-base-uncased-0413/training_args.bin +3 -0
- model/my-unsup-simcse-bert-base-uncased-0413/vocab.txt +0 -0
- model/my-unsup-simcse-bert-base-uncased_ori/config.json +26 -0
- model/my-unsup-simcse-bert-base-uncased_ori/merges.txt +0 -0
- model/my-unsup-simcse-bert-base-uncased_ori/optimizer.pt +3 -0
- model/my-unsup-simcse-bert-base-uncased_ori/pytorch_model.bin +3 -0
- model/my-unsup-simcse-bert-base-uncased_ori/scheduler.pt +3 -0
- model/my-unsup-simcse-bert-base-uncased_ori/special_tokens_map.json +1 -0
- model/my-unsup-simcse-bert-base-uncased_ori/tokenizer_config.json +1 -0
- model/my-unsup-simcse-bert-base-uncased_ori/train_results.txt +3 -0
- model/my-unsup-simcse-bert-base-uncased_ori/trainer_state.json +44 -0
- model/my-unsup-simcse-bert-base-uncased_ori/training_args.bin +3 -0
- model/my-unsup-simcse-bert-base-uncased_ori/vocab.json +0 -0
- model/my-unsup-simcse-bert-base-uncased_ori/vocab.txt +0 -0
model/.DS_Store
ADDED
|
Binary file (8.2 kB). View file
|
|
|
model/my-unsup-simcse-bert-base-uncased-0413/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "bert-base-uncased",
|
| 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": "4.2.2",
|
| 21 |
+
"type_vocab_size": 2,
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 30522
|
| 24 |
+
}
|
model/my-unsup-simcse-bert-base-uncased-0413/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7fe0a42b6c6994f10fd145e4cc4141592b3d6f0ad929d39c3c44f9f7c6e8c5d1
|
| 3 |
+
size 875974661
|
model/my-unsup-simcse-bert-base-uncased-0413/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:97cea0b156dd8b46dcdda7a4b9b9be70ca097d43b955fcc4770c6e00b3ec4c18
|
| 3 |
+
size 438014253
|
model/my-unsup-simcse-bert-base-uncased-0413/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2ede01415a15a84a2fcc3aa958983b10cddb77be0e10206cc7e73b7b95df800
|
| 3 |
+
size 627
|
model/my-unsup-simcse-bert-base-uncased-0413/special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
|
model/my-unsup-simcse-bert-base-uncased-0413/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"}
|
model/my-unsup-simcse-bert-base-uncased-0413/trainer_state.json
ADDED
|
@@ -0,0 +1,741 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.5772300834060221,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-bert-base-uncased-0413",
|
| 4 |
+
"epoch": 1.1541953054078589,
|
| 5 |
+
"global_step": 8900,
|
| 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": 0.43413417287570444,
|
| 13 |
+
"eval_sickr_spearman": 0.4582097432098862,
|
| 14 |
+
"eval_stsb_spearman": 0.4100586025415227,
|
| 15 |
+
"step": 100
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.03,
|
| 19 |
+
"eval_avg_sts": 0.44202934238660074,
|
| 20 |
+
"eval_sickr_spearman": 0.46371670111417157,
|
| 21 |
+
"eval_stsb_spearman": 0.42034198365902986,
|
| 22 |
+
"step": 200
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.04,
|
| 26 |
+
"eval_avg_sts": 0.45392676705884744,
|
| 27 |
+
"eval_sickr_spearman": 0.4647000418539141,
|
| 28 |
+
"eval_stsb_spearman": 0.44315349226378076,
|
| 29 |
+
"step": 300
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.05,
|
| 33 |
+
"eval_avg_sts": 0.46705471818431266,
|
| 34 |
+
"eval_sickr_spearman": 0.4777946649767838,
|
| 35 |
+
"eval_stsb_spearman": 0.4563147713918415,
|
| 36 |
+
"step": 400
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 0.06,
|
| 40 |
+
"learning_rate": 9.783858557039727e-06,
|
| 41 |
+
"loss": 0.0626,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.06,
|
| 46 |
+
"eval_avg_sts": 0.4629420855704067,
|
| 47 |
+
"eval_sickr_spearman": 0.472255862452245,
|
| 48 |
+
"eval_stsb_spearman": 0.4536283086885684,
|
| 49 |
+
"step": 500
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.08,
|
| 53 |
+
"eval_avg_sts": 0.4583178800945026,
|
| 54 |
+
"eval_sickr_spearman": 0.4685936350608932,
|
| 55 |
+
"eval_stsb_spearman": 0.4480421251281119,
|
| 56 |
+
"step": 600
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.09,
|
| 60 |
+
"eval_avg_sts": 0.46463274114411685,
|
| 61 |
+
"eval_sickr_spearman": 0.4777823209837142,
|
| 62 |
+
"eval_stsb_spearman": 0.4514831613045195,
|
| 63 |
+
"step": 700
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.1,
|
| 67 |
+
"eval_avg_sts": 0.4757360446446461,
|
| 68 |
+
"eval_sickr_spearman": 0.4900823655954934,
|
| 69 |
+
"eval_stsb_spearman": 0.46138972369379877,
|
| 70 |
+
"step": 800
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.12,
|
| 74 |
+
"eval_avg_sts": 0.4706008129820051,
|
| 75 |
+
"eval_sickr_spearman": 0.4842226672822465,
|
| 76 |
+
"eval_stsb_spearman": 0.45697895868176375,
|
| 77 |
+
"step": 900
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.13,
|
| 81 |
+
"learning_rate": 9.567717114079454e-06,
|
| 82 |
+
"loss": 0.006,
|
| 83 |
+
"step": 1000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.13,
|
| 87 |
+
"eval_avg_sts": 0.4724716705172539,
|
| 88 |
+
"eval_sickr_spearman": 0.4847777627215667,
|
| 89 |
+
"eval_stsb_spearman": 0.4601655783129412,
|
| 90 |
+
"step": 1000
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.14,
|
| 94 |
+
"eval_avg_sts": 0.46868978609741097,
|
| 95 |
+
"eval_sickr_spearman": 0.4790444342362035,
|
| 96 |
+
"eval_stsb_spearman": 0.4583351379586184,
|
| 97 |
+
"step": 1100
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.16,
|
| 101 |
+
"eval_avg_sts": 0.4739309502400343,
|
| 102 |
+
"eval_sickr_spearman": 0.4857525539252532,
|
| 103 |
+
"eval_stsb_spearman": 0.46210934655481545,
|
| 104 |
+
"step": 1200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.17,
|
| 108 |
+
"eval_avg_sts": 0.47321307094316606,
|
| 109 |
+
"eval_sickr_spearman": 0.4839474971021075,
|
| 110 |
+
"eval_stsb_spearman": 0.4624786447842246,
|
| 111 |
+
"step": 1300
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.18,
|
| 115 |
+
"eval_avg_sts": 0.476232403795022,
|
| 116 |
+
"eval_sickr_spearman": 0.4853778633029345,
|
| 117 |
+
"eval_stsb_spearman": 0.46708694428710956,
|
| 118 |
+
"step": 1400
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.19,
|
| 122 |
+
"learning_rate": 9.35157567111918e-06,
|
| 123 |
+
"loss": 0.0048,
|
| 124 |
+
"step": 1500
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.19,
|
| 128 |
+
"eval_avg_sts": 0.4671866652183253,
|
| 129 |
+
"eval_sickr_spearman": 0.47849875289276517,
|
| 130 |
+
"eval_stsb_spearman": 0.45587457754388544,
|
| 131 |
+
"step": 1500
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.21,
|
| 135 |
+
"eval_avg_sts": 0.4719700135797812,
|
| 136 |
+
"eval_sickr_spearman": 0.48377895596716125,
|
| 137 |
+
"eval_stsb_spearman": 0.46016107119240107,
|
| 138 |
+
"step": 1600
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.22,
|
| 142 |
+
"eval_avg_sts": 0.468405693803544,
|
| 143 |
+
"eval_sickr_spearman": 0.47872507544274157,
|
| 144 |
+
"eval_stsb_spearman": 0.45808631216434653,
|
| 145 |
+
"step": 1700
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.23,
|
| 149 |
+
"eval_avg_sts": 0.4693147295623183,
|
| 150 |
+
"eval_sickr_spearman": 0.4841435120270765,
|
| 151 |
+
"eval_stsb_spearman": 0.4544859470975601,
|
| 152 |
+
"step": 1800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.25,
|
| 156 |
+
"eval_avg_sts": 0.4756334603907184,
|
| 157 |
+
"eval_sickr_spearman": 0.4869564054516981,
|
| 158 |
+
"eval_stsb_spearman": 0.46431051532973877,
|
| 159 |
+
"step": 1900
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.26,
|
| 163 |
+
"learning_rate": 9.135434228158908e-06,
|
| 164 |
+
"loss": 0.0039,
|
| 165 |
+
"step": 2000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.26,
|
| 169 |
+
"eval_avg_sts": 0.47879461277752494,
|
| 170 |
+
"eval_sickr_spearman": 0.4944859050064409,
|
| 171 |
+
"eval_stsb_spearman": 0.463103320548609,
|
| 172 |
+
"step": 2000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.27,
|
| 176 |
+
"eval_avg_sts": 0.4919587799046121,
|
| 177 |
+
"eval_sickr_spearman": 0.50618844271633,
|
| 178 |
+
"eval_stsb_spearman": 0.47772911709289423,
|
| 179 |
+
"step": 2100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.29,
|
| 183 |
+
"eval_avg_sts": 0.4920605376671945,
|
| 184 |
+
"eval_sickr_spearman": 0.5040317021684543,
|
| 185 |
+
"eval_stsb_spearman": 0.48008937316593464,
|
| 186 |
+
"step": 2200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.3,
|
| 190 |
+
"eval_avg_sts": 0.5061405516094162,
|
| 191 |
+
"eval_sickr_spearman": 0.515380875065063,
|
| 192 |
+
"eval_stsb_spearman": 0.4969002281537693,
|
| 193 |
+
"step": 2300
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.31,
|
| 197 |
+
"eval_avg_sts": 0.49937347650708863,
|
| 198 |
+
"eval_sickr_spearman": 0.5125618336594574,
|
| 199 |
+
"eval_stsb_spearman": 0.48618511935471986,
|
| 200 |
+
"step": 2400
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.32,
|
| 204 |
+
"learning_rate": 8.919292785198635e-06,
|
| 205 |
+
"loss": 0.0036,
|
| 206 |
+
"step": 2500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.32,
|
| 210 |
+
"eval_avg_sts": 0.5097962418074036,
|
| 211 |
+
"eval_sickr_spearman": 0.5162478844771226,
|
| 212 |
+
"eval_stsb_spearman": 0.5033445991376845,
|
| 213 |
+
"step": 2500
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.34,
|
| 217 |
+
"eval_avg_sts": 0.5038287826668915,
|
| 218 |
+
"eval_sickr_spearman": 0.5139890778386909,
|
| 219 |
+
"eval_stsb_spearman": 0.4936684874950922,
|
| 220 |
+
"step": 2600
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.35,
|
| 224 |
+
"eval_avg_sts": 0.4876245229950539,
|
| 225 |
+
"eval_sickr_spearman": 0.503567481573055,
|
| 226 |
+
"eval_stsb_spearman": 0.4716815644170527,
|
| 227 |
+
"step": 2700
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.36,
|
| 231 |
+
"eval_avg_sts": 0.48852558904470145,
|
| 232 |
+
"eval_sickr_spearman": 0.5041236336966068,
|
| 233 |
+
"eval_stsb_spearman": 0.472927544392796,
|
| 234 |
+
"step": 2800
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.38,
|
| 238 |
+
"eval_avg_sts": 0.4908887011823253,
|
| 239 |
+
"eval_sickr_spearman": 0.506046366718276,
|
| 240 |
+
"eval_stsb_spearman": 0.4757310356463747,
|
| 241 |
+
"step": 2900
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.39,
|
| 245 |
+
"learning_rate": 8.703151342238363e-06,
|
| 246 |
+
"loss": 0.0031,
|
| 247 |
+
"step": 3000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.39,
|
| 251 |
+
"eval_avg_sts": 0.49885341513488746,
|
| 252 |
+
"eval_sickr_spearman": 0.5072558859146905,
|
| 253 |
+
"eval_stsb_spearman": 0.49045094435508446,
|
| 254 |
+
"step": 3000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.4,
|
| 258 |
+
"eval_avg_sts": 0.5025732287061541,
|
| 259 |
+
"eval_sickr_spearman": 0.5157659404052926,
|
| 260 |
+
"eval_stsb_spearman": 0.4893805170070156,
|
| 261 |
+
"step": 3100
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.41,
|
| 265 |
+
"eval_avg_sts": 0.48566033454734003,
|
| 266 |
+
"eval_sickr_spearman": 0.5039478878964448,
|
| 267 |
+
"eval_stsb_spearman": 0.4673727811982352,
|
| 268 |
+
"step": 3200
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.43,
|
| 272 |
+
"eval_avg_sts": 0.47697364115308033,
|
| 273 |
+
"eval_sickr_spearman": 0.497832856247953,
|
| 274 |
+
"eval_stsb_spearman": 0.4561144260582077,
|
| 275 |
+
"step": 3300
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.44,
|
| 279 |
+
"eval_avg_sts": 0.4905299355880675,
|
| 280 |
+
"eval_sickr_spearman": 0.5084222731820136,
|
| 281 |
+
"eval_stsb_spearman": 0.4726375979941214,
|
| 282 |
+
"step": 3400
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.45,
|
| 286 |
+
"learning_rate": 8.487009899278088e-06,
|
| 287 |
+
"loss": 0.0026,
|
| 288 |
+
"step": 3500
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.45,
|
| 292 |
+
"eval_avg_sts": 0.4927153870008586,
|
| 293 |
+
"eval_sickr_spearman": 0.5115589442614296,
|
| 294 |
+
"eval_stsb_spearman": 0.4738718297402876,
|
| 295 |
+
"step": 3500
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.47,
|
| 299 |
+
"eval_avg_sts": 0.4990059371177902,
|
| 300 |
+
"eval_sickr_spearman": 0.5237218120475015,
|
| 301 |
+
"eval_stsb_spearman": 0.47429006218807884,
|
| 302 |
+
"step": 3600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.48,
|
| 306 |
+
"eval_avg_sts": 0.5081240005291581,
|
| 307 |
+
"eval_sickr_spearman": 0.5338321187447169,
|
| 308 |
+
"eval_stsb_spearman": 0.4824158823135992,
|
| 309 |
+
"step": 3700
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.49,
|
| 313 |
+
"eval_avg_sts": 0.5053613131195629,
|
| 314 |
+
"eval_sickr_spearman": 0.5345850062597594,
|
| 315 |
+
"eval_stsb_spearman": 0.4761376199793665,
|
| 316 |
+
"step": 3800
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.51,
|
| 320 |
+
"eval_avg_sts": 0.5291853264881973,
|
| 321 |
+
"eval_sickr_spearman": 0.5583114097481583,
|
| 322 |
+
"eval_stsb_spearman": 0.5000592432282362,
|
| 323 |
+
"step": 3900
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.52,
|
| 327 |
+
"learning_rate": 8.270868456317816e-06,
|
| 328 |
+
"loss": 0.0043,
|
| 329 |
+
"step": 4000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.52,
|
| 333 |
+
"eval_avg_sts": 0.5303577078371549,
|
| 334 |
+
"eval_sickr_spearman": 0.555212106865661,
|
| 335 |
+
"eval_stsb_spearman": 0.5055033088086488,
|
| 336 |
+
"step": 4000
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.53,
|
| 340 |
+
"eval_avg_sts": 0.5333661650982712,
|
| 341 |
+
"eval_sickr_spearman": 0.5605562393360713,
|
| 342 |
+
"eval_stsb_spearman": 0.506176090860471,
|
| 343 |
+
"step": 4100
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.54,
|
| 347 |
+
"eval_avg_sts": 0.5278812963738058,
|
| 348 |
+
"eval_sickr_spearman": 0.5576400310122562,
|
| 349 |
+
"eval_stsb_spearman": 0.49812256173535524,
|
| 350 |
+
"step": 4200
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.56,
|
| 354 |
+
"eval_avg_sts": 0.5228114162776825,
|
| 355 |
+
"eval_sickr_spearman": 0.549996457591598,
|
| 356 |
+
"eval_stsb_spearman": 0.495626374963767,
|
| 357 |
+
"step": 4300
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.57,
|
| 361 |
+
"eval_avg_sts": 0.5249527931181994,
|
| 362 |
+
"eval_sickr_spearman": 0.5511832580770323,
|
| 363 |
+
"eval_stsb_spearman": 0.4987223281593667,
|
| 364 |
+
"step": 4400
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.58,
|
| 368 |
+
"learning_rate": 8.054727013357542e-06,
|
| 369 |
+
"loss": 0.0024,
|
| 370 |
+
"step": 4500
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.58,
|
| 374 |
+
"eval_avg_sts": 0.5365941902025078,
|
| 375 |
+
"eval_sickr_spearman": 0.5552903014988024,
|
| 376 |
+
"eval_stsb_spearman": 0.5178980789062131,
|
| 377 |
+
"step": 4500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.6,
|
| 381 |
+
"eval_avg_sts": 0.5347280260801861,
|
| 382 |
+
"eval_sickr_spearman": 0.5567227330369913,
|
| 383 |
+
"eval_stsb_spearman": 0.5127333191233809,
|
| 384 |
+
"step": 4600
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.61,
|
| 388 |
+
"eval_avg_sts": 0.5353743763935372,
|
| 389 |
+
"eval_sickr_spearman": 0.5572790772849489,
|
| 390 |
+
"eval_stsb_spearman": 0.5134696755021255,
|
| 391 |
+
"step": 4700
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.62,
|
| 395 |
+
"eval_avg_sts": 0.5372262235055488,
|
| 396 |
+
"eval_sickr_spearman": 0.557209576281168,
|
| 397 |
+
"eval_stsb_spearman": 0.5172428707299297,
|
| 398 |
+
"step": 4800
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.64,
|
| 402 |
+
"eval_avg_sts": 0.5392746843345577,
|
| 403 |
+
"eval_sickr_spearman": 0.5658935994211787,
|
| 404 |
+
"eval_stsb_spearman": 0.5126557692479369,
|
| 405 |
+
"step": 4900
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.65,
|
| 409 |
+
"learning_rate": 7.83858557039727e-06,
|
| 410 |
+
"loss": 0.0035,
|
| 411 |
+
"step": 5000
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.65,
|
| 415 |
+
"eval_avg_sts": 0.5381467627451588,
|
| 416 |
+
"eval_sickr_spearman": 0.5665912992006686,
|
| 417 |
+
"eval_stsb_spearman": 0.5097022262896491,
|
| 418 |
+
"step": 5000
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.66,
|
| 422 |
+
"eval_avg_sts": 0.5437598393325787,
|
| 423 |
+
"eval_sickr_spearman": 0.5716501269285366,
|
| 424 |
+
"eval_stsb_spearman": 0.5158695517366209,
|
| 425 |
+
"step": 5100
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.67,
|
| 429 |
+
"eval_avg_sts": 0.5458859184475997,
|
| 430 |
+
"eval_sickr_spearman": 0.5727973978174873,
|
| 431 |
+
"eval_stsb_spearman": 0.5189744390777121,
|
| 432 |
+
"step": 5200
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.69,
|
| 436 |
+
"eval_avg_sts": 0.5471436350834367,
|
| 437 |
+
"eval_sickr_spearman": 0.5750997686649232,
|
| 438 |
+
"eval_stsb_spearman": 0.5191875015019501,
|
| 439 |
+
"step": 5300
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.7,
|
| 443 |
+
"eval_avg_sts": 0.5362576150135143,
|
| 444 |
+
"eval_sickr_spearman": 0.5684171054596358,
|
| 445 |
+
"eval_stsb_spearman": 0.504098124567393,
|
| 446 |
+
"step": 5400
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.71,
|
| 450 |
+
"learning_rate": 7.622444127436995e-06,
|
| 451 |
+
"loss": 0.0021,
|
| 452 |
+
"step": 5500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.71,
|
| 456 |
+
"eval_avg_sts": 0.5439797124533817,
|
| 457 |
+
"eval_sickr_spearman": 0.5719825982126912,
|
| 458 |
+
"eval_stsb_spearman": 0.5159768266940723,
|
| 459 |
+
"step": 5500
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.73,
|
| 463 |
+
"eval_avg_sts": 0.5336924618718794,
|
| 464 |
+
"eval_sickr_spearman": 0.5533836108650136,
|
| 465 |
+
"eval_stsb_spearman": 0.5140013128787451,
|
| 466 |
+
"step": 5600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.74,
|
| 470 |
+
"eval_avg_sts": 0.5432756107135348,
|
| 471 |
+
"eval_sickr_spearman": 0.5664231903456354,
|
| 472 |
+
"eval_stsb_spearman": 0.5201280310814343,
|
| 473 |
+
"step": 5700
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.75,
|
| 477 |
+
"eval_avg_sts": 0.5362600896652556,
|
| 478 |
+
"eval_sickr_spearman": 0.5664119990990002,
|
| 479 |
+
"eval_stsb_spearman": 0.5061081802315109,
|
| 480 |
+
"step": 5800
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.77,
|
| 484 |
+
"eval_avg_sts": 0.5558055193441328,
|
| 485 |
+
"eval_sickr_spearman": 0.5845013764604193,
|
| 486 |
+
"eval_stsb_spearman": 0.5271096622278463,
|
| 487 |
+
"step": 5900
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.78,
|
| 491 |
+
"learning_rate": 7.406302684476722e-06,
|
| 492 |
+
"loss": 0.0029,
|
| 493 |
+
"step": 6000
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.78,
|
| 497 |
+
"eval_avg_sts": 0.5543715317449052,
|
| 498 |
+
"eval_sickr_spearman": 0.582216825151617,
|
| 499 |
+
"eval_stsb_spearman": 0.5265262383381932,
|
| 500 |
+
"step": 6000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.79,
|
| 504 |
+
"eval_avg_sts": 0.5577833334850189,
|
| 505 |
+
"eval_sickr_spearman": 0.5861351063447354,
|
| 506 |
+
"eval_stsb_spearman": 0.5294315606253022,
|
| 507 |
+
"step": 6100
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.8,
|
| 511 |
+
"eval_avg_sts": 0.5459303241614681,
|
| 512 |
+
"eval_sickr_spearman": 0.5734108990761563,
|
| 513 |
+
"eval_stsb_spearman": 0.5184497492467801,
|
| 514 |
+
"step": 6200
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.82,
|
| 518 |
+
"eval_avg_sts": 0.5527974603364145,
|
| 519 |
+
"eval_sickr_spearman": 0.5769931546836136,
|
| 520 |
+
"eval_stsb_spearman": 0.5286017659892155,
|
| 521 |
+
"step": 6300
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.83,
|
| 525 |
+
"eval_avg_sts": 0.5613325719571378,
|
| 526 |
+
"eval_sickr_spearman": 0.5828151485822315,
|
| 527 |
+
"eval_stsb_spearman": 0.5398499953320443,
|
| 528 |
+
"step": 6400
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.84,
|
| 532 |
+
"learning_rate": 7.190161241516449e-06,
|
| 533 |
+
"loss": 0.0013,
|
| 534 |
+
"step": 6500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.84,
|
| 538 |
+
"eval_avg_sts": 0.5706535000902035,
|
| 539 |
+
"eval_sickr_spearman": 0.5902679904990792,
|
| 540 |
+
"eval_stsb_spearman": 0.5510390096813279,
|
| 541 |
+
"step": 6500
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.86,
|
| 545 |
+
"eval_avg_sts": 0.5588971829128009,
|
| 546 |
+
"eval_sickr_spearman": 0.5792282339579348,
|
| 547 |
+
"eval_stsb_spearman": 0.5385661318676669,
|
| 548 |
+
"step": 6600
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.87,
|
| 552 |
+
"eval_avg_sts": 0.5467875739347965,
|
| 553 |
+
"eval_sickr_spearman": 0.5713499805756498,
|
| 554 |
+
"eval_stsb_spearman": 0.5222251672939431,
|
| 555 |
+
"step": 6700
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.88,
|
| 559 |
+
"eval_avg_sts": 0.5564259742453666,
|
| 560 |
+
"eval_sickr_spearman": 0.5743576881477043,
|
| 561 |
+
"eval_stsb_spearman": 0.5384942603430287,
|
| 562 |
+
"step": 6800
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.89,
|
| 566 |
+
"eval_avg_sts": 0.564484253954618,
|
| 567 |
+
"eval_sickr_spearman": 0.5855938918937308,
|
| 568 |
+
"eval_stsb_spearman": 0.5433746160155053,
|
| 569 |
+
"step": 6900
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.91,
|
| 573 |
+
"learning_rate": 6.974019798556175e-06,
|
| 574 |
+
"loss": 0.0046,
|
| 575 |
+
"step": 7000
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.91,
|
| 579 |
+
"eval_avg_sts": 0.5572693279756483,
|
| 580 |
+
"eval_sickr_spearman": 0.5816440542669675,
|
| 581 |
+
"eval_stsb_spearman": 0.5328946016843291,
|
| 582 |
+
"step": 7000
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.92,
|
| 586 |
+
"eval_avg_sts": 0.5564053626625001,
|
| 587 |
+
"eval_sickr_spearman": 0.5780550262742075,
|
| 588 |
+
"eval_stsb_spearman": 0.5347556990507928,
|
| 589 |
+
"step": 7100
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.93,
|
| 593 |
+
"eval_avg_sts": 0.5595700328398423,
|
| 594 |
+
"eval_sickr_spearman": 0.5775588169652502,
|
| 595 |
+
"eval_stsb_spearman": 0.5415812487144344,
|
| 596 |
+
"step": 7200
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.95,
|
| 600 |
+
"eval_avg_sts": 0.5507423483095826,
|
| 601 |
+
"eval_sickr_spearman": 0.5766743242322674,
|
| 602 |
+
"eval_stsb_spearman": 0.5248103723868978,
|
| 603 |
+
"step": 7300
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.96,
|
| 607 |
+
"eval_avg_sts": 0.5496315444386322,
|
| 608 |
+
"eval_sickr_spearman": 0.5782933085684419,
|
| 609 |
+
"eval_stsb_spearman": 0.5209697803088224,
|
| 610 |
+
"step": 7400
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.97,
|
| 614 |
+
"learning_rate": 6.7578783555959026e-06,
|
| 615 |
+
"loss": 0.0029,
|
| 616 |
+
"step": 7500
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.97,
|
| 620 |
+
"eval_avg_sts": 0.5654071055800436,
|
| 621 |
+
"eval_sickr_spearman": 0.5835302836515436,
|
| 622 |
+
"eval_stsb_spearman": 0.5472839275085436,
|
| 623 |
+
"step": 7500
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 0.99,
|
| 627 |
+
"eval_avg_sts": 0.5573712317131654,
|
| 628 |
+
"eval_sickr_spearman": 0.5770115986265659,
|
| 629 |
+
"eval_stsb_spearman": 0.5377308647997648,
|
| 630 |
+
"step": 7600
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.0,
|
| 634 |
+
"eval_avg_sts": 0.561124258264991,
|
| 635 |
+
"eval_sickr_spearman": 0.5816098080916421,
|
| 636 |
+
"eval_stsb_spearman": 0.5406387084383399,
|
| 637 |
+
"step": 7700
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.01,
|
| 641 |
+
"eval_avg_sts": 0.5704081137772914,
|
| 642 |
+
"eval_sickr_spearman": 0.592072951260022,
|
| 643 |
+
"eval_stsb_spearman": 0.5487432762945608,
|
| 644 |
+
"step": 7800
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.02,
|
| 648 |
+
"eval_avg_sts": 0.5742201136812924,
|
| 649 |
+
"eval_sickr_spearman": 0.5966417656910182,
|
| 650 |
+
"eval_stsb_spearman": 0.5517984616715667,
|
| 651 |
+
"step": 7900
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.04,
|
| 655 |
+
"learning_rate": 6.541736912635629e-06,
|
| 656 |
+
"loss": 0.0041,
|
| 657 |
+
"step": 8000
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.04,
|
| 661 |
+
"eval_avg_sts": 0.578527273351308,
|
| 662 |
+
"eval_sickr_spearman": 0.6026895057663952,
|
| 663 |
+
"eval_stsb_spearman": 0.5543650409362207,
|
| 664 |
+
"step": 8000
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.05,
|
| 668 |
+
"eval_avg_sts": 0.5874722846001152,
|
| 669 |
+
"eval_sickr_spearman": 0.6127884771236713,
|
| 670 |
+
"eval_stsb_spearman": 0.5621560920765593,
|
| 671 |
+
"step": 8100
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.06,
|
| 675 |
+
"eval_avg_sts": 0.5832344728360276,
|
| 676 |
+
"eval_sickr_spearman": 0.6076356044991877,
|
| 677 |
+
"eval_stsb_spearman": 0.5588333411728674,
|
| 678 |
+
"step": 8200
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.08,
|
| 682 |
+
"eval_avg_sts": 0.5855265282675808,
|
| 683 |
+
"eval_sickr_spearman": 0.6116178150883202,
|
| 684 |
+
"eval_stsb_spearman": 0.5594352414468413,
|
| 685 |
+
"step": 8300
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.09,
|
| 689 |
+
"eval_avg_sts": 0.5773281021616433,
|
| 690 |
+
"eval_sickr_spearman": 0.6021348906380893,
|
| 691 |
+
"eval_stsb_spearman": 0.5525213136851973,
|
| 692 |
+
"step": 8400
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.1,
|
| 696 |
+
"learning_rate": 6.325595469675356e-06,
|
| 697 |
+
"loss": 0.0034,
|
| 698 |
+
"step": 8500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 1.1,
|
| 702 |
+
"eval_avg_sts": 0.5733092691598134,
|
| 703 |
+
"eval_sickr_spearman": 0.598557342078573,
|
| 704 |
+
"eval_stsb_spearman": 0.5480611962410536,
|
| 705 |
+
"step": 8500
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 1.12,
|
| 709 |
+
"eval_avg_sts": 0.5735029071550635,
|
| 710 |
+
"eval_sickr_spearman": 0.5968002683257639,
|
| 711 |
+
"eval_stsb_spearman": 0.5502055459843631,
|
| 712 |
+
"step": 8600
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 1.13,
|
| 716 |
+
"eval_avg_sts": 0.5851313052653937,
|
| 717 |
+
"eval_sickr_spearman": 0.6086310971075941,
|
| 718 |
+
"eval_stsb_spearman": 0.5616315134231933,
|
| 719 |
+
"step": 8700
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 1.14,
|
| 723 |
+
"eval_avg_sts": 0.5914101664574409,
|
| 724 |
+
"eval_sickr_spearman": 0.6088285529656062,
|
| 725 |
+
"eval_stsb_spearman": 0.5739917799492756,
|
| 726 |
+
"step": 8800
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 1.15,
|
| 730 |
+
"eval_avg_sts": 0.5948432200811302,
|
| 731 |
+
"eval_sickr_spearman": 0.6124563567562383,
|
| 732 |
+
"eval_stsb_spearman": 0.5772300834060221,
|
| 733 |
+
"step": 8900
|
| 734 |
+
}
|
| 735 |
+
],
|
| 736 |
+
"max_steps": 23133,
|
| 737 |
+
"num_train_epochs": 3,
|
| 738 |
+
"total_flos": 0,
|
| 739 |
+
"trial_name": null,
|
| 740 |
+
"trial_params": null
|
| 741 |
+
}
|
model/my-unsup-simcse-bert-base-uncased-0413/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd222bfb9c26b00bdd9e473783a9c2007e0baf100da40fb19c324977569aa06b
|
| 3 |
+
size 2171
|
model/my-unsup-simcse-bert-base-uncased-0413/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model/my-unsup-simcse-bert-base-uncased_ori/config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "roberta-base",
|
| 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 |
+
}
|
model/my-unsup-simcse-bert-base-uncased_ori/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model/my-unsup-simcse-bert-base-uncased_ori/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d90ba75920b945193134035ade9fb71cf9af547549831752edbe17c72f0b4ff
|
| 3 |
+
size 997281797
|
model/my-unsup-simcse-bert-base-uncased_ori/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29f3ec6385de50439bcc848c54dbeffe904d2e8d9d503c1eb1ad40d49437b75b
|
| 3 |
+
size 498669037
|
model/my-unsup-simcse-bert-base-uncased_ori/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1522b9247a289f4d0d155e0bbe7375547292a9040940f809c7eb3c8cec5b56bf
|
| 3 |
+
size 627
|
model/my-unsup-simcse-bert-base-uncased_ori/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}}
|
model/my-unsup-simcse-bert-base-uncased_ori/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"}
|
model/my-unsup-simcse-bert-base-uncased_ori/train_results.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
epoch = 3.0
|
| 2 |
+
train_runtime = 763.7936
|
| 3 |
+
train_samples_per_second = 1.261
|
model/my-unsup-simcse-bert-base-uncased_ori/trainer_state.json
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7830865408035854,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-bert-base-uncased",
|
| 4 |
+
"epoch": 0.62402496099844,
|
| 5 |
+
"global_step": 400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.16,
|
| 12 |
+
"eval_avg_sts": 0.7354961861232879,
|
| 13 |
+
"eval_sickr_spearman": 0.6918233183223756,
|
| 14 |
+
"eval_stsb_spearman": 0.7791690539242002,
|
| 15 |
+
"step": 100
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.31,
|
| 19 |
+
"eval_avg_sts": 0.7392649185824722,
|
| 20 |
+
"eval_sickr_spearman": 0.7019525973046593,
|
| 21 |
+
"eval_stsb_spearman": 0.7765772398602853,
|
| 22 |
+
"step": 200
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.47,
|
| 26 |
+
"eval_avg_sts": 0.7348363506865718,
|
| 27 |
+
"eval_sickr_spearman": 0.6903618759989183,
|
| 28 |
+
"eval_stsb_spearman": 0.7793108253742252,
|
| 29 |
+
"step": 300
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.62,
|
| 33 |
+
"eval_avg_sts": 0.7360856589231828,
|
| 34 |
+
"eval_sickr_spearman": 0.6890847770427803,
|
| 35 |
+
"eval_stsb_spearman": 0.7830865408035854,
|
| 36 |
+
"step": 400
|
| 37 |
+
}
|
| 38 |
+
],
|
| 39 |
+
"max_steps": 1923,
|
| 40 |
+
"num_train_epochs": 3,
|
| 41 |
+
"total_flos": 0,
|
| 42 |
+
"trial_name": null,
|
| 43 |
+
"trial_params": null
|
| 44 |
+
}
|
model/my-unsup-simcse-bert-base-uncased_ori/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0aa113111f5a68f9866ca380fbcb100457fb0a5f72e4c85cd5385220044e1e55
|
| 3 |
+
size 2107
|
model/my-unsup-simcse-bert-base-uncased_ori/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model/my-unsup-simcse-bert-base-uncased_ori/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|