add model

0_Transformer/config.json

@@ -0,0 +1,23 @@
+{
+  "architectures": [
+    "XLMRobertaModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "xlm-roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "output_past": true,
+  "pad_token_id": 1,
+  "type_vocab_size": 1,
+  "vocab_size": 250002
+}

0_Transformer/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7ed7f322684e11c07b88e96f50a49a6cb92b8b5339923d89d2d7eb683dcc5e2
+size 1112259151

0_Transformer/sentence_bert_config.json

@@ -0,0 +1,3 @@
+{
+  "max_seq_length": 128
+}

0_Transformer/sentencepiece.bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
+size 5069051

0_Transformer/special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": "<mask>"}

0_Transformer/tokenizer_config.json

@@ -0,0 +1 @@
+{"model_max_length": 512, "special_tokens_map_file": "/content/gdrive/My Drive/en2fin-sbert/output/knowledge-distillation/XLM-R-distilled-once/0_Transformer/special_tokens_map.json", "full_tokenizer_file": null}

1_Pooling/config.json

@@ -0,0 +1,7 @@
+{
+  "word_embedding_dimension": 768,
+  "pooling_mode_cls_token": false,
+  "pooling_mode_mean_tokens": true,
+  "pooling_mode_max_tokens": false,
+  "pooling_mode_mean_sqrt_len_tokens": false
+}

config.json

@@ -0,0 +1,3 @@
+{
+  "__version__": "0.3.9"
+}

modules.json

@@ -0,0 +1,14 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "0_Transformer",
+    "type": "sentence_transformers.models.Transformer"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_Pooling",
+    "type": "sentence_transformers.models.Pooling"
+  }
+]

mse_evaluation_Dev-MSE-evaluator_results.csv

@@ -0,0 +1,127 @@
+epoch,steps,MSE
+0,500,8.570442348718643
+0,1000,8.574279397726059
+0,1500,8.557721227407455
+0,2000,8.466233313083649
+0,2500,8.497065305709839
+0,3000,8.495021611452103
+0,3500,8.528045564889908
+0,4000,8.526290953159332
+0,4500,8.51152092218399
+0,5000,8.509846776723862
+0,5500,8.489316701889038
+0,6000,8.47911462187767
+0,6500,8.477255702018738
+0,7000,8.49146693944931
+0,7500,8.450501412153244
+0,8000,8.498767763376236
+0,8500,8.525879681110382
+0,9000,8.5193932056427
+0,9500,8.524371683597565
+0,10000,8.49359780550003
+0,10500,8.501896262168884
+0,11000,8.521968871355057
+0,11500,8.512929081916809
+0,12000,8.464228361845016
+0,-1,8.464228361845016
+1,500,8.439267426729202
+1,1000,8.435093611478806
+1,1500,8.464305847883224
+1,2000,8.429614454507828
+1,2500,8.444567769765854
+1,3000,8.424865454435349
+1,3500,8.420820534229279
+1,4000,8.424185961484909
+1,4500,8.416373282670975
+1,5000,8.425454050302505
+1,5500,8.41192752122879
+1,6000,8.403629809617996
+1,6500,8.41481164097786
+1,7000,8.419384062290192
+1,7500,8.419118076562881
+1,8000,8.427265286445618
+1,8500,8.4264375269413
+1,9000,8.400160074234009
+1,9500,8.40001180768013
+1,10000,8.39078426361084
+1,10500,8.387839049100876
+1,11000,8.400828391313553
+1,11500,8.38041678071022
+1,12000,8.375781774520874
+1,-1,8.375781774520874
+2,500,8.397626876831055
+2,1000,8.38230550289154
+2,1500,8.391416817903519
+2,2000,8.396975696086884
+2,2500,8.38283821940422
+2,3000,8.376853168010712
+2,3500,8.388946205377579
+2,4000,8.390841633081436
+2,4500,8.380627632141113
+2,5000,8.38095024228096
+2,5500,8.381462842226028
+2,6000,8.403825014829636
+2,6500,8.39417278766632
+2,7000,8.38434174656868
+2,7500,8.365185558795929
+2,8000,8.37533250451088
+2,8500,8.382804691791534
+2,9000,8.384855091571808
+2,9500,8.382391184568405
+2,10000,8.367933332920074
+2,10500,8.37000235915184
+2,11000,8.375443518161774
+2,11500,8.379511535167694
+2,12000,8.351945132017136
+2,-1,8.351945132017136
+3,500,8.353012055158615
+3,1000,8.358874171972275
+3,1500,8.356638997793198
+3,2000,8.383019268512726
+3,2500,8.342177420854568
+3,3000,8.356159180402756
+3,3500,8.349952101707458
+3,4000,8.343320339918137
+3,4500,8.349838852882385
+3,5000,8.35258811712265
+3,5500,8.359123021364212
+3,6000,8.366020023822784
+3,6500,8.366061002016068
+3,7000,8.35094228386879
+3,7500,8.335867524147034
+3,8000,8.344996720552444
+3,8500,8.35791900753975
+3,9000,8.349382132291794
+3,9500,8.332888036966324
+3,10000,8.34961012005806
+3,10500,8.342666923999786
+3,11000,8.32587331533432
+3,11500,8.321724086999893
+3,12000,8.316271752119064
+3,-1,8.316271752119064
+4,500,8.327758312225342
+4,1000,8.344313502311707
+4,1500,8.35057869553566
+4,2000,8.335026353597641
+4,2500,8.343841135501862
+4,3000,8.330678194761276
+4,3500,8.342430740594864
+4,4000,8.35268348455429
+4,4500,8.346807211637497
+4,5000,8.33790972828865
+4,5500,8.340930938720703
+4,6000,8.313610404729843
+4,6500,8.340232074260712
+4,7000,8.326788246631622
+4,7500,8.33984762430191
+4,8000,8.319726586341858
+4,8500,8.330386132001877
+4,9000,8.329372107982635
+4,9500,8.320283889770508
+4,10000,8.302289992570877
+4,10500,8.327770233154297
+4,11000,8.320759981870651
+4,11500,8.321399986743927
+4,12000,8.308279514312744
+4,-1,8.308279514312744
+5,500,8.304926007986069