add model

README.md

@@ -14,10 +14,10 @@ probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [hfl/chinese-roberta-wwm-ext](https://huggingface.co/hfl/chinese-roberta-wwm-ext) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Train Loss: 0.4546
-- Train Sparse Categorical Accuracy: 0.8132
-- Validation Loss: 0.5708
-- Validation Sparse Categorical Accuracy: 0.7705
+- Train Loss: 0.5776
+- Train Sparse Categorical Accuracy: 0.7278
+- Validation Loss: 0.6460
+- Validation Sparse Categorical Accuracy: 0.6859
 - Epoch: 2
 
 ## Model description
@@ -44,9 +44,9 @@ The following hyperparameters were used during training:
 
 | Train Loss | Train Sparse Categorical Accuracy | Validation Loss | Validation Sparse Categorical Accuracy | Epoch |
 |:----------:|:---------------------------------:|:---------------:|:--------------------------------------:|:-----:|
-| 0.6264     | 0.7340                            | 0.6649          | 0.7134                                 | 0     |
-| 0.5270     | 0.7727                            | 0.5472          | 0.7394                                 | 1     |
-| 0.4546     | 0.8132                            | 0.5708          | 0.7705                                 | 2     |
+| 0.7207     | 0.6394                            | 0.6930          | 0.6811                                 | 0     |
+| 0.6253     | 0.7033                            | 0.6549          | 0.6872                                 | 1     |
+| 0.5776     | 0.7278                            | 0.6460          | 0.6859                                 | 2     |
 
 
 ### Framework versions
@@ -54,4 +54,4 @@ The following hyperparameters were used during training:
 - Transformers 4.16.2
 - TensorFlow 2.8.0
 - Datasets 1.18.3
-- Tokenizers 0.10.3
+- Tokenizers 0.11.0

config.json

@@ -29,7 +29,7 @@
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
   "output_past": true,
-  "pad_token_id": 1,
+  "pad_token_id": 0,
   "pooler_fc_size": 768,
   "pooler_num_attention_heads": 12,
   "pooler_num_fc_layers": 3,

tf_model.h5

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:909e1d480b6e2a00081d0a785f14538f1a8a40ec474d289205eb42b1e48d2d84
+oid sha256:51a2c50cef3d8d2bdee5c94fdcc3334ce30bfc1b922c71811acb6d252a707fd8
 size 409367900