update model card README.md

README.md

@@ -0,0 +1,61 @@
+---
+license: apache-2.0
+tags:
+- generated_from_trainer
+model-index:
+- name: 20split_dataset
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# 20split_dataset
+
+This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.0446
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-05
+- train_batch_size: 64
+- eval_batch_size: 8
+- seed: 42
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- num_epochs: 6
+
+### Training results
+
+| Training Loss | Epoch | Step  | Validation Loss |
+|:-------------:|:-----:|:-----:|:---------------:|
+| 2.5971        | 1.0   | 11851 | 2.3479          |
+| 2.3773        | 2.0   | 23702 | 2.2446          |
+| 2.2663        | 3.0   | 35553 | 2.1630          |
+| 2.1842        | 4.0   | 47404 | 2.1059          |
+| 2.1145        | 5.0   | 59255 | 2.0626          |
+| 2.0652        | 6.0   | 71106 | 2.0446          |
+
+
+### Framework versions
+
+- Transformers 4.20.1
+- Pytorch 1.12.0+cu113
+- Datasets 2.3.2
+- Tokenizers 0.12.1