update model card README.md

README.md

@@ -0,0 +1,61 @@
+---
+tags:
+- generated_from_trainer
+metrics:
+- accuracy
+- f1
+model-index:
+- name: bert-base-chinese-finetuned-amazon_zh_20000
+  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. -->
+
+# bert-base-chinese-finetuned-amazon_zh_20000
+
+This model is a fine-tuned version of [bert-base-chinese](https://huggingface.co/bert-base-chinese) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.1683
+- Accuracy: 0.5224
+- F1: 0.5194
+
+## 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: 2e-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- num_epochs: 2
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1     |
+|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
+| 1.2051        | 1.0   | 2500 | 1.1717          | 0.506    | 0.4847 |
+| 1.0035        | 2.0   | 5000 | 1.1683          | 0.5224   | 0.5194 |
+
+
+### Framework versions
+
+- Transformers 4.15.0
+- Pytorch 1.9.1
+- Datasets 1.18.3
+- Tokenizers 0.10.3