update model card README.md

README.md

@@ -0,0 +1,57 @@
+---
+license: apache-2.0
+tags:
+- generated_from_trainer
+datasets:
+- glue
+model-index:
+- name: distilbert-sst2
+  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. -->
+
+# distilbert-sst2
+
+This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the glue dataset.
+It achieves the following results on the evaluation set:
+- eval_loss: 0.4182
+- eval_accuracy: 0.8911
+- eval_runtime: 1.8021
+- eval_samples_per_second: 483.882
+- eval_steps_per_second: 60.485
+- epoch: 0.8
+- step: 6700
+
+## 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: 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: 3.0
+
+### Framework versions
+
+- Transformers 4.15.0
+- Pytorch 1.10.1
+- Datasets 1.17.0
+- Tokenizers 0.10.3