Update README.md

### Llama3 CoEdit

This is a Llama3 8B model trained using [torchtune](https://pytorch.org/torchtune) on the `grammarly/coedit` dataset.

### Training details

See `8B_lora.yaml` file for all configuration. In order to use the `grammarly/coedit` dataset, I added this line to the config:

```
dataset:
_component_: torchtune.datasets.instruct_dataset
source: grammarly/coedit
template: GrammarErrorCorrectionTemplate
column_map: {"sentence": "src", "output": "tgt"}
train_on_input: False
split: train
```

### Evaluation

README.md

@@ -6,4 +6,6 @@ language:
 - en
 metrics:
 - accuracy
+tags:
+- torchtune
 ---