From 6b63048562f5be0b638d31b3b8997350d205697a Mon Sep 17 00:00:00 2001
From: Matthew Hayes <matthayes@users.noreply.huggingface.co>
Date: Wed, 12 Apr 2023 09:08:16 +0000
Subject: [PATCH] Update README.md

Add pipeline usage instructions
---
 README.md | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/README.md b/README.md
index f843215..6d82350 100644
--- a/README.md
+++ b/README.md
@@ -24,6 +24,32 @@ high quality instruction following behavior not characteristic of the foundation
 [EleutherAI’s](https://www.eleuther.ai/) [Pythia-12b](https://huggingface.co/EleutherAI/pythia-12b) and fine-tuned 
 on a ~15K record instruction corpus generated by Databricks employees and released under a permissive license (CC-BY-SA)
 
+## Usage
+
+To use the model with the `transformers` library on a machine with GPUs:
+
+```
+from transformers import pipeline
+
+instruct_pipeline = pipeline(model="databricks/dolly-v2-12b", trust_remote_code=True, device_map="auto")
+```
+
+You can then use the pipeline to answer instructions:
+
+```
+instruct_pipeline("Explain to me the difference between nuclear fission and fusion.")
+```
+
+To reduce memory usage you can load the model with `bfloat16`:
+
+```
+import torch
+from transformers import pipeline
+
+instruct_pipeline = pipeline(model="databricks/dolly-v2-12b", torch_dtype=torch.bfloat16, trust_remote_code=True, device_map="auto")
+```
+
+
 ## Known Limitations
 
 ### Performance Limitations