From 712a887dec070c9a32c7b3d184275528ea4b1e03 Mon Sep 17 00:00:00 2001
From: guschmue <guschmue@microsoft.com>
Date: Thu, 16 May 2024 10:25:29 -0700
Subject: [PATCH] add olive instructions to readme

---
 js/chat/README.md | 15 +++++++++------
 js/chat/main.js   |  5 -----
 2 files changed, 9 insertions(+), 11 deletions(-)

diff --git a/js/chat/README.md b/js/chat/README.md
index ff4d1dd6..0b9e2d7f 100644
--- a/js/chat/README.md
+++ b/js/chat/README.md
@@ -40,12 +40,15 @@ npm run dev
 This will build the project and start a dev server.
 Point your browser to http://localhost:8080/.
 
-### The ONNX Model
+### The Phi3 ONNX Model
 
-The model used in this project is hosted on [Hugging Face](https://huggingface.co/schmuell/phi3-int4). It was created using the [onnx model builder](https://github.com/microsoft/onnxruntime-genai/tree/main/src/python/py/models).
+The model used in this example is hosted on [Hugging Face](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct-onnx-web). It is slightly different than the ONNX model for CUDA or CPU:
+1. The model output 'logits' is kept as float32 (even for float16 models) since Javascript does not support float16.
+2. Our WebGPU implementation uses the custom Multiheaded Attention operator instread of Group Query Attention.
+3. Phi3 is larger then 2GB and we need to use external data files. To keep them cacheable in the browser,
+ both model.onnx and model.onnx.data are kept under 2GB.
 
-You can create the model with 
+The model was created using the [ONNX genai model builder](https://github.com/microsoft/onnxruntime-genai/tree/main/src/python/py/models).
 
-```sh
-python builder.py -m microsoft/Phi-3-mini-4k-instruct -o $your_output -p int4 -e web
-```
+If you like to create the model yourself, you can use [Olive](https://github.com/microsoft/Olive/).
+An example how to create the model for ONNX Runtime Web with Olive can be found [here](https://github.com/microsoft/Olive/tree/main/examples/phi3).
diff --git a/js/chat/main.js b/js/chat/main.js
index d2bd3d23..10d7599c 100644
--- a/js/chat/main.js
+++ b/js/chat/main.js
@@ -4,12 +4,7 @@ import { marked } from 'marked';
 
 
 const MODELS = {
-  "tinyllama": { name: "tinyllama", path: "schmuell/TinyLlama-1.1B-Chat-v1.0-int4", file: "decoder_model_merged" },
-  "tinyllama_fp16": { name: "tinyllama-fp16", path: "schmuell/TinyLlama-1.1B-Chat-v1.0-fp16", externaldata: true, file: "decoder_model_merged" },
-  "phi2": { name: "phi2", path: "schmuell/phi2-int4", file: "decoder_model_merged" },
   "phi3": { name: "phi3", path: "schmuell/phi3-int4", externaldata: true },
-  "phi3-1": { name: "phi3-1", path: "schmuell/phi3-1", externaldata: true },
-  "stablelm": { name: "stablelm", path: "schmuell/stablelm-2-zephyr-1_6b-int4", file: "decoder_model_merged" },
 }
 
 const preCannedQueries = {