[Website] Added content for Connecting Llama via Hugging Face

facebookincubator · Mar 25, 2024 · 3d369e5 · 3d369e5
1 parent f50ad94
commit 3d369e5
Show file tree

Hide file tree

Showing 18 changed files with 87 additions and 10 deletions.
diff --git a/docs/docs/connecting-your-llm/connecting-llama-via-huggingface.md b/docs/docs/connecting-your-llm/connecting-llama-via-huggingface.md
@@ -0,0 +1,63 @@
+---
+sidebar_position: 1
+---
+
+# Connecting to Llama via Hugging Face
+
+To connect CP4M to Llama via Hugging Face, you will need:
+1. A Hugging Face account
+2. Access to the Llama model from Meta
+
+## Sign up for a Hugging Face account
+Hugging Face is a platform that helps you build, deploy, and train machine learning models. You can sign up for an account [here](https://huggingface.co/join).
+
+![Hugging Face Home Page](./img/hugging-face-home-page.png)
+
+## Accessing the Llama model on Hugging Face
+Hugging Face has a large repository of machine learning models for a variety of use cases accessible [here](https://huggingface.co/models).
+
+You can search for the LLama model from Meta of your choice. For the purposes of this tutorial, we will be using the [LLama-2-7b-chat-hf model](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf).
+
+In order to use the Llama model, you will need to accept the Llama 2 Community License Agreement on the [model page](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf).
+
+![Gated model agremeent](./img/hugging-face-gated-model-agreement.png)
+
+Once your account is approved, the model page will inform you that you have been granted access.
+
+![Gated model access](./img/hugging-face-gated-model-access.png)
+
+## Deploying model using Inference API
+The fastest way to deploy Llama is through Hugging Face's serverless Inference API.
+
+1. Go to the [Llama model page](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf).
+2. On the top right, click on the **Deploy** dropdown button.
+
+![Model Deploy](./img/hugging-face-model-deploy.png)
+
+
+3. In the dropdown, click the **Inference API (serverless)** option. *Note: Using Inference API requires a [Hugging Face Pro subscription](https://huggingface.co/pricing)*.
+
+![Inference API First Time](./img/hugging-face-inference-api-first-time.png)
+
+4. On the new modal, click the **Token** dropdown then click **+ New access token**.
+
+![Inference API New Token](./img/hugging-face-inference-api-new-token.png)
+
+5. On the access tokens page, click the **New Token** button.
+
+![Access Token Page](./img/hugging-face-access-token.png)
+
+6. Create a new access token with the **Write** permission.
+
+![Create Access Token Modal](./img/hugging-face-create-access-token.png)
+
+7. Go back to the Model Deploy > Inference API page [here](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf?inference_api=true). Your newly created access token should now be populated in the **Token** section.
+
+## Updating your CP4M config from Inference API
+7. Copy and paste the *API_URL* on the Deploy Inference API modal to your CP4M config file in the "\[\[plugins]] endpoint" entry. If you are using the Llama-2-7b-chat-hf model, it should be "https://api-inference.huggingface.co/models/meta-llama/Llama-2-7b-chat-hf".
+
+![Inference API code](./img/hugging-face-inference-api-code.png)
+
+8. Click **Show API token**. Copy and paste this API token to your CP4M config file in the "\[\[plugins]] api_key" entry.
+
+Congratulations! You have now successfully set up CP4M with Llama via Hugging Face.
diff --git a/docs/docs/connecting-your-llm/connecting-openai.md b/docs/docs/connecting-your-llm/connecting-openai.md
@@ -0,0 +1,7 @@
+---
+sidebar_position: 2
+---
+
+# Connecting to OpenAI
+
+Coming soon...
diff --git a/docs/docs/connecting-your-llm/img/docsVersionDropdown.png b/docs/docs/connecting-your-llm/img/docsVersionDropdown.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-access-token.png b/docs/docs/connecting-your-llm/img/hugging-face-access-token.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-create-access-token.png b/docs/docs/connecting-your-llm/img/hugging-face-create-access-token.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-gated-model-access.png b/docs/docs/connecting-your-llm/img/hugging-face-gated-model-access.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-gated-model-agreement.png b/docs/docs/connecting-your-llm/img/hugging-face-gated-model-agreement.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-home-page.png b/docs/docs/connecting-your-llm/img/hugging-face-home-page.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-inference-api-code.png b/docs/docs/connecting-your-llm/img/hugging-face-inference-api-code.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-inference-api-first-time.png b/docs/docs/connecting-your-llm/img/hugging-face-inference-api-first-time.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-inference-api-new-token.png b/docs/docs/connecting-your-llm/img/hugging-face-inference-api-new-token.png
diff --git a/docs/docs/connecting-your-llm/img/hugging-face-model-deploy.png b/docs/docs/connecting-your-llm/img/hugging-face-model-deploy.png
diff --git a/docs/docs/connecting-your-llm/img/localeDropdown.png b/docs/docs/connecting-your-llm/img/localeDropdown.png
diff --git a/docs/docs/connecting-your-llm/llama-via-huggingface.md b/docs/docs/connecting-your-llm/llama-via-huggingface.md
diff --git a/docs/docs/messaging-integrations/connecting-to-instagram-direct.md b/docs/docs/messaging-integrations/connecting-to-instagram-direct.md
@@ -0,0 +1,7 @@
+---
+sidebar_position: 2
+---
+
+# Connecting to Instagram Direct
+
+Coming soon...
diff --git a/docs/docs/messaging-integrations/connecting-to-messenger.md b/docs/docs/messaging-integrations/connecting-to-messenger.md
@@ -6,8 +6,8 @@ sidebar_position: 1
 
 To connect CP4M to Facebook Messenger, you will need:
 1. A Facebook account registered as a Meta developer
-1. A Meta app to integrate with Meta APIs
-2. A Facebook Page for users to interact with over Messenger
+2. A Meta app to integrate with Meta APIs
+3. A Facebook Page for users to interact with over Messenger
 
 ## Getting Meta developer access
 To register as a Meta developer, you can follow the steps [here](https://developers.facebook.com/docs/development/register).

diff --git a/docs/docs/messaging-integrations/connecting-to-whatsapp.md b/docs/docs/messaging-integrations/connecting-to-whatsapp.md
@@ -0,0 +1,7 @@
+---
+sidebar_position: 3
+---
+
+# Connecting to Whatsapp
+
+Coming soon...
diff --git a/docs/src/pages/index.js b/docs/src/pages/index.js
@@ -35,7 +35,7 @@ const features = [
     ),
   },
   {
-    title: 'Free and Open Source',
+    title: 'Welcoming Collaborators',
     imageUrl: 'CP4M/img/undraw_docusaurus_react.svg',
     description: (
       <>