You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Allow manual setting of the OpenAI Deployment Name, Version, Capacity and Model (#1386)
* check for chat or chat4 to enable different deployments
* add docs for deployment
* line ending
* add USE_GPT4
* remove list
* update docs
* Update docs/deploy_features.md
* add version check
* update to use new isAzureOpenAiHost
* precommit formatter
* revert back name of chat and embeddings deployments
* format with black
* Correct names in parameters
* Add embedding version as parameter
* use empty to check for definition
* change 'azure openai chatgpt model' var back to original name
* change 'azure openai chatgpt model' to match embedding model var
* docs: add intruction for gpt-4
* Use previous env var name
---------
Co-authored-by: Anthony Shaw <[email protected]>
Co-authored-by: Pamela Fox <[email protected]>
Copy file name to clipboardExpand all lines: docs/deploy_features.md
+41-2Lines changed: 41 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -16,8 +16,47 @@ You should typically enable these features before running `azd up`. Once you've
16
16
17
17
We generally find that most developers are able to get high quality answers using GPT 3.5. However, if you want to try GPT-4, you can do so by following these steps:
18
18
19
-
* In `infra/main.bicep`, change `chatGptModelName` to 'gpt-4' instead of 'gpt-35-turbo'.
20
-
* You may also need to adjust the capacity above that line depending on how much TPM your account is allowed.
19
+
Execute the following commands inside your terminal:
20
+
21
+
1. To set the name of the deployment, run this command with a new unique name.
22
+
23
+
```bash
24
+
azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT chat4
25
+
```
26
+
27
+
1. To set the GPT model name to a **gpt-4** version from the [available models](https://learn.microsoft.com/azure/ai-services/openai/concepts/models), run this command with the appropriate gpt model name.
28
+
29
+
```bash
30
+
azd env set AZURE_OPENAI_CHATGPT_MODEL gpt-4
31
+
```
32
+
33
+
1. To set the Azure OpenAI deploymemnt capacity, run this command with the desired capacity.
34
+
35
+
```bash
36
+
azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT_CAPACITY 10
37
+
```
38
+
39
+
1. To set the Azure OpenAI deploymemnt version from the [available versions](https://learn.microsoft.com/azure/ai-services/openai/concepts/models), run this command with the appropriate version.
40
+
41
+
```bash
42
+
azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT_VERSION 0125-Preview
43
+
```
44
+
45
+
1. To updat the deployment with the new parameters, run this command.
46
+
47
+
```bash
48
+
azd up
49
+
```
50
+
51
+
> [!NOTE]
52
+
> To revert back to GPT 3.5, run the following commands:
53
+
> - `azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT chat` to set the name of your old GPT 3.5 deployment.
54
+
> - `azd env set AZURE_OPENAI_CHATGPT_MODEL gpt-35-turbo` to set the name of your old GPT 3.5 model.
55
+
> - `azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT_CAPACITY 30` to set the capacity of your old GPT 3.5 deployment.
56
+
> - `azd env set AZURE_OPENAI_CHATGPT_DEPLOYMENT_VERSION 0613` to set the version number of your old GPT 3.5.
57
+
> - `azd up` to update the provisioned resources.
58
+
>
59
+
> Note that this does not delete your GPT-4 deployment; it just makes your application create a new or reuse an old GPT 3.5 deployment. If you want to delete it, you can go to your Azure OpenAI studio and do so.
0 commit comments