Skip to content

Commit

Permalink
Document generate HTTP endpoint (#6412)
Browse files Browse the repository at this point in the history
* Document generate HTTP endpoint

* Address comment

* Fix up

* format

* Address comment
  • Loading branch information
GuanLuo authored and mc-nv committed Oct 27, 2023
1 parent 011b9a5 commit d854e5f
Showing 1 changed file with 178 additions and 0 deletions.
178 changes: 178 additions & 0 deletions docs/protocol/extension_generate.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,178 @@
<!--
# Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
# * Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
# * Neither the name of NVIDIA CORPORATION nor the names of its
# contributors may be used to endorse or promote products derived
# from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-->

# Generate Extension

> [!NOTE]
> The Generate Extension is *provisional* and likely to change in future versions.
This document describes Triton's generate extension. The generate
extension provides a simple text-oriented endpoint schema for interacting with
large language models (LLMs). The generate endpoint is specific to HTTP/REST
frontend.

## HTTP/REST

In all JSON schemas shown in this document, `$number`, `$string`, `$boolean`,
`$object` and `$array` refer to the fundamental JSON types. #optional
indicates an optional JSON field.

Triton exposes the generate endpoint at the following URLs. The client may use
HTTP POST request to different URLs for different response behavior, the
endpoint will return the generate results on success or an error in the case of
failure.

```
POST v2/models/${MODEL_NAME}[/versions/${MODEL_VERSION}]/generate
POST v2/models/${MODEL_NAME}[/versions/${MODEL_VERSION}]/generate_stream
```

### generate v.s. generate_stream

Both URLs expect the same request JSON object, and generate the same response
JSON object. However, `generate` returns exactly 1 response JSON object, while
`generate_stream` may return multiple responses based on the inference
results. `generate_stream` returns the responses as
[Server-Sent Events](https://html.spec.whatwg.org/multipage/server-sent-events.html#server-sent-events)
(SSE), where each response will be a "data" chunk in the HTTP response body.
Also, note that an error may be returned during inference, whereas the HTTP
response code has been set in the first response of the SSE, which can result in
receiving an [error object](#generate-response-json-error-object) while status
code shows success (200). Therefore the user must always check whether an error
object is received when generating responses through `generate_stream`.

### Generate Request JSON Object

The generate request object, identified as *$generate_request*, is
required in the HTTP body of the POST request. The model name and
(optionally) version must be available in the URL. If a version is not
provided, the server may choose a version based on its own policies or
return an error.

$generate_request =
{
"text_input" : $string,
"parameters" : $parameters #optional
}

* "text_input" : The text input that the model should generate output from.
* "parameters" : An optional object containing zero or more parameters for this
generate request expressed as key/value pairs. See
[Parameters](#parameters) for more information.

> [!NOTE]
> Any additional properties in the request object are passed either as
> parameters or tensors based on model specification.
#### Parameters

The *$parameters* JSON describes zero or more “name”/”value” pairs,
where the “name” is the name of the parameter and the “value” is a
$string, $number, or $boolean.

$parameters =
{
$parameter, ...
}

$parameter = $string : $string | $number | $boolean

Parameters are model-specific. The user should check with the model
specification to set the parameters.

#### Example Request

Below is an example to send generate request with additional model parameters `stream` and `temperature`.

```
$ curl -X POST localhost:8000/v2/models/mymodel/generate -d '{"text_input": "client input", "parameters": {"stream": false, "temperature": 0}}'
POST /v2/models/mymodel/generate HTTP/1.1
Host: localhost:8000
Content-Type: application/json
Content-Length: <xx>
{
"text_input": "client input",
"parameters" :
{
"stream": false,
"temperature": 0
}
}
```

### Generate Response JSON Object

A successful generate request is indicated by a 200 HTTP status code.
The generate response object, identified as *$generate_response*, is returned in
the HTTP body.

$generate_response =
{
"model_name" : $string,
"model_version" : $string,
"text_output" : $string
}

* "model_name" : The name of the model used for inference.
* "model_version" : The specific model version used for inference.
* "text_output" : The output of the inference.

#### Example Response

```
200
{
"model_name" : "mymodel",
"model_version" : "1",
"text_output" : "model output"
}
```

### Generate Response JSON Error Object

A failed generate request must be indicated by an HTTP error status
(typically 400). The HTTP body must contain the
*$generate_error_response* object.

$generate_error_response =
{
"error": <error message string>
}

* “error” : The descriptive message for the error.

#### Example Error

```
400
{
"error" : "error message"
}
```

0 comments on commit d854e5f

Please sign in to comment.