GooseAI

https://goose.ai/docs
Admin
GooseAI is a fully managed inference service that provides easy access to open source language...
Tokens:8,039
Snippets:99
Trust Score:8.8
Update:6 months ago
Show doc for...
Tokens:
Raw
### Install OpenAI Python Client

Source: https://goose.ai/docs/api

Install the required package via pip to enable GooseAI API access in Python.

```bash
$ pip install openai
```

--------------------------------

### Install OpenAI NodeJS Client

Source: https://goose.ai/docs/api

Install the required package via npm to enable GooseAI API access in NodeJS.

```bash
$ npm install openai
```

--------------------------------

### Configure stop sequences

Source: https://goose.ai/docs/api/completions

Examples of defining stop sequences as a single string or an array of strings.

```json
{...
"stop": " goose",
...}
```

```json
{...
"stop: [" geese", " gooose", "<|endoftext|>"],
...}
```

--------------------------------

### Configure logit_bias for gpt-neo-20b

Source: https://goose.ai/docs/api/completions

Example of applying logit bias to specific token IDs for the gpt-neo-20b model.

```json
{...
"logit_bias": {"0": -100,
               "187": 0.5},
...}
```

--------------------------------

### Get Engine Details

Source: https://goose.ai/docs/api/engines

Retrieves information about a specific engine available in the API, such as its owner, name, availability, and a brief description.

```APIDOC
## GET /v1/engines/{engine_id}

### Description
Gets information about an individual engine available, such as the owner, name, availability, and description.

### Method
GET

### Endpoint
/v1/engines/{engine_id}

### Parameters
#### Path Parameters
- **engine_id** (string) - Required - The unique identifier of the engine.

### Request Example
```curl
curl https://api.goose.ai/v1/engines/gpt-j-6b \
  -H 'Authorization: Bearer YOUR_API_KEY'
```

### Response
#### Success Response (200)
- **description** (string) - A description of the engine.
- **id** (string) - The unique identifier of the engine.
- **name** (string) - The display name of the engine.
- **object** (string) - The type of object, expected to be "engine".
- **owner** (string) - The owner of the engine.
- **ready** (boolean) - Indicates if the engine is ready for use.
- **tokenizer** (string) - The tokenizer used by the engine.
- **type** (string) - The type of the engine (e.g., "text").

#### Response Example
```json
{
  "description": "6B parameter EleutherAI model trained on the Pile, using the Mesh Transformer JAX framework.",
  "id": "gpt-j-6b",
  "name": "GPT-J 6B",
  "object": "engine",
  "owner": "gooseai",
  "ready": true,
  "tokenizer": "gpt2",
  "type": "text"
}
```
```

--------------------------------

### Configure logit_bias for other models

Source: https://goose.ai/docs/api/completions

Example of applying logit bias to specific token IDs for models other than gpt-neo-20b.

```json
{...
"logit_bias": {"50256": -100,
               "198":  0.5},
...}
```

--------------------------------

### Get Specific Engine Information

Source: https://goose.ai/docs/api/engines

Retrieve details about a single engine by its ID. Ensure you replace `{engine_id}` with the actual engine identifier.

```HTTP
GET https://api.goose.ai/v1/engines/{engine_id}
```

```curl
curl https://api.goose.ai/v1/engines/gpt-j-6b \
  -H 'Authorization: Bearer YOUR_API_KEY'
```

```json
{
  "description": "6B parameter EleutherAI model trained on the Pile, using the Mesh Transformer JAX framework.",
  "id": "gpt-j-6b",
  "name": "GPT-J 6B",
  "object": "engine",
  "owner": "gooseai",
  "ready": true,
  "tokenizer": "gpt2",
  "type": "text"
}
```

--------------------------------

### POST /v1/engines/{engine_id}/completions

Source: https://goose.ai/docs/api/completions

Creates a new completion for the provided prompt and parameters.

```APIDOC
## POST /v1/engines/{engine_id}/completions

### Description
Creates a new completion for the provided prompt and parameters.

### Method
POST

### Endpoint
/v1/engines/{engine_id}/completions

### Parameters
#### Path Parameters
- **engine_id** (string) - Required - The ID of the engine to use for this request.
```

--------------------------------

### Interact with GooseAI API in Python

Source: https://goose.ai/docs/api

Configure the OpenAI client to point to the GooseAI base URL and perform engine listing and completion tasks.

```python
import openai
openai.api_key = "sk-..."
openai.api_base = "https://api.goose.ai/v1"

# List Engines (Models)
engines = openai.Engine.list()
# Print all engines IDs
for engine in engines.data:
  print(engine.id)

# Create a completion, return results streaming as they are generated. Run with `python3 -u` to ensure unbuffered output.
completion = openai.Completion.create(
  engine="gpt-j-6b",
  prompt="Once upon a time there was a Goose. ",
  max_tokens=160,
  stream=True)

# Print each token as it is returned
for c in completion:
  print (c.choices[0].text, end = '')

print("")
```

--------------------------------

### List All Available Engines

Source: https://goose.ai/docs/api/engines

Fetch a list of all engines currently available through the API, including their descriptions and configurations. This endpoint provides an overview of the models you can utilize.

```HTTP
GET https://api.goose.ai/v1/engines
```

```curl
curl https://api.goose.ai/v1/engines \
  -H 'Authorization: Bearer YOUR_API_KEY'
```

```json
{
  "data": [
    {
      "description": "20B parameter EleutherAI model trained on the Pile, using the NeoX framework.",
      "id": "gpt-neo-20b",
      "name": "GPT-NeoX 20B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "pile",
      "type": "text"
    },
    {
      "description": "6B parameter EleutherAI model trained on the Pile, using the Mesh Transformer JAX framework.",
      "id": "gpt-j-6b",
      "name": "GPT-J 6B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "2.7B parameter EleutherAI model trained on the Pile, using the Neo framework.",
      "id": "gpt-neo-2-7b",
      "name": "GPT-Neo 2.7B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "1.3B parameter EleutherAI model trained on the Pile, using the Neo framework.",
      "id": "gpt-neo-1-3b",
      "name": "GPT-Neo 1.3B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "125M parameter EleutherAI model trained on the Pile, using the Neo framework.",
      "id": "gpt-neo-125m",
      "name": "GPT-Neo 125M",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "13B parameter Facebook Mixture of Experts model trained on RoBERTa and CC100 subset data.",
      "id": "fairseq-13b",
      "name": "Fairseq 13B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "6.7B parameter Facebook Mixture of Experts model trained on RoBERTa and CC100 subset data.",
      "id": "fairseq-6-7b",
      "name": "Fairseq 6.7B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "2.7B parameter Facebook Mixture of Experts model trained on RoBERTa and CC100 subset data.",
      "id": "fairseq-2-7b",
      "name": "Fairseq 2.7B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "1.3B parameter Facebook Mixture of Experts model trained on RoBERTa and CC100 subset data.",
      "id": "fairseq-1-3b",
      "name": "Fairseq 1.3B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    },
    {
      "description": "125M parameter Facebook Mixture of Experts model trained on RoBERTa and CC100 subset data.",
      "id": "fairseq-125m",
      "name": "Fairseq 125M",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    }
  ],
  "object": "list"
}
```

--------------------------------

### List All Engines

Source: https://goose.ai/docs/api/engines

Lists all currently available engines and provides basic information about each, including owner and availability.

```APIDOC
## GET /v1/engines

### Description
Lists the currently available engines, and provides basic information about each one such as the owner and availability.

### Method
GET

### Endpoint
/v1/engines

### Request Example
```curl
curl https://api.goose.ai/v1/engines \
  -H 'Authorization: Bearer YOUR_API_KEY'
```

### Response
#### Success Response (200)
- **data** (array) - A list of engine objects.
  - Each object contains:
    - **description** (string) - A description of the engine.
    - **id** (string) - The unique identifier of the engine.
    - **name** (string) - The display name of the engine.
    - **object** (string) - The type of object, expected to be "engine".
    - **owner** (string) - The owner of the engine.
    - **ready** (boolean) - Indicates if the engine is ready for use.
    - **tokenizer** (string) - The tokenizer used by the engine.
    - **type** (string) - The type of the engine (e.g., "text").
- **object** (string) - The type of object, expected to be "list".

#### Response Example
```json
{
  "data": [
    {
      "description": "20B parameter EleutherAI model trained on the Pile, using the NeoX framework.",
      "id": "gpt-neo-20b",
      "name": "GPT-NeoX 20B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "pile",
      "type": "text"
    },
    {
      "description": "6B parameter EleutherAI model trained on the Pile, using the Mesh Transformer JAX framework.",
      "id": "gpt-j-6b",
      "name": "GPT-J 6B",
      "object": "engine",
      "owner": "gooseai",
      "ready": true,
      "tokenizer": "gpt2",
      "type": "text"
    }
  ],
  "object": "list"
}
```
```

--------------------------------

### Interact with GooseAI API in NodeJS

Source: https://goose.ai/docs/api

Configure the OpenAI client with the GooseAI base path and execute a completion request.

```javascript
const { Configuration, OpenAIApi } = require("openai");

const configuration = new Configuration({
  apiKey: 'sk-...',
  basePath: 'https://api.goose.ai/v1',
});

const openai = new OpenAIApi(configuration);

openai.createCompletion("gpt-j-6b", {
  prompt: `Roses are red `,
  max_tokens: 25,
}).then(
  completion =>
    console.log(completion.data.choices[0].text)
);
```

--------------------------------

### Text Completion API

Source: https://goose.ai/docs/api/completions

This endpoint generates text completions based on provided prompts and various configuration parameters.

```APIDOC
## POST /websites/goose_ai

### Description
Generates text completions for one or more prompts.

### Method
POST

### Endpoint
/websites/goose_ai

### Parameters
#### Request Body
- **prompt** (string or array) - Optional - Defaults to <|endoftext|> - The prompt(s) to generate completions for, encoded as a string, array of strings, array of tokens, or array of token arrays. If multiple prompts are provided in a single request, they will be acted upon concurrently.
- **n** (integer) - Optional - Defaults to 1 - The number of completions to perform per prompt. If more than 1 completion is requested, they will be fulfilled concurrently.
- **max_tokens** (integer) - Optional - Defaults to 16 - The maximum number of tokens to generate in the completion. The token count of your prompt plus max_tokens cannot exceed the model's context length.
- **min_tokens** (integer) - Optional - Defaults to 1 - The minimum number of tokens to generate in the completion. If stop tokens are seen before min_tokens, these tokens will be allowed to be generated.
- **temperature** (number) - Optional - Defaults to 1.0 - Number between 0 and 1.0. What sampling temperature to use. Higher values mean the model will take more risks.
- **top_p** (number) - Optional - Defaults to 1.0 - Number between 0 and 1.0. An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass.
- **logit_bias** (map) - Optional - Defaults to null - Bias for or against the specified tokens appearing. Keys are token IDs and values are floating points ranging from -100 to 100.
- **stop** (string or array) - Optional - Defaults to null - Stops completion when a string or one of the strings in the array is encountered.
- **top_k** (integer) - Optional - Defaults to 0 - Truncates logits to the set value.
- **tfs** (number) - Optional - Defaults to 1.0 - Number between 0 and 1.0. Similar to nucleus sampling, but it sets its cutoff point based on the cumulative sum of the accelerations of the sorted token probabilities.
- **top_a** (number) - Optional - Defaults to 1.0 - Number between 0 and 1.0. Remove all tokens that have probability below the threshold of: limit = pow(max(probs), 2.0) * top_a.
- **typical_p** (number) - Optional - Defaults to 1.0 - Number between 0 and 1.0. Selects tokens according to the expected amount of information they contribute.
- **stream** (boolean) - Optional - Defaults to false - Whether to stream back partial progress.
- **logprobs** (integer) - Optional - Defaults to null - Include the log probabilities on the logprobs most likely tokens, as well as the chosen tokens.
- **echo** (boolean) - Optional - Defaults to false - Echo back the prompt in addition to the completion.

### Request Example
```json
{
  "prompt": "The quick brown fox",
  "max_tokens": 50,
  "temperature": 0.7
}
```

### Response
#### Success Response (200)
- **choices** (array) - An array of completion choices.
- **created** (integer) - Timestamp of creation.
- **id** (string) - Unique identifier for the completion.
- **model** (string) - The model used for completion.
- **object** (string) - The type of object returned.
- **usage** (object) - Usage statistics for the completion.

#### Response Example
```json
{
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "logprobs": null,
      "text": " jumps over the lazy dog."
    }
  ],
  "created": 1506817577,
  "id": "cmpl-7q8k3",
  "model": "gpt-3.5-turbo-instruct",
  "object": "text_completion",
  "usage": {
    "completion_tokens": 5,
    "prompt_tokens": 4,
    "total_tokens": 9
  }
}
```
```

--------------------------------

### Execute API Request via Shell

Source: https://goose.ai/docs/api

Use curl to send a completion request directly to the GooseAI API endpoint.

```shell
curl https://api.goose.ai/v1/engines/gpt-j-6b/completions \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer sk-******' \
  -d '{
     "prompt": "Roses are red ",
     "max_tokens": 25
  }'
```
Tokens:
Raw