> ## Documentation Index
> Fetch the complete documentation index at: https://docs.finetunedb.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Web API

> Connect via API.

## Endpoints

### POST /logs

[https://app.finetunedb.com/api/v1/logs](https://app.finetunedb.com/api/v1/logs)

Creates a new log.

#### Parameters

* `projectId`: string
* `parentId` (optional): string
* `name` (optional): string
* `provider`: enum (openai)
* `tags` (optional): array of strings
* `model`: string
* `source`: string
* `modelParameters`: object
  * `model`: string
  * `provider` (optional): string
  * `max_tokens` (optional): number
  * `temperature` (optional): number
  * `top_p` (optional): number
  * `stop_sequences` (optional): array of strings
  * `presence_penalty` (optional): number
  * `frequency_penalty` (optional): number
  * `other` (optional): record of any type
* `input` (optional): union of input types (Documented below)
* `output` (optional): union of output types (Documented below)
* `metadata` (optional): record of any type
* `type`: enum (CHATCOMPLETION, EMBEDDING)
* `error` (optional): string
* `latencyMs` (optional): number

### Input Types

The log.input accepts different kinds of values.

#### OpenAI Completion (If type=CHATCOMPLETION)

Pass the messages value that you use in openai.chat.completions.create
Example:

```json theme={null}
[
  {
    "role": "system",
    "content": "You are an expert assistant"
  },
  {
    "role": "user",
    "content": "Hello, how are you?"
  }
]
```

#### OpenAI Embedding (If type=EMBEDDING)

Pass the input value that you send to openai.embeddings.create
Example:

```json theme={null}
"Hello world!"
```

or

```json theme={null}
["Hello", "world!"]
```

### Output Types

The log.output accepts different kinds of values.

#### OpenAI Completion (If type=CHATCOMPLETION)

Pass the response value that you get from openai.chat.completions.create as an array.
Example:

```javascript theme={null}
const response = await openai.chat.completions.create(...);
const output = [response.choices[0].message];
```

OpenAI Embedding (If type=EMBEDDING)
Pass the response.data value that you get from openai.embeddings.create.

Example:

```javascript theme={null}
const response = await openai.embeddings.create(...);
const output = response.data;
```