OpenAI API参考文档 | OpenAI开发文档|OpenAI中文官方文档|ChatGPT中文版|ChatGPT教程

Fine-tuning

Manage fine-tuning jobs to tailor a model to your specific training data. Related guide: Fine-tune models source

Create fine-tuning job

post https://api.openai.com/v1/fine_tuning/jobs

Creates a fine-tuning job which begins the process of creating a new model from a given dataset.source

Response includes details of the enqueued job including job status and the name of the fine-tuned models once complete.source

Learn more about fine-tuning source

Request body

model

stringsource

Required

The name of the model to fine-tune. You can select one of the supported models.source

training_file

stringsource

Required

The ID of an uploaded file that contains training data.source

See upload file for how to upload a file.source

Your dataset must be formatted as a JSONL file. Additionally, you must upload your file with the purpose fine-tune.source

The contents of the file should differ depending on if the model uses the chat or completions format.source

See the fine-tuning guide for more details.source

hyperparameters

objectsource

Optional

The hyperparameters used for the fine-tuning job.source

suffix

string or nullsource

Optional

Defaults to null

A string of up to 64 characters that will be added to your fine-tuned model name.source

For example, a suffix of "custom-model-name" would produce a model name like ft:gpt-4o-mini:openai:custom-model-name:7p4lURel.source

validation_file

string or nullsource

Optional

The ID of an uploaded file that contains validation data.source

If you provide this file, the data is used to generate validation metrics periodically during fine-tuning. These metrics can be viewed in the fine-tuning results file. The same data should not be present in both train and validation files.source

Your dataset must be formatted as a JSONL file. You must upload your file with the purpose fine-tune.source

See the fine-tuning guide for more details.source

integrations

array or nullsource

Optional

A list of integrations to enable for your fine-tuning job.source

seed

integer or nullsource

Optional

The seed controls the reproducibility of the job. Passing in the same seed and job parameters should produce the same results, but may differ in rare cases. If a seed is not specified, one will be generated for you.source

Returns

A fine-tuning.job object.source

Example request

1
2
3
4
5
6
7
curl https://api.openai.com/v1/fine_tuning/jobs \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "training_file": "file-BK7bzQj3FfZFXr7DbL6xJwfo",
    "model": "gpt-4o-mini"
  }'

1
2
3
4
5
6
7
from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.create(
  training_file="file-abc123",
  model="gpt-4o-mini"
)

1
2
3
4
5
6
7
8
9
10
11
12
13
import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const fineTune = await openai.fineTuning.jobs.create({
    training_file: "file-abc123"
  });

  console.log(fineTune);
}

main();

Response

1
2
3
4
5
6
7
8
9
10
11
12
{
  "object": "fine_tuning.job",
  "id": "ftjob-abc123",
  "model": "gpt-4o-mini-2024-07-18",
  "created_at": 1721764800,
  "fine_tuned_model": null,
  "organization_id": "org-123",
  "result_files": [],
  "status": "queued",
  "validation_file": null,
  "training_file": "file-abc123",
}

List fine-tuning jobs

get https://api.openai.com/v1/fine_tuning/jobs

List your organization's fine-tuning jobssource

Query parameters

after

stringsource

Optional

Identifier for the last job from the previous pagination request.source

limit

integersource

Optional

Defaults to 20

Number of fine-tuning jobs to retrieve.source

Returns

A list of paginated fine-tuning job objects.source

Example request

1
2
curl https://api.openai.com/v1/fine_tuning/jobs?limit=2 \
  -H "Authorization: Bearer $OPENAI_API_KEY"

1
2
3
4
from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.list()

1
2
3
4
5
6
7
8
9
10
11
12
13
import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const list = await openai.fineTuning.jobs.list();

  for await (const fineTune of list) {
    console.log(fineTune);
  }
}

main();

Response

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
{
  "object": "list",
  "data": [
    {
      "object": "fine_tuning.job.event",
      "id": "ft-event-TjX0lMfOniCZX64t9PUQT5hn",
      "created_at": 1689813489,
      "level": "warn",
      "message": "Fine tuning process stopping due to job cancellation",
      "data": null,
      "type": "message"
    },
    { ... },
    { ... }
  ], "has_more": true
}

List fine-tuning events

get https://api.openai.com/v1/fine_tuning/jobs/{fine_tuning_job_id}/events

Get status updates for a fine-tuning job.source

Path parameters

fine_tuning_job_id

stringsource

Required

The ID of the fine-tuning job to get events for.source

Query parameters

after

stringsource

Optional

Identifier for the last event from the previous pagination request.source

limit

integersource

Optional

Defaults to 20

Number of events to retrieve.source

Returns

A list of fine-tuning event objects.source

Example request

1
2
curl https://api.openai.com/v1/fine_tuning/jobs/ftjob-abc123/events \
  -H "Authorization: Bearer $OPENAI_API_KEY"

1
2
3
4
5
6
7
from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.list_events(
  fine_tuning_job_id="ftjob-abc123",
  limit=2
)

1
2
3
4
5
6
7
8
9
10
11
12
13
import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const list = await openai.fineTuning.list_events(id="ftjob-abc123", limit=2);

  for await (const fineTune of list) {
    console.log(fineTune);
  }
}

main();

Response

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
  "object": "list",
  "data": [
    {
      "object": "fine_tuning.job.event",
      "id": "ft-event-ddTJfwuMVpfLXseO0Am0Gqjm",
      "created_at": 1721764800,
      "level": "info",
      "message": "Fine tuning job successfully completed",
      "data": null,
      "type": "message"
    },
    {
      "object": "fine_tuning.job.event",
      "id": "ft-event-tyiGuB72evQncpH87xe505Sv",
      "created_at": 1721764800,
      "level": "info",
      "message": "New fine-tuned model created: ft:gpt-4o-mini:openai::7p4lURel",
      "data": null,
      "type": "message"
    }
  ],
  "has_more": true
}

List fine-tuning checkpoints

get https://api.openai.com/v1/fine_tuning/jobs/{fine_tuning_job_id}/checkpoints

List checkpoints for a fine-tuning job.source

Path parameters

fine_tuning_job_id

stringsource

Required

The ID of the fine-tuning job to get checkpoints for.source

Query parameters

after

stringsource

Optional

Identifier for the last checkpoint ID from the previous pagination request.source

limit

integersource

Optional

Defaults to 10

Number of checkpoints to retrieve.source

Returns

A list of fine-tuning checkpoint objects for a fine-tuning job.source

Example request

1
2
curl https://api.openai.com/v1/fine_tuning/jobs/ftjob-abc123/checkpoints \
  -H "Authorization: Bearer $OPENAI_API_KEY"

Response

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
{
  "object": "list"
  "data": [
    {
      "object": "fine_tuning.job.checkpoint",
      "id": "ftckpt_zc4Q7MP6XxulcVzj4MZdwsAB",
      "created_at": 1721764867,
      "fine_tuned_model_checkpoint": "ft:gpt-4o-mini-2024-07-18:my-org:custom-suffix:96olL566:ckpt-step-2000",
      "metrics": {
        "full_valid_loss": 0.134,
        "full_valid_mean_token_accuracy": 0.874
      },
      "fine_tuning_job_id": "ftjob-abc123",
      "step_number": 2000,
    },
    {
      "object": "fine_tuning.job.checkpoint",
      "id": "ftckpt_enQCFmOTGj3syEpYVhBRLTSy",
      "created_at": 1721764800,
      "fine_tuned_model_checkpoint": "ft:gpt-4o-mini-2024-07-18:my-org:custom-suffix:7q8mpxmy:ckpt-step-1000",
      "metrics": {
        "full_valid_loss": 0.167,
        "full_valid_mean_token_accuracy": 0.781
      },
      "fine_tuning_job_id": "ftjob-abc123",
      "step_number": 1000,
    },
  ],
  "first_id": "ftckpt_zc4Q7MP6XxulcVzj4MZdwsAB",
  "last_id": "ftckpt_enQCFmOTGj3syEpYVhBRLTSy",
  "has_more": true
}

Retrieve fine-tuning job

get https://api.openai.com/v1/fine_tuning/jobs/{fine_tuning_job_id}

Get info about a fine-tuning job.source

Learn more about fine-tuning source

Path parameters

fine_tuning_job_id

stringsource

Required

The ID of the fine-tuning job.source

Returns

The fine-tuning object with the given ID.source

Example request

1
2
curl https://api.openai.com/v1/fine_tuning/jobs/ft-AF1WoRqd3aJAHsqc9NY7iL8F \
  -H "Authorization: Bearer $OPENAI_API_KEY"

1
2
3
4
from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.retrieve("ftjob-abc123")

1
2
3
4
5
6
7
8
9
10
11
import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const fineTune = await openai.fineTuning.jobs.retrieve("ftjob-abc123");

  console.log(fineTune);
}

main();

Response

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
  "object": "fine_tuning.job",
  "id": "ftjob-abc123",
  "model": "davinci-002",
  "created_at": 1692661014,
  "finished_at": 1692661190,
  "fine_tuned_model": "ft:davinci-002:my-org:custom_suffix:7q8mpxmy",
  "organization_id": "org-123",
  "result_files": [
      "file-abc123"
  ],
  "status": "succeeded",
  "validation_file": null,
  "training_file": "file-abc123",
  "hyperparameters": {
      "n_epochs": 4,
      "batch_size": 1,
      "learning_rate_multiplier": 1.0
  },
  "trained_tokens": 5768,
  "integrations": [],
  "seed": 0,
  "estimated_finish": 0
}

Cancel fine-tuning

post https://api.openai.com/v1/fine_tuning/jobs/{fine_tuning_job_id}/cancel

Immediately cancel a fine-tune job.source

Path parameters

fine_tuning_job_id

stringsource

Required

The ID of the fine-tuning job to cancel.source

Returns

The cancelled fine-tuning object.source

Example request

1
2
curl -X POST https://api.openai.com/v1/fine_tuning/jobs/ftjob-abc123/cancel \
  -H "Authorization: Bearer $OPENAI_API_KEY"

1
2
3
4
from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.cancel("ftjob-abc123")

1
2
3
4
5
6
7
8
9
10
import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const fineTune = await openai.fineTuning.jobs.cancel("ftjob-abc123");

  console.log(fineTune);
}
main();

Response

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
{
  "object": "fine_tuning.job",
  "id": "ftjob-abc123",
  "model": "gpt-4o-mini-2024-07-18",
  "created_at": 1721764800,
  "fine_tuned_model": null,
  "organization_id": "org-123",
  "result_files": [],
  "hyperparameters": {
    "n_epochs":  "auto"
  },
  "status": "cancelled",
  "validation_file": "file-abc123",
  "training_file": "file-abc123"
}

Training format for chat models

The per-line training example of a fine-tuning input file for chat modelssource

messages

arraysource

tools

arraysource

A list of tools the model may generate JSON inputs for.source

parallel_tool_calls

booleansource

Whether to enable parallel function calling during tool use.source

functions

Deprecated

arraysource

A list of functions the model may generate JSON inputs for.source

OBJECT Training format for chat models

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
{
  "messages": [
    { "role": "user", "content": "What is the weather in San Francisco?" },
    {
      "role": "assistant",
      "tool_calls": [
        {
          "id": "call_id",
          "type": "function",
          "function": {
            "name": "get_current_weather",
            "arguments": "{\"location\": \"San Francisco, USA\", \"format\": \"celsius\"}"
          }
        }
      ]
    }
  ],
  "parallel_tool_calls": false,
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "get_current_weather",
        "description": "Get the current weather",
        "parameters": {
          "type": "object",
          "properties": {
            "location": {
                "type": "string",
                "description": "The city and country, eg. San Francisco, USA"
            },
            "format": { "type": "string", "enum": ["celsius", "fahrenheit"] }
          },
          "required": ["location", "format"]
        }
      }
    }
  ]
}

Training format for completions models

The per-line training example of a fine-tuning input file for completions modelssource

prompt

stringsource

The input prompt for this training example.source

completion

stringsource

The desired completion for this training example.source

OBJECT Training format for completions models

1
2
3
4
{
  "prompt": "What is the answer to 2+2",
  "completion": "4"
}

The fine-tuning job object

The fine_tuning.job object represents a fine-tuning job that has been created through the API.source

stringsource

The object identifier, which can be referenced in the API endpoints.source

created_at

integersource

The Unix timestamp (in seconds) for when the fine-tuning job was created.source

error

object or nullsource

For fine-tuning jobs that have failed, this will contain more information on the cause of the failure.source

fine_tuned_model

string or nullsource

The name of the fine-tuned model that is being created. The value will be null if the fine-tuning job is still running.source

finished_at

integer or nullsource

The Unix timestamp (in seconds) for when the fine-tuning job was finished. The value will be null if the fine-tuning job is still running.source

hyperparameters

objectsource

The hyperparameters used for the fine-tuning job. See the fine-tuning guide for more details.source

model

stringsource

The base model that is being fine-tuned.source

object

stringsource

The object type, which is always "fine_tuning.job".source

organization_id

stringsource

The organization that owns the fine-tuning job.source

result_files

arraysource

The compiled results file ID(s) for the fine-tuning job. You can retrieve the results with the Files API.source

status

stringsource

The current status of the fine-tuning job, which can be either validating_files, queued, running, succeeded, failed, or cancelled.source

trained_tokens

integer or nullsource

The total number of billable tokens processed by this fine-tuning job. The value will be null if the fine-tuning job is still running.source

training_file

stringsource

The file ID used for training. You can retrieve the training data with the Files API.source

validation_file

string or nullsource

The file ID used for validation. You can retrieve the validation results with the Files API.source

integrations

array or nullsource

A list of integrations to enable for this fine-tuning job.source

seed

integersource

The seed used for the fine-tuning job.source

estimated_finish

integer or nullsource

The Unix timestamp (in seconds) for when the fine-tuning job is estimated to finish. The value will be null if the fine-tuning job is not running.source

OBJECT The fine-tuning job object

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
  "object": "fine_tuning.job",
  "id": "ftjob-abc123",
  "model": "davinci-002",
  "created_at": 1692661014,
  "finished_at": 1692661190,
  "fine_tuned_model": "ft:davinci-002:my-org:custom_suffix:7q8mpxmy",
  "organization_id": "org-123",
  "result_files": [
      "file-abc123"
  ],
  "status": "succeeded",
  "validation_file": null,
  "training_file": "file-abc123",
  "hyperparameters": {
      "n_epochs": 4,
      "batch_size": 1,
      "learning_rate_multiplier": 1.0
  },
  "trained_tokens": 5768,
  "integrations": [],
  "seed": 0,
  "estimated_finish": 0
}

The fine-tuning job event object

Fine-tuning job event objectsource

stringsource

created_at

integersource

level

stringsource

message

stringsource

object

stringsource

OBJECT The fine-tuning job event object

1
2
3
4
5
6
7
{
  "object": "fine_tuning.job.event",
  "id": "ftevent-abc123"
  "created_at": 1677610602,
  "level": "info",
  "message": "Created fine-tuning job"
}

The fine-tuning job checkpoint object

The fine_tuning.job.checkpoint object represents a model checkpoint for a fine-tuning job that is ready to use.source

stringsource

The checkpoint identifier, which can be referenced in the API endpoints.source

created_at

integersource

The Unix timestamp (in seconds) for when the checkpoint was created.source

fine_tuned_model_checkpoint

stringsource

The name of the fine-tuned checkpoint model that is created.source

step_number

integersource

The step number that the checkpoint was created at.source

metrics

objectsource

Metrics at the step number during the fine-tuning job.source

fine_tuning_job_id

stringsource

The name of the fine-tuning job that this checkpoint was created from.source

object

stringsource

The object type, which is always "fine_tuning.job.checkpoint".source

OBJECT The fine-tuning job checkpoint object

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
{
  "object": "fine_tuning.job.checkpoint",
  "id": "ftckpt_qtZ5Gyk4BLq1SfLFWp3RtO3P",
  "created_at": 1712211699,
  "fine_tuned_model_checkpoint": "ft:gpt-4o-mini-2024-07-18:my-org:custom_suffix:9ABel2dg:ckpt-step-88",
  "fine_tuning_job_id": "ftjob-fpbNQ3H1GrMehXRf8cO97xTN",
  "metrics": {
    "step": 88,
    "train_loss": 0.478,
    "train_mean_token_accuracy": 0.924,
    "valid_loss": 10.112,
    "valid_mean_token_accuracy": 0.145,
    "full_valid_loss": 0.567,
    "full_valid_mean_token_accuracy": 0.944
  },
  "step_number": 88
}