---
title: "Trains and deploys a fine-tuned model."
method: POST
path: "/v1/finetuning/finetuned-models"
tags: ["/finetuning"]
deprecated: true
---

# Trains and deploys a fine-tuned model.

`POST /v1/finetuning/finetuned-models`

> **Deprecated.**

Creates a new fine-tuned model. The model will be trained on the dataset specified in the request body. The training process may take some time, and the model will be available once the training is complete.

## Headers

- `X-Client-Name` string

## Request body

- FinetunedModel — This resource represents a fine-tuned model.
  - `id` string — read-only. FinetunedModel ID.
  - `name` string, required — FinetunedModel name (e.g. `foobar`).
  - `creator_id` string — read-only. User ID of the creator.
  - `organization_id` string — read-only. Organization ID.
  - `settings` Settings, required — The configuration used for fine-tuning.
    - `base_model` BaseModel, required — The base model used for fine-tuning.
      - `name` string — The name of the base model.
      - `version` string — read-only. The version of the base model.
      - `base_type` 'BASE_TYPE_UNSPECIFIED' | 'BASE_TYPE_GENERATIVE' | 'BASE_TYPE_CLASSIFICATION' | 'BASE_TYPE_RERANK' | 'BASE_TYPE_CHAT', required — The possible types of fine-tuned models. - BASE_TYPE_UNSPECIFIED: Unspecified model. - BASE_TYPE_GENERATIVE: Deprecated: Generative model. - BASE_TYPE_CLASSIFICATION: Classification model. - BASE_TYPE_RERANK: Rerank model. - BASE_TYPE_CHAT: Chat model.
      - `strategy` 'STRATEGY_UNSPECIFIED' | 'STRATEGY_VANILLA' | 'STRATEGY_TFEW' — The possible strategy used to serve a fine-tuned models. - STRATEGY_UNSPECIFIED: Unspecified strategy. - STRATEGY_VANILLA: Deprecated: Serve the fine-tuned model on a dedicated GPU. - STRATEGY_TFEW: Deprecated: Serve the fine-tuned model on a shared GPU.
    - `dataset_id` string, required — The data used for training and evaluating the fine-tuned model.
    - `hyperparameters` Hyperparameters — The fine-tuning hyperparameters.
      - `early_stopping_patience` integer — Stops training if the loss metric does not improve beyond the value of `early_stopping_threshold` after this many times of evaluation.
      - `early_stopping_threshold` number, double — How much the loss must improve to prevent early stopping.
      - `train_batch_size` integer — The batch size is the number of training examples included in a single training pass.
      - `train_epochs` integer — The number of epochs to train for.
      - `learning_rate` number, double — The learning rate to be used during training.
      - `lora_alpha` integer — Controls the scaling factor for LoRA updates. Higher values make the updates more impactful.
      - `lora_rank` integer — Specifies the rank for low-rank matrices. Lower ranks reduce parameters but may limit model flexibility.
      - `lora_target_modules` 'LORA_TARGET_MODULES_UNSPECIFIED' | 'LORA_TARGET_MODULES_QV' | 'LORA_TARGET_MODULES_QKVO' | 'LORA_TARGET_MODULES_QKVO_FFN' — The possible combinations of LoRA modules to target. - LORA_TARGET_MODULES_UNSPECIFIED: Unspecified LoRA target modules. - LORA_TARGET_MODULES_QV: LoRA adapts the query and value matrices in transformer attention layers. - LORA_TARGET_MODULES_QKVO: LoRA adapts query, key, value, and output matrices in attention layers. - LORA_TARGET_MODULES_QKVO_FFN: LoRA adapts attention projection matrices and feed-forward networks (FFN).
    - `multi_label` boolean — read-only. Whether the model is single-label or multi-label (only for classification).
    - `wandb` WandbConfig — The Weights & Biases configuration.
      - `project` string, required — The WandB project name to be used during training.
      - `api_key` string, required — The WandB API key to be used during training.
      - `entity` string — The WandB entity name to be used during training.
  - `status` 'STATUS_UNSPECIFIED' | 'STATUS_FINETUNING' | 'STATUS_DEPLOYING_API' | 'STATUS_READY' | 'STATUS_FAILED' | 'STATUS_DELETED' | 'STATUS_TEMPORARILY_OFFLINE' | 'STATUS_PAUSED' | 'STATUS_QUEUED' — The possible stages of a fine-tuned model life-cycle. - STATUS_UNSPECIFIED: Unspecified status. - STATUS_FINETUNING: The fine-tuned model is being fine-tuned. - STATUS_DEPLOYING_API: Deprecated: The fine-tuned model is being deployed. - STATUS_READY: The fine-tuned model is ready to receive requests. - STATUS_FAILED: The fine-tuned model failed. - STATUS_DELETED: The fine-tuned model was deleted. - STATUS_TEMPORARILY_OFFLINE: Deprecated: The fine-tuned model is temporarily unavailable. - STATUS_PAUSED: Deprecated: The fine-tuned model is paused (Vanilla only). - STATUS_QUEUED: The fine-tuned model is queued for training.
  - `created_at` string, date-time — read-only. Creation timestamp.
  - `updated_at` string, date-time — read-only. Latest update timestamp.
  - `completed_at` string, date-time — read-only. Timestamp for the completed fine-tuning.
  - `last_used` string, date-time — read-only. Deprecated: Timestamp for the latest request to this fine-tuned model.

## Response `200`

A successful response.

- CreateFinetunedModelResponse — Response to request to create a fine-tuned model.
  - `finetuned_model` FinetunedModel — This resource represents a fine-tuned model.
    - `id` string — read-only. FinetunedModel ID.
    - `name` string, required — FinetunedModel name (e.g. `foobar`).
    - `creator_id` string — read-only. User ID of the creator.
    - `organization_id` string — read-only. Organization ID.
    - `settings` Settings, required — The configuration used for fine-tuning.
      - `base_model` BaseModel, required — The base model used for fine-tuning.
        - `name` string — The name of the base model.
        - `version` string — read-only. The version of the base model.
        - `base_type` 'BASE_TYPE_UNSPECIFIED' | 'BASE_TYPE_GENERATIVE' | 'BASE_TYPE_CLASSIFICATION' | 'BASE_TYPE_RERANK' | 'BASE_TYPE_CHAT', required — The possible types of fine-tuned models. - BASE_TYPE_UNSPECIFIED: Unspecified model. - BASE_TYPE_GENERATIVE: Deprecated: Generative model. - BASE_TYPE_CLASSIFICATION: Classification model. - BASE_TYPE_RERANK: Rerank model. - BASE_TYPE_CHAT: Chat model.
        - `strategy` 'STRATEGY_UNSPECIFIED' | 'STRATEGY_VANILLA' | 'STRATEGY_TFEW' — The possible strategy used to serve a fine-tuned models. - STRATEGY_UNSPECIFIED: Unspecified strategy. - STRATEGY_VANILLA: Deprecated: Serve the fine-tuned model on a dedicated GPU. - STRATEGY_TFEW: Deprecated: Serve the fine-tuned model on a shared GPU.
      - `dataset_id` string, required — The data used for training and evaluating the fine-tuned model.
      - `hyperparameters` Hyperparameters — The fine-tuning hyperparameters.
        - `early_stopping_patience` integer — Stops training if the loss metric does not improve beyond the value of `early_stopping_threshold` after this many times of evaluation.
        - `early_stopping_threshold` number, double — How much the loss must improve to prevent early stopping.
        - `train_batch_size` integer — The batch size is the number of training examples included in a single training pass.
        - `train_epochs` integer — The number of epochs to train for.
        - `learning_rate` number, double — The learning rate to be used during training.
        - `lora_alpha` integer — Controls the scaling factor for LoRA updates. Higher values make the updates more impactful.
        - `lora_rank` integer — Specifies the rank for low-rank matrices. Lower ranks reduce parameters but may limit model flexibility.
        - `lora_target_modules` 'LORA_TARGET_MODULES_UNSPECIFIED' | 'LORA_TARGET_MODULES_QV' | 'LORA_TARGET_MODULES_QKVO' | 'LORA_TARGET_MODULES_QKVO_FFN' — The possible combinations of LoRA modules to target. - LORA_TARGET_MODULES_UNSPECIFIED: Unspecified LoRA target modules. - LORA_TARGET_MODULES_QV: LoRA adapts the query and value matrices in transformer attention layers. - LORA_TARGET_MODULES_QKVO: LoRA adapts query, key, value, and output matrices in attention layers. - LORA_TARGET_MODULES_QKVO_FFN: LoRA adapts attention projection matrices and feed-forward networks (FFN).
      - `multi_label` boolean — read-only. Whether the model is single-label or multi-label (only for classification).
      - `wandb` WandbConfig — The Weights & Biases configuration.
        - `project` string, required — The WandB project name to be used during training.
        - `api_key` string, required — The WandB API key to be used during training.
        - `entity` string — The WandB entity name to be used during training.
    - `status` 'STATUS_UNSPECIFIED' | 'STATUS_FINETUNING' | 'STATUS_DEPLOYING_API' | 'STATUS_READY' | 'STATUS_FAILED' | 'STATUS_DELETED' | 'STATUS_TEMPORARILY_OFFLINE' | 'STATUS_PAUSED' | 'STATUS_QUEUED' — The possible stages of a fine-tuned model life-cycle. - STATUS_UNSPECIFIED: Unspecified status. - STATUS_FINETUNING: The fine-tuned model is being fine-tuned. - STATUS_DEPLOYING_API: Deprecated: The fine-tuned model is being deployed. - STATUS_READY: The fine-tuned model is ready to receive requests. - STATUS_FAILED: The fine-tuned model failed. - STATUS_DELETED: The fine-tuned model was deleted. - STATUS_TEMPORARILY_OFFLINE: Deprecated: The fine-tuned model is temporarily unavailable. - STATUS_PAUSED: Deprecated: The fine-tuned model is paused (Vanilla only). - STATUS_QUEUED: The fine-tuned model is queued for training.
    - `created_at` string, date-time — read-only. Creation timestamp.
    - `updated_at` string, date-time — read-only. Latest update timestamp.
    - `completed_at` string, date-time — read-only. Timestamp for the completed fine-tuning.
    - `last_used` string, date-time — read-only. Deprecated: Timestamp for the latest request to this fine-tuned model.

## Other responses

- `400` — Bad Request
- `401` — Unauthorized
- `403` — Forbidden
- `404` — Not Found
- `500` — Internal Server Error
- `503` — Status Service Unavailable

---

[API](https://skmtc.net/cohere-ai/apis/api-reference.md) · [All operations](https://skmtc.net/cohere-ai/apis/api-reference/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/cohere-ai/api-reference/revisions/dbed69eda210/schema)