ai.z.openapi.api.fine_tuning

Interface FineTuningApi

public interface FineTuningApi

Fine-tuning API for advanced model customization Enables training custom models based on GLM-4, CodeGeeX-4, and other foundation models Supports domain-specific fine-tuning for chat, code generation, embedding, and specialized tasks Features LoRA (Low-Rank Adaptation) and full parameter fine-tuning with efficient data processing Optimized for Chinese and multilingual scenarios with robust training pipeline

Method Summary

Modifier and Type	Method and Description
io.reactivex.Single<FineTuningJob>	cancelFineTuningJob(String fineTuningJobId) Cancel a running fine-tuning job Stops the training process and marks the job as cancelled
io.reactivex.Single<FineTuningJob>	createFineTuningJob(FineTuningJobRequest request) Create a new fine-tuning job for model customization Initiates the training process for customizing GLM-4, CodeGeeX-4, or other base models Supports various fine-tuning strategies including LoRA, full parameter tuning, and adapter methods Enables domain adaptation for specific use cases like coding, conversation, or specialized knowledge
io.reactivex.Single<FineTuningJob>	deleteFineTuningJob(String fineTuningJobId) Delete a fine-tuning job Permanently removes the fine-tuning job and associated metadata
io.reactivex.Single<FineTunedModelsStatus>	deleteFineTuningModel(String fineTunedModel) Delete a fine-tuned model Permanently removes the custom model created from fine-tuning
io.reactivex.Single<FineTuningEvent>	listFineTuningJobEvents(String fineTuningJobId, Integer limit, String after) List events for a specific fine-tuning job with detailed monitoring Retrieves comprehensive training progress including loss metrics, validation scores, and status updates Provides real-time monitoring of training convergence, learning rate schedules, and performance indicators Essential for tracking model quality and identifying potential training issues
io.reactivex.Single<PersonalFineTuningJob>	queryPersonalFineTuningJobs(Integer limit, String after) Query all personal fine-tuning jobs Lists all fine-tuning jobs created by the current user with pagination support
io.reactivex.Single<FineTuningJob>	retrieveFineTuningJob(String fineTuningJobId, Integer limit, String after) Retrieve comprehensive details of a specific fine-tuning job Gets complete information about job status, training progress, model performance, and configuration Includes training metrics, validation results, estimated completion time, and resource usage Provides insights into model convergence and fine-tuning effectiveness

Method Detail

createFineTuningJob

@POST(value="fine_tuning/jobs")
io.reactivex.Single<FineTuningJob> createFineTuningJob(@Body
                                                                                       FineTuningJobRequest request)

Create a new fine-tuning job for model customization Initiates the training process for customizing GLM-4, CodeGeeX-4, or other base models Supports various fine-tuning strategies including LoRA, full parameter tuning, and adapter methods Enables domain adaptation for specific use cases like coding, conversation, or specialized knowledge

Parameters:

request - Fine-tuning job parameters including base model selection, training dataset, hyperparameters, and optimization settings

Returns:

Fine-tuning job information with unique job ID, status, and initial training configuration details

listFineTuningJobEvents

@GET(value="fine_tuning/jobs/{fine_tuning_job_id}/events")
io.reactivex.Single<FineTuningEvent> listFineTuningJobEvents(@Path(value="fine_tuning_job_id")
                                                                                                                        String fineTuningJobId,
                                                                                                                        @Query(value="limit")
                                                                                                                        Integer limit,
                                                                                                                        @Query(value="after")
                                                                                                                        String after)

List events for a specific fine-tuning job with detailed monitoring Retrieves comprehensive training progress including loss metrics, validation scores, and status updates Provides real-time monitoring of training convergence, learning rate schedules, and performance indicators Essential for tracking model quality and identifying potential training issues

Parameters:

fineTuningJobId - Unique identifier of the fine-tuning job

limit - Maximum number of events to return (recommended: 50-100 for efficient monitoring)

after - Cursor for pagination to get events after this timestamp

Returns:

List of fine-tuning events with timestamps, training metrics, validation results, and status details

retrieveFineTuningJob

@GET(value="fine_tuning/jobs/{fine_tuning_job_id}")
io.reactivex.Single<FineTuningJob> retrieveFineTuningJob(@Path(value="fine_tuning_job_id")
                                                                                                             String fineTuningJobId,
                                                                                                             @Query(value="limit")
                                                                                                             Integer limit,
                                                                                                             @Query(value="after")
                                                                                                             String after)

Retrieve comprehensive details of a specific fine-tuning job Gets complete information about job status, training progress, model performance, and configuration Includes training metrics, validation results, estimated completion time, and resource usage Provides insights into model convergence and fine-tuning effectiveness

Parameters:

fineTuningJobId - Unique identifier of the fine-tuning job

limit - Maximum number of items to return in nested collections (events, checkpoints)

after - Cursor for pagination in nested collections

Returns:

Complete fine-tuning job details including status, metrics, configuration, and performance indicators

queryPersonalFineTuningJobs

@GET(value="fine_tuning/jobs")
io.reactivex.Single<PersonalFineTuningJob> queryPersonalFineTuningJobs(@Query(value="limit")
                                                                                                      Integer limit,
                                                                                                      @Query(value="after")
                                                                                                      String after)

Query all personal fine-tuning jobs Lists all fine-tuning jobs created by the current user with pagination support

Parameters:

limit - Maximum number of jobs to return per page

after - Cursor for pagination to get jobs after this point

Returns:

Paginated list of personal fine-tuning jobs

cancelFineTuningJob

@POST(value="fine_tuning/jobs/{fine_tuning_job_id}/cancel")
io.reactivex.Single<FineTuningJob> cancelFineTuningJob(@Path(value="fine_tuning_job_id")
                                                                                                                   String fineTuningJobId)

Cancel a running fine-tuning job Stops the training process and marks the job as cancelled

Parameters:

fineTuningJobId - Unique identifier of the fine-tuning job to cancel

Returns:

Updated fine-tuning job status after cancellation

deleteFineTuningJob

@DELETE(value="fine_tuning/jobs/{fine_tuning_job_id}")
io.reactivex.Single<FineTuningJob> deleteFineTuningJob(@Path(value="fine_tuning_job_id")
                                                                                                              String fineTuningJobId)

Delete a fine-tuning job Permanently removes the fine-tuning job and associated metadata

Parameters:

fineTuningJobId - Unique identifier of the fine-tuning job to delete

Returns:

Confirmation of job deletion

deleteFineTuningModel

@DELETE(value="fine_tuning/fine_tuned_models/{fine_tuned_model}")
io.reactivex.Single<FineTunedModelsStatus> deleteFineTuningModel(@Path(value="fine_tuned_model")
                                                                                                                                   String fineTunedModel)

Delete a fine-tuned model Permanently removes the custom model created from fine-tuning

Parameters:

fineTunedModel - Identifier of the fine-tuned model to delete

Returns:

Status confirmation of model deletion