Add file and fine-tune api support (#6)

api/src/main/java/com/theokanning/openai/DeleteResult.java

@@ -0,0 +1,24 @@
+package com.theokanning.openai;
+
+import lombok.Data;
+
+/**
+ * A response when deleting an object
+ */
+@Data
+public class DeleteResult {
+    /**
+     * The id of the object.
+     */
+    String id;
+
+    /**
+     * The type of object deleted, for example "file" or "model"
+     */
+    String object;
+
+    /**
+     * True if successfully deleted
+     */
+    boolean deleted;
+}

api/src/main/java/com/theokanning/openai/completion/CompletionRequest.java

@@ -20,6 +20,11 @@ import java.util.List;
 @Data
 public class CompletionRequest {
 
+    /**
+     * The name of the model to use, only used if specifying a fine tuned model.
+     */
+    String model;
+
     /**
      * An optional prompt to complete from
      */

api/src/main/java/com/theokanning/openai/completion/CompletionResult.java

@@ -22,7 +22,7 @@ public class CompletionResult {
     String object;
 
     /**
-     * The creation time in epoch milliseconds.
+     * The creation time in epoch seconds.
      */
     long created;
 

api/src/main/java/com/theokanning/openai/file/File.java

@@ -0,0 +1,42 @@
+package com.theokanning.openai.file;
+
+import lombok.Data;
+
+/**
+ * A file uploaded to OpenAi
+ *
+ * https://beta.openai.com/docs/api-reference/files
+ */
+@Data
+public class File {
+
+    /**
+     * The unique id of this file.
+     */
+    String id;
+
+    /**
+     * The type of object returned, should be "file".
+     */
+    String object;
+
+    /**
+     * File size in bytes.
+     */
+    Long bytes;
+
+    /**
+     * The creation time in epoch seconds.
+     */
+    Long createdAt;
+
+    /**
+     * The name of the file.
+     */
+    String filename;
+
+    /**
+     * Description of the file's purpose.
+     */
+    String purpose;
+}

api/src/main/java/com/theokanning/openai/finetune/FineTuneEvent.java

@@ -0,0 +1,31 @@
+package com.theokanning.openai.finetune;
+
+import lombok.Data;
+
+/**
+ * An object representing an event in the lifecycle of a fine-tuning job
+ *
+ * https://beta.openai.com/docs/api-reference/fine-tunes
+ */
+@Data
+public class FineTuneEvent {
+    /**
+     * The type of object returned, should be "fine-tune-event".
+     */
+    String object;
+
+    /**
+     * The creation time in epoch seconds.
+     */
+    Long createdAt;
+
+    /**
+     * The log level of this message.
+     */
+    String level;
+
+    /**
+     * The event message.
+     */
+    String message;
+}

api/src/main/java/com/theokanning/openai/finetune/FineTuneRequest.java

@@ -0,0 +1,111 @@
+package com.theokanning.openai.finetune;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+import java.util.List;
+
+/**
+ * A request for OpenAi to create a fine-tuned model
+ * All fields except trainingFile are nullable.
+ *
+ * https://beta.openai.com/docs/api-reference/fine-tunes/create
+ */
+@Builder
+@NoArgsConstructor
+@AllArgsConstructor
+@Data
+public class FineTuneRequest {
+
+    /**
+     * The ID of an uploaded file that contains training data.
+     */
+    String trainingFile;
+
+    /**
+     * The ID of an uploaded file that contains validation data.
+     */
+    String validationFile;
+
+    /**
+     * The name of the base model to fine-tune. You can select one of "ada", "babbage", "curie", or "davinci".
+     * To learn more about these models, see the Engines documentation.
+     */
+    String model;
+
+    /**
+     * The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.
+     */
+    Integer nEpochs;
+
+    /**
+     * The batch size to use for training.
+     * The batch size is the number of training examples used to train a single forward and backward pass.
+     *
+     * By default, the batch size will be dynamically configured to be ~0.2% of the number of examples in the training
+     * set, capped at 256 - in general, we've found that larger batch sizes tend to work better for larger datasets.
+     */
+    Integer batchSize;
+
+    /**
+     * The learning rate multiplier to use for training.
+     * The fine-tuning learning rate is the original learning rate used for pretraining multiplied by this value.
+     *
+     * By default, the learning rate multiplier is the 0.05, 0.1, or 0.2 depending on final batch_size
+     * (larger learning rates tend to perform better with larger batch sizes).
+     * We recommend experimenting with values in the range 0.02 to 0.2 to see what produces the best results.
+     */
+    Double learningRateMultiplier;
+
+    /**
+     * The weight to use for loss on the prompt tokens.
+     * This controls how much the model tries to learn to generate the prompt
+     * (as compared to the completion which always has a weight of 1.0),
+     * and can add a stabilizing effect to training when completions are short.
+     *
+     * If prompts are extremely long (relative to completions), it may make sense to reduce this weight so as to
+     * avoid over-prioritizing learning the prompt.
+     */
+    Double promptLossWeight;
+
+    /**
+     * If set, we calculate classification-specific metrics such as accuracy and F-1 score using the validation set
+     * at the end of every epoch. These metrics can be viewed in the results file.
+     *
+     * In order to compute classification metrics, you must provide a validation_file.
+     * Additionally, you must specify {@link FineTuneRequest#classificationNClasses} for multiclass
+     * classification or {@link FineTuneRequest#classificationPositiveClass} for binary classification.
+     */
+    Boolean computeClassificationMetrics;
+
+    /**
+     * The number of classes in a classification task.
+     *
+     * This parameter is required for multiclass classification.
+     */
+    Integer classificationNClasses; // todo verify snake case
+
+    /**
+     * The positive class in binary classification.
+     *
+     * This parameter is needed to generate precision, recall, and F1 metrics when doing binary classification.
+     */
+    String classificationPositiveClass;
+
+    /**
+     * If this is provided, we calculate F-beta scores at the specified beta values.
+     * The F-beta score is a generalization of F-1 score. This is only used for binary classification.
+     *
+     * With a beta of 1 (i.e. the F-1 score), precision and recall are given the same weight.
+     * A larger beta score puts more weight on recall and less on precision.
+     * A smaller beta score puts more weight on precision and less on recall.
+     */
+    List<Double> classificationBetas;
+
+    /**
+     * A string of up to 40 characters that will be added to your fine-tuned model name.
+     */
+    String suffix;
+}

api/src/main/java/com/theokanning/openai/finetune/FineTuneResult.java

@@ -0,0 +1,80 @@
+package com.theokanning.openai.finetune;
+
+import com.theokanning.openai.file.File;
+import lombok.Data;
+
+import java.util.List;
+
+/**
+ * An object describing an fine-tuned model. Returned by multiple fine-tune requests.
+ *
+ * https://beta.openai.com/docs/api-reference/fine-tunes
+ */
+@Data
+public class FineTuneResult {
+    /**
+     * The ID of the fine-tuning job.
+     */
+    String id;
+
+    /**
+     * The type of object returned, should be "fine-tune".
+     */
+    String object;
+
+    /**
+     * The name of the base model.
+     */
+    String model;
+
+    /**
+     * The creation time in epoch seconds.
+     */
+    Long createdAt;
+
+    /**
+     * List of events in this job's lifecycle. Null when getting a list of fine-tune jobs.
+     */
+    List<FineTuneEvent> events;
+
+    /**
+     * The ID of the fine-tuned model, null if tuning job is not finished.
+     * This is the id used to call the model.
+     */
+    String fineTunedModel;
+
+    /**
+     * The specified hyper-parameters for the tuning job.
+     */
+    HyperParameters hyperparams;
+
+    /**
+     * The ID of the organization this model belongs to.
+     */
+    String organizationId;
+
+    /**
+     * Result files for this fine-tune job.
+     */
+    List<File> resultFiles;
+
+    /**
+     * The status os the fine-tune job. "pending", "succeeded", or "cancelled"
+     */
+    String status;
+
+    /**
+     * Training files for this fine-tune job.
+     */
+    List<File> trainingFiles;
+
+    /**
+     * The last update time in epoch seconds.
+     */
+    Long updatedAt;
+
+    /**
+     * Validation files for this fine-tune job.
+     */
+    List<File> validationFiles;
+}

api/src/main/java/com/theokanning/openai/finetune/HyperParameters.java

@@ -0,0 +1,32 @@
+package com.theokanning.openai.finetune;
+
+import lombok.Data;
+
+/**
+ * Fine-tuning job hyperparameters
+ *
+ * https://beta.openai.com/docs/api-reference/fine-tunes
+ */
+@Data
+public class HyperParameters {
+
+    /**
+     * The batch size to use for training.
+     */
+    String batchSize;
+
+    /**
+     * The learning rate multiplier to use for training.
+     */
+    Double learningRateMultiplier;
+
+    /**
+     * The number of epochs to train the model for.
+     */
+    Integer nEpochs;
+
+    /**
+     * The weight to use for loss on the prompt tokens.
+     */
+    Double promptLossWeight;
+}

client/build.gradle

@@ -7,6 +7,16 @@ dependencies {
     api 'com.squareup.retrofit2:retrofit:2.9.0'
     implementation 'com.squareup.retrofit2:adapter-rxjava2:2.9.0'
     implementation 'com.squareup.retrofit2:converter-jackson:2.9.0'
+
+    testImplementation(platform('org.junit:junit-bom:5.8.2'))
+    testImplementation('org.junit.jupiter:junit-jupiter')
+}
+
+test {
+    useJUnitPlatform()
+    testLogging {
+        events "passed", "skipped", "failed"
+    }
 }
 
 ext {

client/src/main/java/com/theokanning/openai/OpenAiApi.java

@@ -1,15 +1,18 @@
 package com.theokanning.openai;
 
 import com.theokanning.openai.engine.Engine;
+import com.theokanning.openai.file.File;
+import com.theokanning.openai.finetune.FineTuneRequest;
+import com.theokanning.openai.finetune.FineTuneEvent;
+import com.theokanning.openai.finetune.FineTuneResult;
 import com.theokanning.openai.search.SearchRequest;
 import io.reactivex.Single;
 import com.theokanning.openai.completion.CompletionRequest;
 import com.theokanning.openai.completion.CompletionResult;
 import com.theokanning.openai.search.SearchResult;
-import retrofit2.http.Body;
+import okhttp3.MultipartBody;
-import retrofit2.http.GET;
+import okhttp3.RequestBody;
-import retrofit2.http.POST;
+import retrofit2.http.*;
-import retrofit2.http.Path;
 
 public interface OpenAiApi {
 
@@ -24,4 +27,39 @@ public interface OpenAiApi {
 
     @POST("/v1/engines/{engine_id}/search")
     Single<OpenAiResponse<SearchResult>> search(@Path("engine_id") String engineId, @Body SearchRequest request);
+
+    @GET("/v1/files")
+    Single<OpenAiResponse<File>> listFiles();
+
+    @Multipart
+    @POST("/v1/files")
+    Single<File> uploadFile(@Part("purpose") RequestBody purpose, @Part MultipartBody.Part file);
+
+    @DELETE("/v1/files/{file_id}")
+    Single<DeleteResult> deleteFile(@Path("file_id") String fileId);
+
+    @GET("/v1/files/{file_id}")
+    Single<File> retrieveFile(@Path("file_id") String fileId);
+
+    @POST("/v1/fine-tunes")
+    Single<FineTuneResult> createFineTune(@Body FineTuneRequest request);
+
+    @POST("/v1/completions")
+    Single<CompletionResult> createFineTuneCompletion(@Body CompletionRequest request);
+
+    @GET("/v1/fine-tunes")
+    Single<OpenAiResponse<FineTuneResult>> listFineTunes();
+
+    @GET("/v1/fine-tunes/{fine_tune_id}")
+    Single<FineTuneResult> retrieveFineTune(@Path("fine_tune_id") String fineTuneId);
+
+    @POST("/v1/fine-tunes/{fine_tune_id}/cancel")
+    Single<FineTuneResult> cancelFineTune(@Path("fine_tune_id") String fineTuneId);
+
+    @GET("/v1/fine-tunes/{fine_tune_id}/events")
+    Single<OpenAiResponse<FineTuneEvent>> listFineTuneEvents(@Path("fine_tune_id") String fineTuneId);
+
+    @DELETE("/v1/models/{fine_tune_id}")
+    Single<DeleteResult> deleteFineTune(@Path("fine_tune_id") String fineTuneId);
+
 }

client/src/main/java/com/theokanning/openai/OpenAiService.java

@@ -4,9 +4,12 @@ import com.fasterxml.jackson.annotation.JsonInclude;
 import com.fasterxml.jackson.databind.DeserializationFeature;
 import com.fasterxml.jackson.databind.ObjectMapper;
 import com.fasterxml.jackson.databind.PropertyNamingStrategy;
+import com.theokanning.openai.file.File;
+import com.theokanning.openai.finetune.FineTuneRequest;
+import com.theokanning.openai.finetune.FineTuneEvent;
+import com.theokanning.openai.finetune.FineTuneResult;
 import com.theokanning.openai.search.SearchRequest;
-import okhttp3.ConnectionPool;
+import okhttp3.*;
-import okhttp3.OkHttpClient;
 import com.theokanning.openai.completion.CompletionRequest;
 import com.theokanning.openai.completion.CompletionResult;
 import com.theokanning.openai.engine.Engine;
@@ -58,4 +61,53 @@ public class OpenAiService {
     public List<SearchResult> search(String engineId, SearchRequest request) {
         return api.search(engineId, request).blockingGet().data;
     }
+
+    public List<File> listFiles() {
+        return api.listFiles().blockingGet().data;
+    }
+
+    public File uploadFile(String purpose, String filepath) {
+        java.io.File file = new java.io.File(filepath);
+        RequestBody purposeBody = RequestBody.create(okhttp3.MultipartBody.FORM, purpose);
+        RequestBody fileBody = RequestBody.create(MediaType.parse("text"), file);
+        MultipartBody.Part body = MultipartBody.Part.createFormData("file", filepath, fileBody);
+
+        return api.uploadFile(purposeBody, body).blockingGet();
+    }
+
+    public DeleteResult deleteFile(String fileId) {
+        return api.deleteFile(fileId).blockingGet();
+    }
+
+    public File retrieveFile(String fileId) {
+        return api.retrieveFile(fileId).blockingGet();
+    }
+
+    public FineTuneResult createFineTune(FineTuneRequest request) {
+        return api.createFineTune(request).blockingGet();
+    }
+
+    public CompletionResult createFineTuneCompletion(CompletionRequest request) {
+        return api.createFineTuneCompletion(request).blockingGet();
+    }
+
+    public List<FineTuneResult> listFineTunes() {
+        return api.listFineTunes().blockingGet().data;
+    }
+
+    public FineTuneResult retrieveFineTune(String fineTuneId) {
+        return api.retrieveFineTune(fineTuneId).blockingGet();
+    }
+
+    public FineTuneResult cancelFineTune(String fineTuneId) {
+        return api.cancelFineTune(fineTuneId).blockingGet();
+    }
+
+    public List<FineTuneEvent> listFineTuneEvents(String fineTuneId) {
+        return api.listFineTuneEvents(fineTuneId).blockingGet().data;
+    }
+
+    public DeleteResult deleteFineTune(String fineTuneId) {
+        return api.deleteFineTune(fineTuneId).blockingGet();
+    }
 }

client/src/test/java/com/theokanning/openai/FileTest.java

@@ -0,0 +1,55 @@
+package com.theokanning.openai;
+
+import com.theokanning.openai.file.File;
+import org.junit.jupiter.api.*;
+
+import java.util.List;
+import java.util.concurrent.TimeUnit;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+@TestMethodOrder(MethodOrderer.OrderAnnotation.class)
+public class FileTest {
+    static String filePath = "src/test/resources/fine-tuning-data.jsonl";
+
+    String token = System.getenv("OPENAI_TOKEN");
+    OpenAiService service = new OpenAiService(token);
+    static String fileId;
+
+    @Test
+    @Order(1)
+    void uploadFile() throws Exception {
+        File file = service.uploadFile("fine-tune", filePath);
+        fileId = file.getId();
+
+        assertEquals("fine-tune", file.getPurpose());
+        assertEquals(filePath, file.getFilename());
+
+        // wait for file to be processed
+        TimeUnit.SECONDS.sleep(10);
+    }
+
+    @Test
+    @Order(2)
+    void listFiles() {
+        List<File> files = service.listFiles();
+
+        assertTrue(files.stream().anyMatch(file -> file.getId().equals(fileId)));
+    }
+
+    @Test
+    @Order(3)
+    void retrieveFile() {
+        File file = service.retrieveFile(fileId);
+
+        assertEquals(filePath, file.getFilename());
+    }
+
+    @Test
+    @Order(4)
+    void deleteFile() {
+        DeleteResult result = service.deleteFile(fileId);
+        assertTrue(result.isDeleted());
+    }
+}

client/src/test/java/com/theokanning/openai/FineTuneTest.java

@@ -0,0 +1,76 @@
+package com.theokanning.openai;
+
+import com.theokanning.openai.finetune.FineTuneRequest;
+import com.theokanning.openai.finetune.FineTuneEvent;
+import com.theokanning.openai.finetune.FineTuneResult;
+import org.junit.jupiter.api.*;
+
+import java.util.List;
+
+import static org.junit.jupiter.api.Assertions.*;
+
+@TestMethodOrder(MethodOrderer.OrderAnnotation.class)
+public class FineTuneTest {
+    static OpenAiService service;
+    static String fileId;
+    static String fineTuneId;
+
+
+    @BeforeAll
+   static void setup() {
+        String token = System.getenv("OPENAI_TOKEN");
+        service = new OpenAiService(token);
+        fileId = service.uploadFile("fine-tune", "src/test/resources/fine-tuning-data.jsonl").getId();
+    }
+
+    @AfterAll
+    static void teardown() {
+        service.deleteFile(fileId);
+    }
+
+    @Test
+    @Order(1)
+    void createFineTune() {
+        FineTuneRequest request = FineTuneRequest.builder()
+                .trainingFile(fileId)
+                .model("ada")
+                .build();
+
+        FineTuneResult fineTune = service.createFineTune(request);
+        fineTuneId = fineTune.getId();
+
+        assertEquals("pending", fineTune.getStatus());
+    }
+
+    @Test
+    @Order(2)
+    void listFineTunes() {
+        List<FineTuneResult> fineTunes = service.listFineTunes();
+
+        assertTrue(fineTunes.stream().anyMatch(fineTune -> fineTune.getId().equals(fineTuneId)));
+    }
+
+    @Test
+    @Order(3)
+    void listFineTuneEvents() {
+        List<FineTuneEvent> events = service.listFineTuneEvents(fineTuneId);
+
+        assertFalse(events.isEmpty());
+    }
+
+    @Test
+    @Order(3)
+    void retrieveFineTune() {
+        FineTuneResult fineTune = service.retrieveFineTune(fineTuneId);
+
+        assertEquals("ada", fineTune.getModel());
+    }
+
+    @Test
+    @Order(4)
+    void cancelFineTune() {
+        FineTuneResult fineTune = service.cancelFineTune(fineTuneId);
+
+        assertEquals("cancelled", fineTune.getStatus());
+    }
+}

client/src/test/resources/fine-tuning-data.jsonl

@@ -0,0 +1,2 @@
+{"prompt": "prompt", "completion": "text"}
+{"prompt": "prompt", "completion": "text"}