Upload training data through Azure Blob Storage

This commit is contained in:
David Corbitt
2023-09-07 05:12:26 -07:00
parent 4fddc80dc5
commit 071ce47411
20 changed files with 1029 additions and 146 deletions

View File

@@ -0,0 +1,23 @@
-- CreateEnum
CREATE TYPE "DatasetFileUploadStatus" AS ENUM ('PENDING', 'DOWNLOADING', 'PROCESSING', 'SAVING', 'COMPLETE', 'ERROR');
-- CreateTable
CREATE TABLE "DatasetFileUpload" (
"id" UUID NOT NULL,
"datasetId" UUID NOT NULL,
"blobName" TEXT NOT NULL,
"fileName" TEXT NOT NULL,
"fileSize" INTEGER NOT NULL,
"progress" INTEGER NOT NULL DEFAULT 0,
"status" "DatasetFileUploadStatus" NOT NULL DEFAULT 'PENDING',
"uploadedAt" TIMESTAMP(3) NOT NULL,
"visible" BOOLEAN NOT NULL DEFAULT true,
"errorMessage" TEXT,
"createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
"updatedAt" TIMESTAMP(3) NOT NULL,
CONSTRAINT "DatasetFileUpload_pkey" PRIMARY KEY ("id")
);
-- AddForeignKey
ALTER TABLE "DatasetFileUpload" ADD CONSTRAINT "DatasetFileUpload_datasetId_fkey" FOREIGN KEY ("datasetId") REFERENCES "Dataset"("id") ON DELETE CASCADE ON UPDATE CASCADE;

View File

@@ -176,12 +176,41 @@ model OutputEvaluation {
@@unique([modelResponseId, evaluationId])
}
enum DatasetFileUploadStatus {
PENDING
DOWNLOADING
PROCESSING
SAVING
COMPLETE
ERROR
}
model DatasetFileUpload {
id String @id @default(uuid()) @db.Uuid
datasetId String @db.Uuid
dataset Dataset @relation(fields: [datasetId], references: [id], onDelete: Cascade)
blobName String
fileName String
fileSize Int
progress Int @default(0) // Percentage
status DatasetFileUploadStatus @default(PENDING)
uploadedAt DateTime
visible Boolean @default(true)
errorMessage String?
createdAt DateTime @default(now())
updatedAt DateTime @updatedAt
}
model Dataset {
id String @id @default(uuid()) @db.Uuid
name String
datasetEntries DatasetEntry[]
fineTunes FineTune[]
datasetFileUploads DatasetFileUpload[]
trainingRatio Float @default(0.8)
projectId String @db.Uuid