Skip to content

DONT MERGE: Spot jobs #140

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
stranavad wants to merge 1 commit into
base: dev
Choose a base branch
from
Open

DONT MERGE: Spot jobs #140

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions apps/api/src/modules/training-external/schema/training-external.schema.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,10 +131,16 @@ export const trainingOutputUploadSchema = z.object({
object_path: z.string(),
});

export const trainingCheckpointConfigSchema = z.object({
put_url: z.string(),
get_url: z.string(),
});

const basePayload = z.object({
id: z.number(),
training_config: trainingConfigSchema,
output_config: trainingOutputUploadSchema.array(),
checkpoint_config: trainingCheckpointConfigSchema.optional(),
});

const createDataSchema = <T extends z.ZodTypeAny>(labelSchema: T) =>
Expand Down
36 changes: 35 additions & 1 deletion apps/api/src/modules/training-external/services/training-external.service.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -324,6 +324,7 @@ export class TrainingExternalService {
const instancePolicy: protos.google.cloud.batch.v1.AllocationPolicy.IInstancePolicy = {
machineType: mlBatchMachineType,
bootDisk: { sizeGb: String(mlBatchBootDiskGb) },
provisioningModel: "SPOT",
};
if (mlBatchGpuType && mlBatchGpuCount > 0) {
instancePolicy.accelerators = [
Expand Down Expand Up @@ -380,6 +381,15 @@ export class TrainingExternalService {
memoryMib: mlBatchTaskMemoryMib,
},
maxRunDuration: { seconds: String(mlBatchMaxRunSeconds) },
maxRetryCount: 3,
lifecyclePolicies: [
{
action: protos.google.cloud.batch.v1.LifecyclePolicy.Action.RETRY_TASK,
actionCondition: {
exitCodes: [50001],
},
},
],
},
},
],
Expand Down Expand Up @@ -464,16 +474,40 @@ export class TrainingExternalService {
}
})

const isUltralytics = model.backend === ModelBackendEnum.ULTRALYTICS;

const checkpointObjectPath = `${model.projectId}/model/${model.id}/checkpoint_last.pt`;
const [checkpointPutUrl, checkpointGetUrl] = isUltralytics
? await Promise.all([
this.assetsService.generateSignedUploadUrl(
checkpointObjectPath,
"application/octet-stream",
UPLOAD_URL_TTL_MS,
),
this.assetsService.generateSignedDownloadUrl(
checkpointObjectPath,
UPLOAD_URL_TTL_MS,
),
])
: [undefined, undefined];

const basePayload: TrainingBasePayload = {
id: model.id,
output_config: outputUploads,
...(checkpointPutUrl &&
checkpointGetUrl && {
checkpoint_config: {
put_url: checkpointPutUrl,
get_url: checkpointGetUrl,
},
}),
training_config: {
output_types: model.outputTypes,
epochs: model.epochs,
// ml-yolo consumes this to flip HubAI's quantization_mode between
// FP16_STANDARD and INT8_STANDARD. Omit for Luxonis — its Pydantic
// model rejects unknown keys (extra="forbid").
...(model.backend === ModelBackendEnum.ULTRALYTICS && {
...(isUltralytics && {
quantization: model.quantization,
}),
dataset_config: {
Expand Down
Loading