Skip to content

Commit c751a01

Browse files
Google APIscopybara-github
authored andcommitted
feat: Expose FullFineTunedResources for full fine tuned deployments
PiperOrigin-RevId: 839371231
1 parent 9ae82e1 commit c751a01

File tree

2 files changed

+40
-0
lines changed

2 files changed

+40
-0
lines changed

google/cloud/aiplatform/v1beta1/endpoint.proto

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -225,6 +225,10 @@ message DeployedModel {
225225
string shared_resources = 17 [(google.api.resource_reference) = {
226226
type: "aiplatform.googleapis.com/DeploymentResourcePool"
227227
}];
228+
229+
// Optional. Resources for a full fine tuned model.
230+
FullFineTunedResources full_fine_tuned_resources = 36
231+
[(google.api.field_behavior) = OPTIONAL];
228232
}
229233

230234
// Immutable. The ID of the DeployedModel. If not provided upon deployment,

google/cloud/aiplatform/v1beta1/machine_resources.proto

Lines changed: 36 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -270,6 +270,42 @@ message BatchDedicatedResources {
270270
bool spot = 5 [(google.api.field_behavior) = OPTIONAL];
271271
}
272272

273+
// Resources for an fft model.
274+
message FullFineTunedResources {
275+
// The type of deployment.
276+
enum DeploymentType {
277+
// Unspecified deployment type.
278+
DEPLOYMENT_TYPE_UNSPECIFIED = 0;
279+
280+
// Eval deployment type.
281+
DEPLOYMENT_TYPE_EVAL = 1;
282+
283+
// Prod deployment type.
284+
DEPLOYMENT_TYPE_PROD = 2;
285+
}
286+
287+
// Required. The kind of deployment.
288+
DeploymentType deployment_type = 1 [(google.api.field_behavior) = REQUIRED];
289+
290+
// Optional. The number of model inference units to use for this deployment.
291+
// This can only be specified for DEPLOYMENT_TYPE_PROD.
292+
// The following table lists the number of model inference units for different
293+
// model types:
294+
// * Gemini 2.5 Flash
295+
// * Foundation FMIU: 25
296+
// * Expansion FMIU: 4
297+
// * Gemini 2.5 Pro
298+
// * Foundation FMIU: 32
299+
// * Expansion FMIU: 16
300+
// * Veo 3.0 (undistilled)
301+
// * Foundation FMIU: 63
302+
// * Expansion FMIU: 7
303+
// * Veo 3.0 (distilled)
304+
// * Foundation FMIU: 30
305+
// * Expansion FMIU: 10
306+
int32 model_inference_unit_count = 2 [(google.api.field_behavior) = OPTIONAL];
307+
}
308+
273309
// Statistics information about resource consumption.
274310
message ResourcesConsumed {
275311
// Output only. The number of replica hours used. Note that many replicas may

0 commit comments

Comments
 (0)