API Overview

`class` `ActionsExecuteBatchReq`

Pydantic 필드:

project_id: <class 'str'>
action_ref: <class 'str'>
call_ids: list[str]
wb_user_id: str | None

`class` `ActionsExecuteBatchRes`

`class` `AggregationType`

피드백 및 호출 통계 메트릭이 지원하는 집계 함수입니다.

`class` `AliasesListReq`

Pydantic 필드:

project_id: <class 'str'>
wb_user_id: str | None

`class` `AliasesListRes`

Pydantic 필드:

aliases: list[str]

`class` `AnnotationQueueAddCallsReq`

어노테이션 큐에 call을 일괄 추가하는 요청입니다. 내부 API에서 사용하기 위해 queue_id를 추가해 AnnotationQueueAddCallsBody를 확장합니다. Pydantic 필드:

project_id: <class 'str'>
call_ids: list[str]
display_fields: list[str]
queue_id: <class 'str'>
wb_user_id: str | None

`class` `AnnotationQueueAddCallsRes`

큐에 call을 추가할 때의 응답입니다. Pydantic 필드:

added_count: <class 'int'>
duplicates: <class 'int'>

`class` `AnnotationQueueCreateReq`

어노테이션 큐를 새로 생성하는 요청입니다. Pydantic 필드:

project_id: <class 'str'>
name: <class 'str'>
description: <class 'str'>
scorer_refs: list[str]
wb_user_id: str | None

`class` `AnnotationQueueCreateRes`

어노테이션 큐를 생성할 때의 응답입니다. Pydantic 필드:

id: <class 'str'>

`class` `AnnotationQueueDeleteReq`

어노테이션 큐를 삭제(소프트 삭제)하는 요청입니다. Pydantic 필드:

project_id: <class 'str'>
queue_id: <class 'str'>
wb_user_id: str | None

`class` `AnnotationQueueDeleteRes`

어노테이션 큐 삭제 응답입니다. Pydantic 필드:

queue: <class 'AnnotationQueueSchema'>

`class` `AnnotationQueueItemSchema`

어노테이션 큐 항목 응답 스키마입니다. Pydantic 필드:

id: <class 'str'>
project_id: <class 'str'>
queue_id: <class 'str'>
call_id: <class 'str'>
call_started_at: <class 'datetime.datetime'>
call_ended_at: datetime.datetime | None
call_op_name: <class 'str'>
call_trace_id: <class 'str'>
display_fields: list[str]
added_by: str | None
annotation_state: typing.Literal['unstarted', 'in_progress', 'completed', 'skipped']
annotator_user_id: str | None
created_at: <class 'datetime.datetime'>
created_by: <class 'str'>
updated_at: <class 'datetime.datetime'>
deleted_at: datetime.datetime | None
position_in_queue: int | None

`class` `AnnotationQueueItemsQueryReq`

어노테이션 큐의 항목을 쿼리하기 위한 요청입니다. 내부 API 사용을 위해 queue_id를 추가하여 AnnotationQueueItemsQueryBody를 확장합니다. Pydantic 필드:

project_id: <class 'str'>
filter: weave.trace_server.common_interface.AnnotationQueueItemsFilter | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
limit: int | None
offset: int | None
include_position: <class 'bool'>
queue_id: <class 'str'>

`class` `AnnotationQueueItemsQueryRes`

어노테이션 큐 항목 쿼리에 대한 응답입니다. Pydantic 필드:

items: list[AnnotationQueueItemSchema]

`class` `AnnotationQueueReadReq`

특정 어노테이션 큐를 조회하는 요청입니다. Pydantic 필드:

project_id: <class 'str'>
queue_id: <class 'str'>

`class` `AnnotationQueueReadRes`

어노테이션 큐 조회 응답입니다. Pydantic 필드:

queue: <class 'AnnotationQueueSchema'>

`class` `AnnotationQueueSchema`

어노테이션 큐 응답을 위한 스키마입니다. Pydantic 필드:

id: <class 'str'>
project_id: <class 'str'>
name: <class 'str'>
description: <class 'str'>
scorer_refs: list[str]
created_at: <class 'datetime.datetime'>
created_by: <class 'str'>
updated_at: <class 'datetime.datetime'>
deleted_at: datetime.datetime | None

`class` `AnnotationQueueStatsSchema`

단일 어노테이션 큐에 대한 통계입니다. Pydantic 필드:

queue_id: <class 'str'>
total_items: <class 'int'>
completed_items: <class 'int'>

`class` `AnnotationQueueUpdateReq`

어노테이션 큐를 업데이트하는 요청입니다. project_id와 queue_id를 제외한 모든 필드는 선택 사항이며, 제공된 필드만 업데이트됩니다. Pydantic 필드:

project_id: <class 'str'>
queue_id: <class 'str'>
name: str | None
description: str | None
scorer_refs: list[str] | None
wb_user_id: str | None

`class` `AnnotationQueueUpdateRes`

어노테이션 큐 업데이트 응답입니다. Pydantic 필드:

queue: <class 'AnnotationQueueSchema'>

`class` `AnnotationQueuesQueryReq`

프로젝트의 어노테이션 큐를 쿼리하는 요청입니다. Pydantic 필드:

project_id: <class 'str'>
name: str | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
limit: int | None
offset: int | None

`class` `AnnotationQueuesQueryRes`

어노테이션 큐 쿼리 응답입니다. Pydantic 필드:

queues: list[AnnotationQueueSchema]

`class` `AnnotationQueuesStatsReq`

여러 어노테이션 큐의 통계를 쿼리하는 요청입니다. Pydantic 필드:

project_id: <class 'str'>
queue_ids: list[str]

`class` `AnnotationQueuesStatsRes`

여러 어노테이션 큐에 대한 통계 응답입니다. Pydantic 필드:

stats: list[AnnotationQueueStatsSchema]

`class` `AnnotatorQueueItemsProgressUpdateReq`

현재 어노테이터의 큐 항목에 대한 어노테이션 상태를 업데이트하는 요청입니다. 유효한 상태 전이:

(없음) -> ‘in_progress’: 항목을 진행 중으로 표시합니다(기록이 아직 없을 때만)
(없음) -> ‘completed’ 또는 ‘skipped’: 항목을 바로 완료하거나 건너뜁니다
‘in_progress’ 또는 ‘unstarted’ -> ‘completed’ 또는 ‘skipped’: 시작된 항목을 완료하거나 건너뜁니다
same_state -> same_state: 멱등적인 no-op입니다(기존 항목을 변경 없이 반환)

Pydantic 필드:

project_id: <class 'str'>
queue_id: <class 'str'>
item_id: <class 'str'>
annotation_state: <class 'str'>
wb_user_id: str | None

`class` `AnnotatorQueueItemsProgressUpdateRes`

어노테이션 상태를 업데이트한 응답입니다. Pydantic 필드:

item: <class 'AnnotationQueueItemSchema'>

`class` `CallBatchEndMode`

Pydantic 필드:

mode: <class 'str'>
req: <class 'CallEndReq'>

`class` `CallBatchStartMode`

Pydantic 필드:

mode: <class 'str'>
req: <class 'CallStartReq'>

`class` `CallCreateBatchReq`

Pydantic 필드:

batch: list[CallBatchStartMode | CallBatchEndMode]

`class` `CallCreateBatchRes`

Pydantic 필드:

res: list[CallStartRes | CallEndRes]

`class` `CallEndReq`

Pydantic 필드:

end: <class 'EndedCallSchemaForInsert'>

`class` `CallEndRes`

`class` `CallEndV2Req`

v2 API를 통해 단일 Call을 종료하는 요청입니다. Pydantic 필드:

end: <class 'EndedCallSchemaForInsertWithStartedAt'>

`class` `CallEndV2Res`

v2 API를 통해 단일 Call을 종료할 때의 응답입니다.

`class` `CallMetricSpec`

호출 단위에서 집계할 메트릭에 대한 명세입니다(모델별로 그룹화하지 않음). Pydantic 필드:

metric: typing.Literal['latency_ms', 'call_count', 'error_count']
aggregations: list[AggregationType]
percentiles: list[float]

`class` `CallReadReq`

Pydantic 필드:

project_id: <class 'str'>
id: <class 'str'>
include_costs: bool | None
include_storage_size: bool | None
include_total_storage_size: bool | None

`class` `CallReadRes`

Pydantic 필드:

call: CallSchema | None

`class` `CallSchema`

Pydantic 필드:

id: <class 'str'>
project_id: <class 'str'>
op_name: <class 'str'>
display_name: str | None
trace_id: <class 'str'>
parent_id: str | None
thread_id: str | None
turn_id: str | None
started_at: <class 'datetime.datetime'>
attributes: dict[str, typing.Any]
inputs: dict[str, typing.Any]
ended_at: datetime.datetime | None
exception: str | None
output: typing.Any | None
summary: SummaryMap | None
wb_user_id: str | None
wb_run_id: str | None
wb_run_step: int | None
wb_run_step_end: int | None
deleted_at: datetime.datetime | None
storage_size_bytes: int | None
total_storage_size_bytes: int | None

`방법` `serialize_typed_dicts`

serialize_typed_dicts(v: dict[str, Any]) → dict[str, Any]

`class` `CallStartReq`

Pydantic 필드:

start: <class 'StartedCallSchemaForInsert'>

`class` `CallStartRes`

Pydantic 필드:

id: <class 'str'>
trace_id: <class 'str'>

`class` `CallStartV2Req`

v2 API를 통해 단일 Call을 시작하는 요청입니다. Pydantic 필드:

start: <class 'StartedCallSchemaForInsert'>

`class` `CallStartV2Res`

v2 API를 통해 단일 Call을 시작할 때 반환되는 응답입니다. Pydantic 필드:

id: <class 'str'>
trace_id: <class 'str'>

`class` `CallStatsReq`

일정 시간 범위의 집계된 호출 통계에 대한 요청입니다. Pydantic 필드:

project_id: <class 'str'>
start: <class 'datetime.datetime'>
end: datetime.datetime | None
granularity: int | None
usage_metrics: list[UsageMetricSpec] | None
call_metrics: list[CallMetricSpec] | None
filter: CallsFilter | None
timezone: <class 'str'>

`방법` `validate_date_range`

validate_date_range() → CallStatsReq

호출 통계 요청이 안전한 날짜 범위 내로 제한되도록 합니다.

`class` `CallStatsRes`

시계열 호출 통계를 담은 응답입니다. Pydantic 필드:

start: <class 'datetime.datetime'>
end: <class 'datetime.datetime'>
granularity: <class 'int'>
timezone: <class 'str'>
usage_buckets: list[dict[str, typing.Any]]
call_buckets: list[dict[str, typing.Any]]

`class` `CallUpdateReq`

Pydantic 필드:

project_id: <class 'str'>
call_id: <class 'str'>
display_name: str | None
wb_user_id: str | None

`class` `CallUpdateRes`

`class` `CallsDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
call_ids: list[str]
wb_user_id: str | None

`class` `CallsDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `CallsFilter`

Pydantic 필드:

op_names: list[str] | None
input_refs: list[str] | None
output_refs: list[str] | None
parent_ids: list[str] | None
trace_ids: list[str] | None
call_ids: list[str] | None
thread_ids: list[str] | None
turn_ids: list[str] | None
trace_roots_only: bool | None
wb_user_ids: list[str] | None
wb_run_ids: list[str] | None

`class` `CallsQueryReq`

Pydantic 필드:

project_id: <class 'str'>
filter: CallsFilter | None
limit: int | None
offset: int | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
query: weave.trace_server.interface.query.Query | None
include_costs: bool | None
include_feedback: bool | None
include_storage_size: bool | None
include_total_storage_size: bool | None
columns: list[str] | None
expand_columns: list[str] | None
return_expanded_column_values: bool | None

`class` `CallsQueryRes`

Pydantic 필드:

calls: list[CallSchema]

`class` `CallsQueryStatsReq`

Pydantic 필드:

project_id: <class 'str'>
filter: CallsFilter | None
query: weave.trace_server.interface.query.Query | None
limit: int | None
include_total_storage_size: bool | None
expand_columns: list[str] | None

`class` `CallsQueryStatsRes`

Pydantic 필드:

count: <class 'int'>
total_storage_size_bytes: int | None

`class` `CallsScoreReq`

호출 목록의 스코어링 작업을 큐에 추가하는 요청입니다. 스코어링은 Kafka에서 메시지를 소비하고 각 call_id에 각 scorer_ref를 적용하는 call_scoring_worker에 의해 비동기적으로 수행됩니다. Pydantic 필드:

project_id: <class 'str'>
call_ids: list[str]
scorer_refs: list[str]
wb_user_id: str | None

`class` `CallsScoreRes`

calls_score에 대한 빈 응답입니다. 이 인터페이스 전반에서 사용하는 규칙을 따르고, 이후 호환성이 깨지는 변경 없이 필드를 추가할 수 있도록 None을 반환하는 대신 모델로 정의했습니다.

`class` `CallsUpsertCompleteReq`

완료된 call 배치를 업서트하기 위한 요청입니다. Pydantic 필드:

batch: list[CompletedCallSchemaForInsert]

`class` `CallsUpsertCompleteRes`

완료된 call 배치를 업서트할 때의 응답입니다.

`class` `CallsUsageReq`

여러 루트 call의 집계 사용량을 계산하기 위한 요청입니다. 이 엔드포인트는 요청된 각 루트 call의 사용 메트릭을 반환합니다. 각 루트의 메트릭에는 해당 루트 자체의 사용량과 모든 하위 call의 사용량 합계가 포함됩니다. 참고: 집계를 위해 일치하는 모든 call을 메모리에 로드합니다. 결과 집합이 매우 큰 경우(>10k calls)에는 루트 call ID를 여러 배치로 나누거나 애플리케이션 레이어에서 더 좁은 필터를 사용하는 것을 고려하세요. Pydantic 필드:

project_id: <class 'str'>
call_ids: list[str]
include_costs: <class 'bool'>
limit: <class 'int'>

`class` `CallsUsageRes`

루트 call별로 집계된 사용 메트릭이 포함된 응답입니다. Pydantic 필드:

call_usage: dict[str, dict[str, LLMAggregatedUsage]]
unfinished_call_ids: list[str]

`class` `CompletedCallSchemaForInsert`

완료된 call을 직접 삽입하기 위한 스키마입니다. 이 스키마는 삽입 시점에 이미 완료된 call을 나타내며, 시작 정보와 종료 정보가 함께 제공됩니다. calls_complete 엔드포인트에서 사용됩니다. Pydantic 필드:

project_id: <class 'str'>
id: <class 'str'>
trace_id: <class 'str'>
op_name: <class 'str'>
started_at: <class 'datetime.datetime'>
ended_at: <class 'datetime.datetime'>
display_name: str | None
parent_id: str | None
thread_id: str | None
turn_id: str | None
attributes: dict[str, typing.Any]
inputs: dict[str, typing.Any]
output: typing.Any | None
summary: <class 'SummaryInsertMap'>
otel_dump: dict[str, typing.Any] | None
exception: str | None
wb_user_id: str | None
wb_run_id: str | None
wb_run_step: int | None
wb_run_step_end: int | None

`방법` `serialize_typed_dicts`

serialize_typed_dicts(v: dict[str, Any]) → dict[str, Any]

`class` `CompletionsCreateReq`

Pydantic 필드:

project_id: <class 'str'>
inputs: <class 'CompletionsCreateRequestInputs'>
wb_user_id: str | None
track_llm_call: bool | None
trace_id: str | None
parent_id: str | None

`class` `CompletionsCreateRequestInputs`

Pydantic 필드:

model: <class 'str'>
messages: <class 'list'>
timeout: float | str | None
temperature: float | None
top_p: float | None
n: int | None
stop: str | list | None
max_completion_tokens: int | None
max_tokens: int | None
modalities: list | None
presence_penalty: float | None
frequency_penalty: float | None
stream: bool | None
logit_bias: dict | None
user: str | None
response_format: dict | type[pydantic.main.BaseModel] | None
seed: int | None
tools: list | None
tool_choice: str | dict | None
logprobs: bool | None
top_logprobs: int | None
parallel_tool_calls: bool | None
extra_headers: dict | None
functions: list | None
function_call: str | None
api_version: str | None
prompt: str | None
template_vars: dict[str, typing.Any] | None
vertex_credentials: str | None

`class` `CompletionsCreateRes`

Pydantic 필드:

response: dict[str, typing.Any]
weave_call_id: str | None

`class` `CostCreateInput`

Pydantic 필드:

prompt_token_cost: <class 'float'>
completion_token_cost: <class 'float'>
prompt_token_cost_unit: str | None
completion_token_cost_unit: str | None
effective_date: datetime.datetime | None
provider_id: str | None

`class` `CostCreateReq`

Pydantic 필드:

project_id: <class 'str'>
costs: dict[str, CostCreateInput]
wb_user_id: str | None

`class` `CostCreateRes`

Pydantic 필드:

ids: list[tuple[str, str]]

`class` `CostPurgeReq`

Pydantic 필드:

project_id: <class 'str'>
query: <class 'weave.trace_server.interface.query.Query'>

`class` `CostPurgeRes`

`class` `CostQueryOutput`

Pydantic 필드:

id: str | None
llm_id: str | None
prompt_token_cost: float | None
completion_token_cost: float | None
prompt_token_cost_unit: str | None
completion_token_cost_unit: str | None
effective_date: datetime.datetime | None
provider_id: str | None

`class` `CostQueryReq`

Pydantic 필드:

project_id: <class 'str'>
fields: list[str] | None
query: weave.trace_server.interface.query.Query | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
limit: int | None
offset: int | None

`class` `CostQueryRes`

Pydantic 필드:

results: list[CostQueryOutput]

`class` `DatasetCreateBody`

Pydantic 필드:

name: str | None
description: str | None
rows: list[dict[str, typing.Any]]

`class` `DatasetCreateReq`

Pydantic 필드:

name: str | None
description: str | None
rows: list[dict[str, typing.Any]]
project_id: <class 'str'>
wb_user_id: str | None

`class` `DatasetCreateRes`

Pydantic 필드:

digest: <class 'str'>
object_id: <class 'str'>
version_index: <class 'int'>

`class` `DatasetDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None
wb_user_id: str | None

`class` `DatasetDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `DatasetListReq`

Pydantic 필드:

project_id: <class 'str'>
limit: int | None
offset: int | None
wb_user_id: str | None

`class` `DatasetReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
wb_user_id: str | None

`class` `DatasetReadRes`

Pydantic 필드:

object_id: <class 'str'>
digest: <class 'str'>
version_index: <class 'int'>
created_at: <class 'datetime.datetime'>
name: <class 'str'>
description: str | None
rows: <class 'str'>

`class` `EndedCallSchemaForInsert`

Pydantic 필드:

project_id: <class 'str'>
id: <class 'str'>
ended_at: <class 'datetime.datetime'>
exception: str | None
output: typing.Any | None
summary: <class 'SummaryInsertMap'>
wb_run_step_end: int | None

`방법` `serialize_typed_dicts`

serialize_typed_dicts(v: dict[str, Any]) → dict[str, Any]

`class` `EndedCallSchemaForInsertWithStartedAt`

v2 종료 업데이트를 위한 선택적 started_at이 포함된 종료 call 스키마입니다. started_at이 제공되면 기본 키 (project_id, started_at, id)를 활용해 ClickHouse 쿼리를 더 효율적으로 수행할 수 있습니다. 제공되지 않으면 쿼리는 (project_id, id)만 사용하도록 대체됩니다. Pydantic 필드:

project_id: <class 'str'>
id: <class 'str'>
ended_at: <class 'datetime.datetime'>
exception: str | None
output: typing.Any | None
summary: <class 'SummaryInsertMap'>
wb_run_step_end: int | None
started_at: datetime.datetime | None

`방법` `serialize_typed_dicts`

serialize_typed_dicts(v: dict[str, Any]) → dict[str, Any]

`class` `EvalResultsEvaluationSummary`

Pydantic 필드:

evaluation_call_id: <class 'str'>
trial_count: <class 'int'>
scorer_stats: list[EvalResultsScorerStats]
evaluation_ref: str | None
model_ref: str | None
display_name: str | None
trace_id: str | None
started_at: str | None

`class` `EvalResultsQueryBody`

Pydantic 필드:

evaluation_call_ids: list[str] | None
evaluation_run_ids: list[str] | None
require_intersection: <class 'bool'>
include_raw_data_rows: <class 'bool'>
resolve_row_refs: <class 'bool'>
include_rows: <class 'bool'>
include_summary: <class 'bool'>
summary_require_intersection: bool | None
limit: int | None
offset: <class 'int'>

`방법` `validate_identifiers`

validate_identifiers() → EvalResultsQueryBody

평가 식별자가 하나 이상 제공되었는지 확인합니다.

`class` `EvalResultsQueryReq`

Pydantic 필드:

evaluation_call_ids: list[str] | None
evaluation_run_ids: list[str] | None
require_intersection: <class 'bool'>
include_raw_data_rows: <class 'bool'>
resolve_row_refs: <class 'bool'>
include_rows: <class 'bool'>
include_summary: <class 'bool'>
summary_require_intersection: bool | None
limit: int | None
offset: <class 'int'>
project_id: <class 'str'>

`방법` `validate_identifiers`

validate_identifiers() → EvalResultsQueryBody

평가 식별자가 하나 이상 제공되었는지 검증합니다.

`class` `EvalResultsQueryRes`

Pydantic 필드:

rows: list[EvalResultsRow]
total_rows: <class 'int'>
summary: ForwardRef('EvalResultsSummaryRes | None')
warnings: list[str]

`class` `EvalResultsRow`

Pydantic 필드:

row_digest: <class 'str'>
raw_data_row: typing.Any | None
evaluations: list[EvalResultsRowEvaluation]

`class` `EvalResultsRowEvaluation`

Pydantic 필드:

evaluation_call_id: <class 'str'>
trials: list[EvalResultsTrial]

`class` `EvalResultsScorerStats`

단일 평탄화 점수 차원(scorer_key 또는 scorer_key.path.to.leaf)에 대한 통계입니다. Pydantic 필드:

scorer_key: <class 'str'>
path: str | None
value_type: typing.Optional[typing.Literal['binary', 'continuous']]
trial_count: <class 'int'>
numeric_count: <class 'int'>
numeric_mean: float | None
pass_true_count: <class 'int'>
pass_known_count: <class 'int'>
pass_rate: float | None
pass_signal_coverage: float | None

`class` `EvalResultsSummaryRes`

Pydantic 필드:

row_count: <class 'int'>
evaluations: list[EvalResultsEvaluationSummary]

`class` `EvalResultsTrial`

Pydantic 필드:

predict_and_score_call_id: <class 'str'>
predict_call_id: str | None
model_output: typing.Any | None
scores: dict[str, typing.Any]
model_latency_seconds: float | None
total_tokens: int | None
scorer_call_ids: dict[str, str]

`class` `EvaluateModelReq`

Pydantic 필드:

project_id: <class 'str'>
evaluation_ref: <class 'str'>
model_ref: <class 'str'>
wb_user_id: str | None

`class` `EvaluateModelRes`

Pydantic 필드:

call_id: <class 'str'>

`class` `EvaluationCreateBody`

Pydantic 필드:

name: <class 'str'>
description: str | None
dataset: <class 'str'>
scorers: list[str] | None
trials: <class 'int'>
evaluation_name: str | None
eval_attributes: dict[str, typing.Any] | None

`class` `EvaluationCreateReq`

Pydantic 필드:

name: <class 'str'>
description: str | None
dataset: <class 'str'>
scorers: list[str] | None
trials: <class 'int'>
evaluation_name: str | None
eval_attributes: dict[str, typing.Any] | None
project_id: <class 'str'>
wb_user_id: str | None

`class` `EvaluationCreateRes`

Pydantic 필드:

digest: <class 'str'>
object_id: <class 'str'>
version_index: <class 'int'>
evaluation_ref: <class 'str'>

`class` `EvaluationDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None
wb_user_id: str | None

`class` `EvaluationDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `EvaluationListReq`

Pydantic 필드:

project_id: <class 'str'>
limit: int | None
offset: int | None
wb_user_id: str | None

`class` `EvaluationReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
wb_user_id: str | None

`class` `EvaluationReadRes`

Pydantic 필드:

object_id: <class 'str'>
digest: <class 'str'>
version_index: <class 'int'>
created_at: <class 'datetime.datetime'>
name: <class 'str'>
description: str | None
dataset: <class 'str'>
scorers: list[str]
trials: <class 'int'>
evaluation_name: str | None
evaluate_op: str | None
predict_and_score_op: str | None
summarize_op: str | None

`class` `EvaluationRunCreateBody`

Pydantic 필드:

evaluation: <class 'str'>
model: <class 'str'>

`class` `EvaluationRunCreateReq`

Pydantic 필드:

evaluation: <class 'str'>
model: <class 'str'>
project_id: <class 'str'>
wb_user_id: str | None

`class` `EvaluationRunCreateRes`

Pydantic 필드:

evaluation_run_id: <class 'str'>

`class` `EvaluationRunDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
evaluation_run_ids: list[str]
wb_user_id: str | None

`class` `EvaluationRunDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `EvaluationRunFilter`

Pydantic 필드:

evaluations: list[str] | None
models: list[str] | None
evaluation_run_ids: list[str] | None

`class` `EvaluationRunFinishBody`

REST API를 통해 evaluation run을 완료할 때의 요청 본문입니다. 이 모델은 project_id와 evaluation_run_id를 제외합니다. 이 값들은 RESTful 엔드포인트의 URL 경로에서 가져오기 때문입니다. Pydantic 필드:

summary: dict[str, typing.Any] | None

`class` `EvaluationRunFinishReq`

Pydantic 필드:

summary: dict[str, typing.Any] | None
project_id: <class 'str'>
evaluation_run_id: <class 'str'>
wb_user_id: str | None

`class` `EvaluationRunFinishRes`

Pydantic 필드:

success: <class 'bool'>

`class` `EvaluationRunListReq`

Pydantic 필드:

project_id: <class 'str'>
filter: EvaluationRunFilter | None
limit: int | None
offset: int | None

`class` `EvaluationRunReadReq`

Pydantic 필드:

project_id: <class 'str'>
evaluation_run_id: <class 'str'>

`class` `EvaluationRunReadRes`

Pydantic 필드:

evaluation_run_id: <class 'str'>
evaluation: <class 'str'>
model: <class 'str'>
status: str | None
started_at: datetime.datetime | None
finished_at: datetime.datetime | None
summary: dict[str, typing.Any] | None

`class` `EvaluationStatusComplete`

Pydantic 필드:

code: typing.Literal['complete']
output: dict[str, typing.Any]

`class` `EvaluationStatusFailed`

Pydantic 필드:

code: typing.Literal['failed']
error: str | None

`class` `EvaluationStatusNotFound`

Pydantic 필드:

code: typing.Literal['not_found']

`class` `EvaluationStatusReq`

Pydantic 필드:

project_id: <class 'str'>
call_id: <class 'str'>

`class` `EvaluationStatusRes`

Pydantic 필드:

status: EvaluationStatusNotFound | EvaluationStatusRunning | EvaluationStatusFailed | EvaluationStatusComplete

`class` `EvaluationStatusRunning`

Pydantic 필드:

code: typing.Literal['running']
completed_rows: <class 'int'>
total_rows: <class 'int'>

`class` `ExportTracePartialSuccess`

Pydantic 필드:

rejected_spans: <class 'int'>
error_message: <class 'str'>

`class` `ExtraKeysTypedDict`

`class` `Feedback`

Pydantic 필드:

id: <class 'str'>
project_id: <class 'str'>
weave_ref: <class 'str'>
creator: str | None
feedback_type: <class 'str'>
payload: dict[str, typing.Any]
annotation_ref: str | None
runnable_ref: str | None
call_ref: str | None
trigger_ref: str | None
queue_id: str | None
wb_user_id: str | None
created_at: <class 'datetime.datetime'>

`class` `FeedbackCreateBatchReq`

Pydantic 필드:

batch: list[FeedbackCreateReq]

`class` `FeedbackCreateBatchRes`

Pydantic 필드:

res: list[FeedbackCreateRes]

`class` `FeedbackCreateReq`

Pydantic 필드:

id: str | None
project_id: <class 'str'>
weave_ref: <class 'str'>
creator: str | None
feedback_type: <class 'str'>
payload: dict[str, typing.Any]
annotation_ref: str | None
runnable_ref: str | None
call_ref: str | None
trigger_ref: str | None
queue_id: str | None
wb_user_id: str | None

`class` `FeedbackCreateRes`

Pydantic 필드:

id: <class 'str'>
created_at: <class 'datetime.datetime'>
wb_user_id: <class 'str'>
payload: dict[str, typing.Any]

`class` `FeedbackDict`

`class` `FeedbackMetricSpec`

집계 대상 피드백 페이로드 메트릭의 사양입니다. Pydantic 필드:

json_path: <class 'str'>
value_type: typing.Literal['numeric', 'boolean', 'categorical']
aggregations: list[AggregationType]
percentiles: list[float]

`class` `FeedbackPayloadPath`

추론된 유형이 지정된 피드백 페이로드 내 경로입니다. Pydantic 필드:

json_path: <class 'str'>
value_type: typing.Literal['numeric', 'boolean', 'categorical']

`class` `FeedbackPayloadSchemaReq`

피드백 페이로드 스키마 조회 요청입니다. Pydantic 필드:

project_id: <class 'str'>
start: <class 'datetime.datetime'>
end: datetime.datetime | None
feedback_type: str | None
trigger_ref: str | None
sample_limit: <class 'int'>

`방법` `validate_date_range`

validate_date_range() → _FeedbackFilterBase

피드백 요청이 안전한 날짜 범위 내로 제한되도록 하세요.

`class` `FeedbackPayloadSchemaRes`

발견된 피드백 페이로드 경로와 유형을 포함한 응답입니다. Pydantic 필드:

paths: list[FeedbackPayloadPath]

`class` `FeedbackPurgeReq`

Pydantic 필드:

project_id: <class 'str'>
query: <class 'weave.trace_server.interface.query.Query'>

`class` `FeedbackPurgeRes`

`class` `FeedbackQueryReq`

Pydantic 필드:

project_id: <class 'str'>
fields: list[str] | None
query: weave.trace_server.interface.query.Query | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
limit: int | None
offset: int | None

`class` `FeedbackQueryRes`

Pydantic 필드:

result: list[dict[str, typing.Any]]

`class` `FeedbackReplaceReq`

Pydantic 필드:

id: str | None
project_id: <class 'str'>
weave_ref: <class 'str'>
creator: str | None
feedback_type: <class 'str'>
payload: dict[str, typing.Any]
annotation_ref: str | None
runnable_ref: str | None
call_ref: str | None
trigger_ref: str | None
queue_id: str | None
wb_user_id: str | None
feedback_id: <class 'str'>

`class` `FeedbackReplaceRes`

Pydantic 필드:

id: <class 'str'>
created_at: <class 'datetime.datetime'>
wb_user_id: <class 'str'>
payload: dict[str, typing.Any]

`class` `FeedbackStatsReq`

시간 버킷별 집계 피드백 통계를 요청합니다. Pydantic 필드:

project_id: <class 'str'>
start: <class 'datetime.datetime'>
end: datetime.datetime | None
feedback_type: str | None
trigger_ref: str | None
granularity: int | None
timezone: <class 'str'>
metrics: list[FeedbackMetricSpec]

`방법` `validate_date_range`

validate_date_range() → _FeedbackFilterBase

피드백 요청이 안전한 날짜 범위 내로 제한되도록 하세요.

`class` `FeedbackStatsRes`

시계열 피드백 통계를 담은 응답입니다. Pydantic 필드:

start: <class 'datetime.datetime'>
end: <class 'datetime.datetime'>
granularity: <class 'int'>
timezone: <class 'str'>
buckets: list[dict[str, typing.Any]]
window_stats: dict[str, dict[str, float | None]] | None

`class` `FileContentReadReq`

Pydantic 필드:

project_id: <class 'str'>
digest: <class 'str'>

`class` `FileContentReadRes`

Pydantic 필드:

content: <class 'bytes'>

`class` `FileCreateReq`

Pydantic 필드:

project_id: <class 'str'>
name: <class 'str'>
content: <class 'bytes'>
expected_digest: str | None

`class` `FileCreateRes`

Pydantic 필드:

digest: <class 'str'>

`class` `FilesStatsReq`

Pydantic 필드:

project_id: <class 'str'>

`class` `FilesStatsRes`

Pydantic 필드:

total_size_bytes: <class 'int'>

`class` `FullTraceServerInterface`

V1 및 Object API를 모두 지원하는 전체 트레이스 서버 인터페이스입니다. 이 프로토콜은 레거시 V1 엔드포인트와 최신 Object 엔드포인트를 포함해 전체 API 집합을 지원하는 트레이스 서버 구현을 나타냅니다. 두 API 버전을 모두 지원해야 하는 구현에는 이 유형을 사용하세요.

`방법` `actions_execute_batch`

actions_execute_batch(req: ActionsExecuteBatchReq) → ActionsExecuteBatchRes

`방법` `aliases_list`

aliases_list(req: AliasesListReq) → AliasesListRes

`방법` `annotation_queue_add_calls`

annotation_queue_add_calls(
    req: AnnotationQueueAddCallsReq
) → AnnotationQueueAddCallsRes

`방법` `annotation_queue_create`

annotation_queue_create(
    req: AnnotationQueueCreateReq
) → AnnotationQueueCreateRes

`방법` `annotation_queue_delete`

annotation_queue_delete(
    req: AnnotationQueueDeleteReq
) → AnnotationQueueDeleteRes

`방법` `annotation_queue_items_query`

annotation_queue_items_query(
    req: AnnotationQueueItemsQueryReq
) → AnnotationQueueItemsQueryRes

`방법` `annotation_queue_read`

annotation_queue_read(req: AnnotationQueueReadReq) → AnnotationQueueReadRes

`방법` `annotation_queue_update`

annotation_queue_update(
    req: AnnotationQueueUpdateReq
) → AnnotationQueueUpdateRes

`방법` `annotation_queues_query_stream`

annotation_queues_query_stream(
    req: AnnotationQueuesQueryReq
) → Iterator[AnnotationQueueSchema]

`방법` `annotation_queues_stats`

annotation_queues_stats(
    req: AnnotationQueuesStatsReq
) → AnnotationQueuesStatsRes

`방법` `annotator_queue_items_progress_update`

annotator_queue_items_progress_update(
    req: AnnotatorQueueItemsProgressUpdateReq
) → AnnotatorQueueItemsProgressUpdateRes

`방법` `call_end`

call_end(req: CallEndReq) → CallEndRes

`방법` `call_end_v2`

call_end_v2(req: CallEndV2Req) → CallEndV2Res

`방법` `call_read`

call_read(req: CallReadReq) → CallReadRes

`방법` `call_start`

call_start(req: CallStartReq) → CallStartRes

`방법` `call_start_batch`

call_start_batch(req: CallCreateBatchReq) → CallCreateBatchRes

`방법` `call_start_v2`

call_start_v2(req: CallStartV2Req) → CallStartV2Res

`방법` `call_stats`

call_stats(req: 'CallStatsReq') → CallStatsRes

`방법` `call_update`

call_update(req: CallUpdateReq) → CallUpdateRes

`방법` `calls_complete`

calls_complete(req: CallsUpsertCompleteReq) → CallsUpsertCompleteRes

`방법` `calls_delete`

calls_delete(req: CallsDeleteReq) → CallsDeleteRes

`방법` `calls_query`

calls_query(req: CallsQueryReq) → CallsQueryRes

`방법` `calls_query_stats`

calls_query_stats(req: CallsQueryStatsReq) → CallsQueryStatsRes

`방법` `calls_query_stream`

calls_query_stream(req: CallsQueryReq) → Iterator[CallSchema]

`방법` `calls_score`

calls_score(req: CallsScoreReq) → CallsScoreRes

`방법` `calls_usage`

calls_usage(req: 'CallsUsageReq') → CallsUsageRes

`방법` `completions_create`

completions_create(req: CompletionsCreateReq) → CompletionsCreateRes

`방법` `completions_create_stream`

completions_create_stream(req: CompletionsCreateReq) → Iterator[dict[str, Any]]

`방법` `cost_create`

cost_create(req: CostCreateReq) → CostCreateRes

`방법` `cost_purge`

cost_purge(req: CostPurgeReq) → CostPurgeRes

`방법` `cost_query`

cost_query(req: CostQueryReq) → CostQueryRes

`방법` `dataset_create`

dataset_create(req: DatasetCreateReq) → DatasetCreateRes

`방법` `dataset_delete`

dataset_delete(req: DatasetDeleteReq) → DatasetDeleteRes

`방법` `dataset_list`

dataset_list(req: DatasetListReq) → Iterator[DatasetReadRes]

`방법` `dataset_read`

dataset_read(req: DatasetReadReq) → DatasetReadRes

`방법` `eval_results_query`

eval_results_query(req: EvalResultsQueryReq) → EvalResultsQueryRes

`방법` `evaluate_model`

evaluate_model(req: EvaluateModelReq) → EvaluateModelRes

`방법` `evaluation_create`

evaluation_create(req: EvaluationCreateReq) → EvaluationCreateRes

`방법` `evaluation_delete`

evaluation_delete(req: EvaluationDeleteReq) → EvaluationDeleteRes

`방법` `evaluation_list`

evaluation_list(req: EvaluationListReq) → Iterator[EvaluationReadRes]

`방법` `evaluation_read`

evaluation_read(req: EvaluationReadReq) → EvaluationReadRes

`방법` `evaluation_run_create`

evaluation_run_create(req: EvaluationRunCreateReq) → EvaluationRunCreateRes

`방법` `evaluation_run_delete`

evaluation_run_delete(req: EvaluationRunDeleteReq) → EvaluationRunDeleteRes

`방법` `evaluation_run_finish`

evaluation_run_finish(req: EvaluationRunFinishReq) → EvaluationRunFinishRes

`방법` `evaluation_run_list`

evaluation_run_list(req: EvaluationRunListReq) → Iterator[EvaluationRunReadRes]

`방법` `evaluation_run_read`

evaluation_run_read(req: EvaluationRunReadReq) → EvaluationRunReadRes

`방법` `evaluation_status`

evaluation_status(req: EvaluationStatusReq) → EvaluationStatusRes

`방법` `feedback_create`

feedback_create(req: FeedbackCreateReq) → FeedbackCreateRes

`방법` `feedback_create_batch`

feedback_create_batch(req: FeedbackCreateBatchReq) → FeedbackCreateBatchRes

`방법` `feedback_payload_schema`

feedback_payload_schema(
    req: FeedbackPayloadSchemaReq
) → FeedbackPayloadSchemaRes

`방법` `feedback_purge`

feedback_purge(req: FeedbackPurgeReq) → FeedbackPurgeRes

`방법` `feedback_query`

feedback_query(req: FeedbackQueryReq) → FeedbackQueryRes

`방법` `feedback_replace`

feedback_replace(req: FeedbackReplaceReq) → FeedbackReplaceRes

`방법` `feedback_stats`

feedback_stats(req: FeedbackStatsReq) → FeedbackStatsRes

`방법` `file_content_read`

file_content_read(req: FileContentReadReq) → FileContentReadRes

`방법` `file_create`

file_create(req: FileCreateReq) → FileCreateRes

`방법` `files_stats`

files_stats(req: FilesStatsReq) → FilesStatsRes

`방법` `image_create`

image_create(req: ImageGenerationCreateReq) → ImageGenerationCreateRes

`방법` `model_create`

model_create(req: ModelCreateReq) → ModelCreateRes

`방법` `model_delete`

model_delete(req: ModelDeleteReq) → ModelDeleteRes

`방법` `model_list`

model_list(req: ModelListReq) → Iterator[ModelReadRes]

`방법` `model_read`

model_read(req: ModelReadReq) → ModelReadRes

`방법` `obj_add_tags`

obj_add_tags(req: ObjAddTagsReq) → ObjAddTagsRes

`방법` `obj_create`

obj_create(req: ObjCreateReq) → ObjCreateRes

`방법` `obj_delete`

obj_delete(req: ObjDeleteReq) → ObjDeleteRes

`방법` `obj_read`

obj_read(req: ObjReadReq) → ObjReadRes

`방법` `obj_remove_aliases`

obj_remove_aliases(req: ObjRemoveAliasesReq) → ObjRemoveAliasesRes

`방법` `obj_remove_tags`

obj_remove_tags(req: ObjRemoveTagsReq) → ObjRemoveTagsRes

`방법` `obj_set_aliases`

obj_set_aliases(req: ObjSetAliasesReq) → ObjSetAliasesRes

`방법` `objs_query`

objs_query(req: ObjQueryReq) → ObjQueryRes

`방법` `op_create`

op_create(req: OpCreateReq) → OpCreateRes

`방법` `op_delete`

op_delete(req: OpDeleteReq) → OpDeleteRes

`방법` `op_list`

op_list(req: OpListReq) → Iterator[OpReadRes]

`방법` `op_read`

op_read(req: OpReadReq) → OpReadRes

`방법` `otel_export`

otel_export(req: OTelExportReq) → OTelExportRes

`방법` `prediction_create`

prediction_create(req: PredictionCreateReq) → PredictionCreateRes

`방법` `prediction_delete`

prediction_delete(req: PredictionDeleteReq) → PredictionDeleteRes

`방법` `prediction_finish`

prediction_finish(req: PredictionFinishReq) → PredictionFinishRes

`방법` `prediction_list`

prediction_list(req: PredictionListReq) → Iterator[PredictionReadRes]

`방법` `prediction_read`

prediction_read(req: PredictionReadReq) → PredictionReadRes

`방법` `project_stats`

project_stats(req: ProjectStatsReq) → ProjectStatsRes

`방법` `refs_read_batch`

refs_read_batch(req: RefsReadBatchReq) → RefsReadBatchRes

`방법` `score_create`

score_create(req: ScoreCreateReq) → ScoreCreateRes

`방법` `score_delete`

score_delete(req: ScoreDeleteReq) → ScoreDeleteRes

`방법` `score_list`

score_list(req: ScoreListReq) → Iterator[ScoreReadRes]

`방법` `score_read`

score_read(req: ScoreReadReq) → ScoreReadRes

`방법` `scorer_create`

scorer_create(req: ScorerCreateReq) → ScorerCreateRes

`방법` `scorer_delete`

scorer_delete(req: ScorerDeleteReq) → ScorerDeleteRes

`방법` `scorer_list`

scorer_list(req: ScorerListReq) → Iterator[ScorerReadRes]

`방법` `scorer_read`

scorer_read(req: ScorerReadReq) → ScorerReadRes

`방법` `table_create`

table_create(req: TableCreateReq) → TableCreateRes

`방법` `table_create_from_digests`

table_create_from_digests(
    req: TableCreateFromDigestsReq
) → TableCreateFromDigestsRes

`방법` `table_query`

table_query(req: TableQueryReq) → TableQueryRes

`방법` `table_query_stats`

table_query_stats(req: TableQueryStatsReq) → TableQueryStatsRes

`방법` `table_query_stats_batch`

table_query_stats_batch(req: TableQueryStatsBatchReq) → TableQueryStatsBatchRes

`방법` `table_query_stream`

table_query_stream(req: TableQueryReq) → Iterator[TableRowSchema]

`방법` `table_update`

table_update(req: TableUpdateReq) → TableUpdateRes

`방법` `tags_list`

tags_list(req: TagsListReq) → TagsListRes

`방법` `threads_query_stream`

threads_query_stream(req: ThreadsQueryReq) → Iterator[ThreadSchema]

`방법` `trace_usage`

trace_usage(req: 'TraceUsageReq') → TraceUsageRes

`클래스` `ImageGenerationCreateReq`

Pydantic 필드:

project_id: <class 'str'>
inputs: <class 'ImageGenerationRequestInputs'>
wb_user_id: str | None
track_llm_call: bool | None

`클래스` `ImageGenerationCreateRes`

Pydantic 필드:

response: dict[str, typing.Any]
weave_call_id: str | None

`클래스` `ImageGenerationRequestInputs`

Pydantic 필드:

model: <class 'str'>
prompt: <class 'str'>
n: int | None

`클래스` `LLMAggregatedUsage`

특정 LLM의 집계 사용 메트릭입니다. Pydantic 필드:

requests: <class 'int'>
prompt_tokens: <class 'int'>
completion_tokens: <class 'int'>
total_tokens: <class 'int'>
prompt_tokens_total_cost: float | None
completion_tokens_total_cost: float | None

`클래스` `LLMCostSchema`

`클래스` `LLMUsageSchema`

`클래스` `ModelCreateBody`

Pydantic 필드:

name: <class 'str'>
description: str | None
source_code: <class 'str'>
attributes: dict[str, typing.Any] | None

`클래스` `ModelCreateReq`

Pydantic 필드:

name: <class 'str'>
description: str | None
source_code: <class 'str'>
attributes: dict[str, typing.Any] | None
project_id: <class 'str'>
wb_user_id: str | None

`클래스` `ModelCreateRes`

Pydantic 필드:

digest: <class 'str'>
object_id: <class 'str'>
version_index: <class 'int'>
model_ref: <class 'str'>

`클래스` `ModelDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None
wb_user_id: str | None

`클래스` `ModelDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`클래스` `ModelListReq`

Pydantic 필드:

project_id: <class 'str'>
limit: int | None
offset: int | None

`클래스` `ModelReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>

`클래스` `ModelReadRes`

Pydantic 필드:

object_id: <class 'str'>
digest: <class 'str'>
version_index: <class 'int'>
created_at: <class 'datetime.datetime'>
name: <class 'str'>
description: str | None
source_code: <class 'str'>
attributes: dict[str, typing.Any] | None

`클래스` `OTelExportReq`

Pydantic 필드:

processed_spans: list[ProcessedResourceSpans]
project_id: <class 'str'>
wb_user_id: str | None

`클래스` `OTelExportRes`

Pydantic 필드:

partial_success: ExportTracePartialSuccess | None

`클래스` `ObjAddTagsReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
tags: list[str]
wb_user_id: str | None

`방법` `validate_tags`

validate_tags() → ObjAddTagsReq

`클래스` `ObjAddTagsRes`

`클래스` `ObjCreateReq`

Pydantic 필드:

obj: <class 'ObjSchemaForInsert'>

`클래스` `ObjCreateRes`

Pydantic 필드:

digest: <class 'str'>
object_id: str | None

`클래스` `ObjDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None

`클래스` `ObjDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`클래스` `ObjQueryReq`

Pydantic 필드:

project_id: <class 'str'>
filter: ObjectVersionFilter | None
limit: int | None
offset: int | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None
metadata_only: bool | None
include_storage_size: bool | None
include_tags_and_aliases: bool | None

`클래스` `ObjQueryRes`

Pydantic 필드:

objs: list[ObjSchema]

`클래스` `ObjReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
metadata_only: bool | None
include_tags_and_aliases: bool | None

`클래스` `ObjReadRes`

Pydantic 필드:

obj: <class 'ObjSchema'>

`클래스` `ObjRemoveAliasesReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
aliases: list[str]
wb_user_id: str | None

`방법` `validate_aliases`

validate_aliases() → ObjRemoveAliasesReq

`클래스` `ObjRemoveAliasesRes`

`클래스` `ObjRemoveTagsReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
tags: list[str]
wb_user_id: str | None

`클래스` `ObjRemoveTagsRes`

`클래스` `ObjSchema`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
created_at: <class 'datetime.datetime'>
deleted_at: datetime.datetime | None
digest: <class 'str'>
version_index: <class 'int'>
is_latest: <class 'int'>
kind: <class 'str'>
base_object_class: str | None
leaf_object_class: str | None
val: typing.Any
wb_user_id: str | None
size_bytes: int | None
tags: list[str] | None
aliases: list[str] | None

`클래스` `ObjSchemaForInsert`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
val: typing.Any
builtin_object_class: str | None
set_base_object_class: str | None
expected_digest: str | None
wb_user_id: str | None

`방법` `model_post_init`

model_post_init(_ObjSchemaForInsert__context: Any) → None

`클래스` `ObjSetAliasesReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
aliases: list[str]
wb_user_id: str | None

`방법` `validate_aliases`

validate_aliases() → ObjSetAliasesReq

`클래스` `ObjSetAliasesRes`

`클래스` `ObjectInterface`

Trace Server용 객체 API 엔드포인트입니다. 이 프로토콜은 더 깔끔하고 RESTful한 인터페이스를 제공하는 객체 관리 API를 정의합니다. 이전 버전과의 호환성을 유지하려면 구현체는 이 프로토콜과 TraceServerInterface를 모두 지원해야 합니다.

`방법` `call_end_v2`

call_end_v2(req: CallEndV2Req) → CallEndV2Res

`방법` `call_start_v2`

call_start_v2(req: CallStartV2Req) → CallStartV2Res

`방법` `calls_complete`

calls_complete(req: CallsUpsertCompleteReq) → CallsUpsertCompleteRes

`방법` `dataset_create`

dataset_create(req: DatasetCreateReq) → DatasetCreateRes

`방법` `dataset_delete`

dataset_delete(req: DatasetDeleteReq) → DatasetDeleteRes

`방법` `dataset_list`

dataset_list(req: DatasetListReq) → Iterator[DatasetReadRes]

`방법` `dataset_read`

dataset_read(req: DatasetReadReq) → DatasetReadRes

`방법` `eval_results_query`

eval_results_query(req: EvalResultsQueryReq) → EvalResultsQueryRes

`방법` `evaluation_create`

evaluation_create(req: EvaluationCreateReq) → EvaluationCreateRes

`방법` `evaluation_delete`

evaluation_delete(req: EvaluationDeleteReq) → EvaluationDeleteRes

`방법` `evaluation_list`

evaluation_list(req: EvaluationListReq) → Iterator[EvaluationReadRes]

`방법` `evaluation_read`

evaluation_read(req: EvaluationReadReq) → EvaluationReadRes

`방법` `evaluation_run_create`

evaluation_run_create(req: EvaluationRunCreateReq) → EvaluationRunCreateRes

`방법` `evaluation_run_delete`

evaluation_run_delete(req: EvaluationRunDeleteReq) → EvaluationRunDeleteRes

`방법` `evaluation_run_finish`

evaluation_run_finish(req: EvaluationRunFinishReq) → EvaluationRunFinishRes

`방법` `evaluation_run_list`

evaluation_run_list(req: EvaluationRunListReq) → Iterator[EvaluationRunReadRes]

`방법` `evaluation_run_read`

evaluation_run_read(req: EvaluationRunReadReq) → EvaluationRunReadRes

`방법` `model_create`

model_create(req: ModelCreateReq) → ModelCreateRes

`방법` `model_delete`

model_delete(req: ModelDeleteReq) → ModelDeleteRes

`방법` `model_list`

model_list(req: ModelListReq) → Iterator[ModelReadRes]

`방법` `model_read`

model_read(req: ModelReadReq) → ModelReadRes

`방법` `op_create`

op_create(req: OpCreateReq) → OpCreateRes

`방법` `op_delete`

op_delete(req: OpDeleteReq) → OpDeleteRes

`방법` `op_list`

op_list(req: OpListReq) → Iterator[OpReadRes]

`방법` `op_read`

op_read(req: OpReadReq) → OpReadRes

`방법` `prediction_create`

prediction_create(req: PredictionCreateReq) → PredictionCreateRes

`방법` `prediction_delete`

prediction_delete(req: PredictionDeleteReq) → PredictionDeleteRes

`방법` `prediction_finish`

prediction_finish(req: PredictionFinishReq) → PredictionFinishRes

`방법` `prediction_list`

prediction_list(req: PredictionListReq) → Iterator[PredictionReadRes]

`방법` `prediction_read`

prediction_read(req: PredictionReadReq) → PredictionReadRes

`방법` `score_create`

score_create(req: ScoreCreateReq) → ScoreCreateRes

`방법` `score_delete`

score_delete(req: ScoreDeleteReq) → ScoreDeleteRes

`방법` `score_list`

score_list(req: ScoreListReq) → Iterator[ScoreReadRes]

`방법` `score_read`

score_read(req: ScoreReadReq) → ScoreReadRes

`방법` `scorer_create`

scorer_create(req: ScorerCreateReq) → ScorerCreateRes

`방법` `scorer_delete`

scorer_delete(req: ScorerDeleteReq) → ScorerDeleteRes

`방법` `scorer_list`

scorer_list(req: ScorerListReq) → Iterator[ScorerReadRes]

`방법` `scorer_read`

scorer_read(req: ScorerReadReq) → ScorerReadRes

`클래스` `ObjectVersionFilter`

Pydantic 필드:

base_object_classes: list[str] | None
exclude_base_object_classes: list[str] | None
leaf_object_classes: list[str] | None
object_ids: list[str] | None
is_op: bool | None
latest_only: bool | None
tags: list[str] | None
aliases: list[str] | None

`클래스` `OpCreateBody`

REST API를 통해 Op 객체를 생성할 때 사용하는 요청 본문입니다. 이 모델에는 project_id가 포함되지 않습니다. project_id는 RESTful 엔드포인트의 URL 경로에서 가져오기 때문입니다. Pydantic 필드:

name: str | None
source_code: str | None

`클래스` `OpCreateReq`

Op 객체를 생성하는 요청 모델입니다. 내부 API에서 사용하기 위해 project_id를 추가해 OpCreateBody를 확장한 모델입니다. Pydantic Fields:

name: str | None
source_code: str | None
project_id: <class 'str'>
wb_user_id: str | None

`클래스` `OpCreateRes`

Op 객체를 생성할 때 사용하는 응답 모델입니다. Pydantic 필드:

digest: <class 'str'>
object_id: <class 'str'>
version_index: <class 'int'>

`클래스` `OpDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None
wb_user_id: str | None

`클래스` `OpDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`클래스` `OpListReq`

Pydantic 필드:

project_id: <class 'str'>
limit: int | None
offset: int | None
wb_user_id: str | None

`클래스` `OpReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
wb_user_id: str | None

`클래스` `OpReadRes`

Op 객체 조회용 응답 모델입니다. code 필드에는 op의 실제 소스 코드가 포함됩니다. Pydantic 필드:

object_id: <class 'str'>
digest: <class 'str'>
version_index: <class 'int'>
created_at: <class 'datetime.datetime'>
code: <class 'str'>

`클래스` `PredictionCreateBody`

REST API를 통해 Prediction을 생성할 때 사용하는 요청 본문입니다. 이 모델은 project_id를 제외합니다. project_id는 RESTful 엔드포인트의 URL 경로에서 전달되기 때문입니다. Pydantic 필드:

model: <class 'str'>
inputs: dict[str, typing.Any]
output: typing.Any
evaluation_run_id: str | None

`클래스` `PredictionCreateReq`

Prediction 생성 요청 모델입니다. 내부 API 사용을 위해 project_id를 추가해 PredictionCreateBody를 확장한 모델입니다. Pydantic 필드:

model: <class 'str'>
inputs: dict[str, typing.Any]
output: typing.Any
evaluation_run_id: str | None
project_id: <class 'str'>
wb_user_id: str | None

`클래스` `PredictionCreateRes`

Pydantic 필드:

prediction_id: <class 'str'>

`클래스` `PredictionDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
prediction_ids: list[str]
wb_user_id: str | None

`클래스` `PredictionDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`클래스` `PredictionFinishReq`

Pydantic 필드:

project_id: <class 'str'>
prediction_id: <class 'str'>
wb_user_id: str | None

`클래스` `PredictionFinishRes`

Pydantic 필드:

success: <class 'bool'>

`클래스` `PredictionListReq`

Pydantic 필드:

project_id: <class 'str'>
evaluation_run_id: str | None
limit: int | None
offset: int | None
wb_user_id: str | None

`클래스` `PredictionListRes`

Pydantic 필드:

predictions: list[PredictionReadRes]

`클래스` `PredictionReadReq`

Pydantic 필드:

project_id: <class 'str'>
prediction_id: <class 'str'>
wb_user_id: str | None

`클래스` `PredictionReadRes`

Pydantic 필드:

prediction_id: <class 'str'>
model: <class 'str'>
inputs: dict[str, typing.Any]
output: typing.Any
evaluation_run_id: str | None
wb_user_id: str | None

`클래스` `ProcessedResourceSpans`

Pydantic 필드:

entity: <class 'str'>
project: <class 'str'>
run_id: str | None
resource_spans: typing.Any

`class` `ProjectStatsReq`

Pydantic 필드:

project_id: <class 'str'>
include_trace_storage_size: bool | None
include_object_storage_size: bool | None
include_table_storage_size: bool | None
include_file_storage_size: bool | None

`class` `ProjectStatsRes`

Pydantic 필드:

trace_storage_size_bytes: <class 'int'>
objects_storage_size_bytes: <class 'int'>
tables_storage_size_bytes: <class 'int'>
files_storage_size_bytes: <class 'int'>

`class` `RefsReadBatchReq`

Pydantic 필드:

refs: list[str]

`class` `RefsReadBatchRes`

Pydantic 필드:

vals: list[typing.Any]

`class` `ScoreCreateBody`

REST API를 통해 Score를 생성할 때 사용하는 요청 본문입니다. 이 모델에는 project_id가 포함되지 않습니다. project_id는 RESTful 엔드포인트의 URL 경로에서 가져오기 때문입니다. Pydantic 필드:

prediction_id: <class 'str'>
scorer: <class 'str'>
value: <class 'float'>
evaluation_run_id: str | None

`class` `ScoreCreateReq`

Score를 생성하는 요청 모델입니다. 내부 API에서 사용하기 위해 project_id를 추가해 ScoreCreateBody를 확장합니다. Pydantic 필드:

prediction_id: <class 'str'>
scorer: <class 'str'>
value: <class 'float'>
evaluation_run_id: str | None
project_id: <class 'str'>
wb_user_id: str | None

`class` `ScoreCreateRes`

Pydantic 필드:

score_id: <class 'str'>

`class` `ScoreDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
score_ids: list[str]
wb_user_id: str | None

`class` `ScoreDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `ScoreListReq`

Pydantic 필드:

project_id: <class 'str'>
evaluation_run_id: str | None
limit: int | None
offset: int | None
wb_user_id: str | None

`class` `ScoreReadReq`

Pydantic 필드:

project_id: <class 'str'>
score_id: <class 'str'>
wb_user_id: str | None

`class` `ScoreReadRes`

Pydantic 필드:

score_id: <class 'str'>
scorer: <class 'str'>
value: <class 'float'>
evaluation_run_id: str | None
wb_user_id: str | None

`class` `ScorerCreateBody`

Pydantic 필드:

name: <class 'str'>
description: str | None
op_source_code: <class 'str'>

`class` `ScorerCreateReq`

Pydantic 필드:

name: <class 'str'>
description: str | None
op_source_code: <class 'str'>
project_id: <class 'str'>
wb_user_id: str | None

`class` `ScorerCreateRes`

Pydantic 필드:

digest: <class 'str'>
object_id: <class 'str'>
version_index: <class 'int'>
scorer: <class 'str'>

`class` `ScorerDeleteReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digests: list[str] | None
wb_user_id: str | None

`class` `ScorerDeleteRes`

Pydantic 필드:

num_deleted: <class 'int'>

`class` `ScorerListReq`

Pydantic 필드:

project_id: <class 'str'>
limit: int | None
offset: int | None
wb_user_id: str | None

`class` `ScorerReadReq`

Pydantic 필드:

project_id: <class 'str'>
object_id: <class 'str'>
digest: <class 'str'>
wb_user_id: str | None

`class` `ScorerReadRes`

Pydantic 필드:

object_id: <class 'str'>
digest: <class 'str'>
version_index: <class 'int'>
created_at: <class 'datetime.datetime'>
name: <class 'str'>
description: str | None
score_op: <class 'str'>

`class` `StartedCallSchemaForInsert`

Pydantic 필드:

project_id: <class 'str'>
id: str | None
op_name: <class 'str'>
display_name: str | None
trace_id: str | None
parent_id: str | None
thread_id: str | None
turn_id: str | None
started_at: <class 'datetime.datetime'>
attributes: dict[str, typing.Any]
inputs: dict[str, typing.Any]
otel_dump: dict[str, typing.Any] | None
wb_user_id: str | None
wb_run_id: str | None
wb_run_step: int | None

`class` `SummaryInsertMap`

`class` `SummaryMap`

`class` `TableAppendSpec`

Pydantic 필드:

append: <class 'TableAppendSpecPayload'>

`class` `TableAppendSpecPayload`

Pydantic 필드:

row: dict[str, typing.Any]

`class` `TableCreateFromDigestsReq`

Pydantic 필드:

project_id: <class 'str'>
row_digests: list[str]
expected_digest: str | None

`class` `TableCreateFromDigestsRes`

Pydantic 필드:

digest: <class 'str'>

`class` `TableCreateReq`

Pydantic 필드:

table: <class 'TableSchemaForInsert'>

`class` `TableCreateRes`

Pydantic 필드:

digest: <class 'str'>
row_digests: list[str]

`class` `TableInsertSpec`

Pydantic 필드:

insert: <class 'TableInsertSpecPayload'>

`class` `TableInsertSpecPayload`

Pydantic 필드:

index: <class 'int'>
row: dict[str, typing.Any]

`class` `TablePopSpec`

Pydantic 필드:

pop: <class 'TablePopSpecPayload'>

`class` `TablePopSpecPayload`

Pydantic 필드:

index: <class 'int'>

`class` `TableQueryReq`

Pydantic 필드:

project_id: <class 'str'>
digest: <class 'str'>
filter: TableRowFilter | None
limit: int | None
offset: int | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None

`class` `TableQueryRes`

Pydantic 필드:

rows: list[TableRowSchema]

`class` `TableQueryStatsBatchReq`

Pydantic 필드:

project_id: <class 'str'>
digests: list[str] | None
include_storage_size: bool | None

`class` `TableQueryStatsBatchRes`

Pydantic 필드:

tables: list[TableStatsRow]

`class` `TableQueryStatsReq`

Pydantic 필드:

project_id: <class 'str'>
digest: <class 'str'>

`class` `TableQueryStatsRes`

Pydantic 필드:

count: <class 'int'>

`class` `TableRowFilter`

Pydantic 필드:

row_digests: list[str] | None

`class` `TableRowSchema`

Pydantic 필드:

digest: <class 'str'>
val: typing.Any
original_index: int | None

`class` `TableSchemaForInsert`

Pydantic 필드:

project_id: <class 'str'>
rows: list[dict[str, typing.Any]]
expected_digest: str | None

`class` `TableStatsRow`

Pydantic 필드:

count: <class 'int'>
digest: <class 'str'>
storage_size_bytes: int | None

`class` `TableUpdateReq`

Pydantic 필드:

project_id: <class 'str'>
base_digest: <class 'str'>
updates: list[TableAppendSpec | TablePopSpec | TableInsertSpec]

`class` `TableUpdateRes`

Pydantic 필드:

digest: <class 'str'>
updated_row_digests: list[str]

`class` `TagsListReq`

Pydantic 필드:

project_id: <class 'str'>
wb_user_id: str | None

`class` `TagsListRes`

Pydantic 필드:

tags: list[str]

`class` `ThreadSchema`

Pydantic 필드:

thread_id: <class 'str'>
turn_count: <class 'int'>
start_time: <class 'datetime.datetime'>
last_updated: <class 'datetime.datetime'>
first_turn_id: str | None
last_turn_id: str | None
p50_turn_duration_ms: float | None
p99_turn_duration_ms: float | None

`class` `ThreadsQueryFilter`

Pydantic 필드:

after_datetime: datetime.datetime | None
before_datetime: datetime.datetime | None
thread_ids: list[str] | None

`class` `ThreadsQueryReq`

turn call만 기준으로 집계한 통계와 함께 스레드를 쿼리합니다. turn call은 thread context의 바로 아래 자식 call입니다(call.id == turn_id인 경우). 이렇게 하면 중첩된 모든 구현 세부 사항을 포함하지 않고, 대화 수준에서 의미 있는 통계를 제공할 수 있습니다. Pydantic 필드:

project_id: <class 'str'>
filter: ThreadsQueryFilter | None
limit: int | None
offset: int | None
sort_by: list[weave.trace_server.common_interface.SortBy] | None

`class` `TraceServerInterface`

`방법` `actions_execute_batch`

actions_execute_batch(req: ActionsExecuteBatchReq) → ActionsExecuteBatchRes

`방법` `aliases_list`

aliases_list(req: AliasesListReq) → AliasesListRes

`방법` `annotation_queue_add_calls`

annotation_queue_add_calls(
    req: AnnotationQueueAddCallsReq
) → AnnotationQueueAddCallsRes

`방법` `annotation_queue_create`

annotation_queue_create(
    req: AnnotationQueueCreateReq
) → AnnotationQueueCreateRes

`방법` `annotation_queue_delete`

annotation_queue_delete(
    req: AnnotationQueueDeleteReq
) → AnnotationQueueDeleteRes

`방법` `annotation_queue_items_query`

annotation_queue_items_query(
    req: AnnotationQueueItemsQueryReq
) → AnnotationQueueItemsQueryRes

`방법` `annotation_queue_read`

annotation_queue_read(req: AnnotationQueueReadReq) → AnnotationQueueReadRes

`방법` `annotation_queue_update`

annotation_queue_update(
    req: AnnotationQueueUpdateReq
) → AnnotationQueueUpdateRes

`방법` `annotation_queues_query_stream`

annotation_queues_query_stream(
    req: AnnotationQueuesQueryReq
) → Iterator[AnnotationQueueSchema]

`방법` `annotation_queues_stats`

annotation_queues_stats(
    req: AnnotationQueuesStatsReq
) → AnnotationQueuesStatsRes

`방법` `annotator_queue_items_progress_update`

annotator_queue_items_progress_update(
    req: AnnotatorQueueItemsProgressUpdateReq
) → AnnotatorQueueItemsProgressUpdateRes

`방법` `call_end`

call_end(req: CallEndReq) → CallEndRes

`방법` `call_read`

call_read(req: CallReadReq) → CallReadRes

`방법` `call_start`

call_start(req: CallStartReq) → CallStartRes

`방법` `call_start_batch`

call_start_batch(req: CallCreateBatchReq) → CallCreateBatchRes

`방법` `call_stats`

call_stats(req: 'CallStatsReq') → CallStatsRes

`방법` `call_update`

call_update(req: CallUpdateReq) → CallUpdateRes

`방법` `calls_delete`

calls_delete(req: CallsDeleteReq) → CallsDeleteRes

`방법` `calls_query`

calls_query(req: CallsQueryReq) → CallsQueryRes

`방법` `calls_query_stats`

calls_query_stats(req: CallsQueryStatsReq) → CallsQueryStatsRes

`방법` `calls_query_stream`

calls_query_stream(req: CallsQueryReq) → Iterator[CallSchema]

`방법` `calls_score`

calls_score(req: CallsScoreReq) → CallsScoreRes

`방법` `calls_usage`

calls_usage(req: 'CallsUsageReq') → CallsUsageRes

`방법` `completions_create`

completions_create(req: CompletionsCreateReq) → CompletionsCreateRes

`방법` `completions_create_stream`

completions_create_stream(req: CompletionsCreateReq) → Iterator[dict[str, Any]]

`방법` `cost_create`

cost_create(req: CostCreateReq) → CostCreateRes

`방법` `cost_purge`

cost_purge(req: CostPurgeReq) → CostPurgeRes

`방법` `cost_query`

cost_query(req: CostQueryReq) → CostQueryRes

`방법` `evaluate_model`

evaluate_model(req: EvaluateModelReq) → EvaluateModelRes

`방법` `evaluation_status`

evaluation_status(req: EvaluationStatusReq) → EvaluationStatusRes

`방법` `feedback_create`

feedback_create(req: FeedbackCreateReq) → FeedbackCreateRes

`방법` `feedback_create_batch`

feedback_create_batch(req: FeedbackCreateBatchReq) → FeedbackCreateBatchRes

`방법` `feedback_payload_schema`

feedback_payload_schema(
    req: FeedbackPayloadSchemaReq
) → FeedbackPayloadSchemaRes

`방법` `feedback_purge`

feedback_purge(req: FeedbackPurgeReq) → FeedbackPurgeRes

`방법` `feedback_query`

feedback_query(req: FeedbackQueryReq) → FeedbackQueryRes

`방법` `feedback_replace`

feedback_replace(req: FeedbackReplaceReq) → FeedbackReplaceRes

`방법` `feedback_stats`

feedback_stats(req: FeedbackStatsReq) → FeedbackStatsRes

`방법` `file_content_read`

file_content_read(req: FileContentReadReq) → FileContentReadRes

`방법` `file_create`

file_create(req: FileCreateReq) → FileCreateRes

`방법` `files_stats`

files_stats(req: FilesStatsReq) → FilesStatsRes

`방법` `image_create`

image_create(req: ImageGenerationCreateReq) → ImageGenerationCreateRes

`방법` `obj_add_tags`

obj_add_tags(req: ObjAddTagsReq) → ObjAddTagsRes

`방법` `obj_create`

obj_create(req: ObjCreateReq) → ObjCreateRes

`방법` `obj_delete`

obj_delete(req: ObjDeleteReq) → ObjDeleteRes

`방법` `obj_read`

obj_read(req: ObjReadReq) → ObjReadRes

`방법` `obj_remove_aliases`

obj_remove_aliases(req: ObjRemoveAliasesReq) → ObjRemoveAliasesRes

`방법` `obj_remove_tags`

obj_remove_tags(req: ObjRemoveTagsReq) → ObjRemoveTagsRes

`방법` `obj_set_aliases`

obj_set_aliases(req: ObjSetAliasesReq) → ObjSetAliasesRes

`방법` `objs_query`

objs_query(req: ObjQueryReq) → ObjQueryRes

`방법` `otel_export`

otel_export(req: OTelExportReq) → OTelExportRes

`방법` `project_stats`

project_stats(req: ProjectStatsReq) → ProjectStatsRes

`방법` `refs_read_batch`

refs_read_batch(req: RefsReadBatchReq) → RefsReadBatchRes

`방법` `table_create`

table_create(req: TableCreateReq) → TableCreateRes

`방법` `table_create_from_digests`

table_create_from_digests(
    req: TableCreateFromDigestsReq
) → TableCreateFromDigestsRes

`방법` `table_query`

table_query(req: TableQueryReq) → TableQueryRes

`방법` `table_query_stats`

table_query_stats(req: TableQueryStatsReq) → TableQueryStatsRes

`방법` `table_query_stats_batch`

table_query_stats_batch(req: TableQueryStatsBatchReq) → TableQueryStatsBatchRes

`방법` `table_query_stream`

table_query_stream(req: TableQueryReq) → Iterator[TableRowSchema]

`방법` `table_update`

table_update(req: TableUpdateReq) → TableUpdateRes

`방법` `tags_list`

tags_list(req: TagsListReq) → TagsListRes

`방법` `threads_query_stream`

threads_query_stream(req: ThreadsQueryReq) → Iterator[ThreadSchema]

`방법` `trace_usage`

trace_usage(req: 'TraceUsageReq') → TraceUsageRes

`class` `TraceStatus`

`class` `TraceUsageReq`

하위 호출까지 롤업하여 트레이스의 호출별 사용량을 계산하기 위한 요청입니다. 이 엔드포인트는 트레이스의 각 호출에 대한 사용 메트릭을 반환하며, 각 호출의 메트릭에는 해당 호출 자체의 사용량과 모든 하위 호출의 사용량 합계가 포함됩니다. 호출별로 롤업된 메트릭을 확인하려는 트레이스 뷰에 사용하세요. 참고: 집계를 위해 일치하는 모든 호출을 메모리에 로드합니다. 결과 세트가 매우 큰 경우(호출 수 10k 초과), 더 구체적인 필터를 사용하거나 애플리케이션 계층에서 페이지네이션을 사용하는 것을 고려하세요. Pydantic 필드:

project_id: <class 'str'>
filter: CallsFilter | None
query: weave.trace_server.interface.query.Query | None
include_costs: <class 'bool'>
limit: <class 'int'>

`class` `TraceUsageRes`

호출별 사용 메트릭이 포함된 응답입니다(각 메트릭에는 하위 호출의 기여분도 포함됩니다). Pydantic 필드:

call_usage: dict[str, dict[str, LLMAggregatedUsage]]
unfinished_call_ids: list[str]

`class` `UsageMetricSpec`

모델별로 그룹화하여 집계할 사용 메트릭 사양입니다. Pydantic 필드:

metric: typing.Literal['input_tokens', 'output_tokens', 'total_tokens', 'input_cost', 'output_cost', 'total_cost']
aggregations: list[AggregationType]
percentiles: list[float]

Get Started

Guides

Cookbooks

Reference

Details & Support

​API Overview

​class ActionsExecuteBatchReq

​class ActionsExecuteBatchRes

​class AggregationType

​class AliasesListReq

​class AliasesListRes

​class AnnotationQueueAddCallsReq

​class AnnotationQueueAddCallsRes

​class AnnotationQueueCreateReq

​class AnnotationQueueCreateRes

​class AnnotationQueueDeleteReq

​class AnnotationQueueDeleteRes

​class AnnotationQueueItemSchema

​class AnnotationQueueItemsQueryReq

​class AnnotationQueueItemsQueryRes

​class AnnotationQueueReadReq

​class AnnotationQueueReadRes

​class AnnotationQueueSchema

​class AnnotationQueueStatsSchema

​class AnnotationQueueUpdateReq

​class AnnotationQueueUpdateRes

​class AnnotationQueuesQueryReq

​class AnnotationQueuesQueryRes

​class AnnotationQueuesStatsReq

​class AnnotationQueuesStatsRes

​class AnnotatorQueueItemsProgressUpdateReq

​class AnnotatorQueueItemsProgressUpdateRes

​class CallBatchEndMode

​class CallBatchStartMode

​class CallCreateBatchReq

​class CallCreateBatchRes

​class CallEndReq

​class CallEndRes

​class CallEndV2Req

​class CallEndV2Res

​class CallMetricSpec

​class CallReadReq

​class CallReadRes

​class CallSchema

​방법 serialize_typed_dicts

​class CallStartReq

​class CallStartRes

​class CallStartV2Req

​class CallStartV2Res

​class CallStatsReq

​방법 validate_date_range

​class CallStatsRes

​class CallUpdateReq

​class CallUpdateRes

​class CallsDeleteReq

​class CallsDeleteRes

​class CallsFilter

​class CallsQueryReq

​class CallsQueryRes

​class CallsQueryStatsReq

​class CallsQueryStatsRes

​class CallsScoreReq

​class CallsScoreRes

​class CallsUpsertCompleteReq

​class CallsUpsertCompleteRes

​class CallsUsageReq

​class CallsUsageRes

​class CompletedCallSchemaForInsert

​방법 serialize_typed_dicts

​class CompletionsCreateReq

​class CompletionsCreateRequestInputs

​class CompletionsCreateRes

​class CostCreateInput

​class CostCreateReq

​class CostCreateRes

​class CostPurgeReq

​class CostPurgeRes

​class CostQueryOutput

​class CostQueryReq

​class CostQueryRes

API Overview

`class` `ActionsExecuteBatchReq`

`class` `ActionsExecuteBatchRes`

`class` `AggregationType`

`class` `AliasesListReq`

`class` `AliasesListRes`

`class` `AnnotationQueueAddCallsReq`

`class` `AnnotationQueueAddCallsRes`

`class` `AnnotationQueueCreateReq`

`class` `AnnotationQueueCreateRes`

`class` `AnnotationQueueDeleteReq`

`class` `AnnotationQueueDeleteRes`

`class` `AnnotationQueueItemSchema`

`class` `AnnotationQueueItemsQueryReq`

`class` `AnnotationQueueItemsQueryRes`

`class` `AnnotationQueueReadReq`

`class` `AnnotationQueueReadRes`

`class` `AnnotationQueueSchema`

`class` `AnnotationQueueStatsSchema`

`class` `AnnotationQueueUpdateReq`

`class` `AnnotationQueueUpdateRes`

`class` `AnnotationQueuesQueryReq`

`class` `AnnotationQueuesQueryRes`

`class` `AnnotationQueuesStatsReq`

`class` `AnnotationQueuesStatsRes`

`class` `AnnotatorQueueItemsProgressUpdateReq`

`class` `AnnotatorQueueItemsProgressUpdateRes`

`class` `CallBatchEndMode`

`class` `CallBatchStartMode`

`class` `CallCreateBatchReq`

`class` `CallCreateBatchRes`

`class` `CallEndReq`

`class` `CallEndRes`

`class` `CallEndV2Req`

`class` `CallEndV2Res`

`class` `CallMetricSpec`

`class` `CallReadReq`

`class` `CallReadRes`

`class` `CallSchema`

`방법` `serialize_typed_dicts`

`class` `CallStartReq`

`class` `CallStartRes`

`class` `CallStartV2Req`

`class` `CallStartV2Res`

`class` `CallStatsReq`

`방법` `validate_date_range`

`class` `CallStatsRes`

`class` `CallUpdateReq`

`class` `CallUpdateRes`

`class` `CallsDeleteReq`

`class` `CallsDeleteRes`

`class` `CallsFilter`

`class` `CallsQueryReq`

`class` `CallsQueryRes`

`class` `CallsQueryStatsReq`

`class` `CallsQueryStatsRes`

`class` `CallsScoreReq`

`class` `CallsScoreRes`

`class` `CallsUpsertCompleteReq`

`class` `CallsUpsertCompleteRes`

`class` `CallsUsageReq`

`class` `CallsUsageRes`

`class` `CompletedCallSchemaForInsert`

`방법` `serialize_typed_dicts`

`class` `CompletionsCreateReq`

`class` `CompletionsCreateRequestInputs`

`class` `CompletionsCreateRes`

`class` `CostCreateInput`

`class` `CostCreateReq`

`class` `CostCreateRes`

`class` `CostPurgeReq`

`class` `CostPurgeRes`

`class` `CostQueryOutput`

`class` `CostQueryReq`

`class` `CostQueryRes`

`class` `DatasetCreateBody`

`class` `DatasetCreateReq`

`class` `DatasetCreateRes`

`class` `DatasetDeleteReq`

`class` `DatasetDeleteRes`