phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-08-07 11:08:20 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

jobs eval scoring

Browse source
This commit is contained in:
Xi Yan 2025-03-13 11:47:42 -07:00
parent 36320728bf
commit 775e8514b7
4 changed files with 92 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

58
docs/_static/llama-stack-spec.html vendored
View file

@ -230,7 +230,7 @@
}
}
},
"/v1/eval/job/{job_id}/cancel": {
"/v1/eval/jobs/{job_id}/cancel": {
"post": {
"responses": {
"200": {
@ -280,7 +280,7 @@
]
}
},
"/v1/scoring/job/{job_id}/cancel": {
"/v1/scoring/jobs/{job_id}/cancel": {
"post": {
"responses": {
"200": {
@ -923,7 +923,7 @@
]
}
},
"/v1/eval/job/{job_id}": {
"/v1/eval/jobs/{job_id}": {
"get": {
"responses": {
"200": {
@ -1123,7 +1123,7 @@
]
}
},
"/v1/scoring/job/{job_id}": {
"/v1/scoring/jobs/{job_id}": {
"get": {
"responses": {
"200": {
@ -5160,26 +5160,36 @@
},
"type": {
"type": "string",
"const": "eval",
"default": "eval"
"enum": [
"batch_inference",
"scoring",
"evaluation",
"post_training"
],
"default": "evaluation",
"description": "The type of the job."
},
"result_files": {
"type": "array",
"items": {
"type": "string"
}
},
"description": "The file ids of the eval results."
},
"result_datasets": {
"type": "array",
"items": {
"type": "string"
}
},
"description": "The ids of the datasets containing the eval results."
},
"benchmark_id": {
"type": "string"
"type": "string",
"description": "The id of the benchmark to evaluate on."
},
"candidate": {
"$ref": "#/components/schemas/EvalCandidate"
"$ref": "#/components/schemas/EvalCandidate",
"description": "The candidate to evaluate on."
}
},
"additionalProperties": false,
@ -5193,7 +5203,8 @@
"benchmark_id",
"candidate"
],
"title": "EvalJob"
"title": "EvalJob",
"description": "An evaluation job."
},
"ModelCandidate": {
"type": "object",
@ -5399,29 +5410,39 @@
},
"type": {
"type": "string",
"const": "scoring",
"default": "scoring"
"enum": [
"batch_inference",
"scoring",
"evaluation",
"post_training"
],
"default": "scoring",
"description": "The type of the job."
},
"result_files": {
"type": "array",
"items": {
"type": "string"
}
},
"description": "The file ids of the scoring results."
},
"result_datasets": {
"type": "array",
"items": {
"type": "string"
}
},
"description": "The ids of the datasets containing the scoring results."
},
"dataset_id": {
"type": "string"
"type": "string",
"description": "The id of the dataset used for scoring."
},
"scoring_fn_ids": {
"type": "array",
"items": {
"type": "string"
}
},
"description": "The ids of the scoring functions used."
}
},
"additionalProperties": false,
@ -5435,7 +5456,8 @@
"dataset_id",
"scoring_fn_ids"
],
"title": "ScoringJob"
"title": "ScoringJob",
"description": "A scoring job."
},
"CancelTrainingJobRequest": {
"type": "object",

36
docs/_static/llama-stack-spec.yaml vendored
View file

@ -142,7 +142,7 @@ paths:
schema:
$ref: '#/components/schemas/BatchCompletionRequest'
required: true
/v1/eval/job/{job_id}/cancel:
/v1/eval/jobs/{job_id}/cancel:
post:
responses:
'200':
@ -173,7 +173,7 @@ paths:
required: true
schema:
type: string
/v1/scoring/job/{job_id}/cancel:
/v1/scoring/jobs/{job_id}/cancel:
post:
responses:
'200':
@ -622,7 +622,7 @@ paths:
required: true
schema:
type: string
/v1/eval/job/{job_id}:
/v1/eval/jobs/{job_id}:
get:
responses:
'200':
@ -756,7 +756,7 @@ paths:
required: true
schema:
type: string
/v1/scoring/job/{job_id}:
/v1/scoring/jobs/{job_id}:
get:
responses:
'200':
@ -3514,20 +3514,30 @@ components:
If status of the job is failed, this will contain the error message.
type:
type: string
const: eval
default: eval
enum:
- batch_inference
- scoring
- evaluation
- post_training
default: evaluation
description: The type of the job.
result_files:
type: array
items:
type: string
description: The file ids of the eval results.
result_datasets:
type: array
items:
type: string
description: >-
The ids of the datasets containing the eval results.
benchmark_id:
type: string
description: The id of the benchmark to evaluate on.
candidate:
$ref: '#/components/schemas/EvalCandidate'
description: The candidate to evaluate on.
additionalProperties: false
required:
- id
@ -3539,6 +3549,7 @@ components:
- benchmark_id
- candidate
title: EvalJob
description: An evaluation job.
ModelCandidate:
type: object
properties:
@ -3693,22 +3704,32 @@ components:
If status of the job is failed, this will contain the error message.
type:
type: string
const: scoring
enum:
- batch_inference
- scoring
- evaluation
- post_training
default: scoring
description: The type of the job.
result_files:
type: array
items:
type: string
description: The file ids of the scoring results.
result_datasets:
type: array
items:
type: string
description: >-
The ids of the datasets containing the scoring results.
dataset_id:
type: string
description: The id of the dataset used for scoring.
scoring_fn_ids:
type: array
items:
type: string
description: The ids of the scoring functions used.
additionalProperties: false
required:
- id
@ -3720,6 +3741,7 @@ components:
- dataset_id
- scoring_fn_ids
title: ScoringJob
description: A scoring job.
CancelTrainingJobRequest:
type: object
properties:

14
llama_stack/apis/eval/eval.py
View file

@ -10,7 +10,7 @@ from pydantic import BaseModel, Field
from typing_extensions import Annotated
from llama_stack.apis.agents import AgentConfig
from llama_stack.apis.common.job_types import CommonJobFields, JobStatus
from llama_stack.apis.common.job_types import CommonJobFields, JobType
from llama_stack.apis.inference import SamplingParams, SystemMessage
from llama_stack.apis.scoring import ScoringResult
from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
@ -63,7 +63,17 @@ class EvaluateResponse(BaseModel):
@json_schema_type
class EvalJob(CommonJobFields):
type: Literal["eval"] = "eval"
"""
An evaluation job.
:param type: The type of the job.
:param result_files: The file ids of the eval results.
:param result_datasets: The ids of the datasets containing the eval results.
:param benchmark_id: The id of the benchmark to evaluate on.
:param candidate: The candidate to evaluate on.
"""
type: JobType = JobType.evaluation.value
result_files: List[str] = Field(
description="The file ids of the eval results.",
default_factory=list,

12
llama_stack/apis/scoring/scoring.py
View file

@ -50,7 +50,17 @@ class ScoreResponse(BaseModel):
@json_schema_type
class ScoringJob(CommonJobFields):
type: Literal["scoring"] = "scoring"
"""
A scoring job.
:param type: The type of the job.
:param result_files: The file ids of the scoring results.
:param result_datasets: The ids of the datasets containing the scoring results.
:param dataset_id: The id of the dataset used for scoring.
:param scoring_fn_ids: The ids of the scoring functions used.
"""
type: JobType = JobType.scoring.value
result_files: List[str] = Field(
description="The file ids of the scoring results.",