phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-07-16 09:58:10 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Update reward_model_scoring.yaml

Browse source 
add reward model identifier to the reward model spec
This commit is contained in:
Hardik Shah 2024-06-25 18:18:43 -07:00 • committed by GitHub
parent 26d780809c
commit 47390bc9a8
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 26 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

47
reward_model_scoring.yaml
View file

@ -100,27 +100,29 @@ components:
description: User provided metadata carried forward in the response.
paths:
/reward_model_scoring/:
get:
post:
summary: Score a pair of prompt and response using a reward model
description: Take a pair of prompt and responses and score them using a reward model
parameters:
- in: query
name: messages
schema:
type: array
items:
$ref: '#/components/schemas/Message'
required: true
- in: query
name: response
schema:
$ref: '#/components/schemas/Completion'
required: true
- in: query
name: scoring_function
schema:
type: string
required: true
requestBody:
required: true
content:
application/json:
schema:
type: object
properties:
reward_model:
type: string
description: Unique identifier for the reward model
messages:
type: array
items:
$ref: '#/components/schemas/Message'
description: List[Message] same as what is used in /chat_completion api
response:
$ref: '#/components/schemas/Completion'
description: Completion same as what is used in /chat_completion api
scoring_function:
type: string
responses:
'200':
description: Scoring completed successfully
@ -138,7 +140,7 @@ paths:
score:
type: number
format: float
/batch_reward_model_scoring/:
post:
summary: Batch score multiple pairs of prompts and responses
@ -150,7 +152,10 @@ paths:
schema:
type: object
properties:
# TODO: Maybe take local path but api first uploads and generates handle
reward_model:
type: string
description: Unique identifier for the reward model
# TODO: Maybe take local path and api first uploads and generates handle
# which is returned in the repsonse.
prompt_response_path:
type: string