phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-07-29 15:23:51 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

kill reward scoring

Browse source
This commit is contained in:
Xi Yan 2024-10-22 08:48:08 -07:00
parent 02f4c3a081
commit 1dc2962a33
2 changed files with 0 additions and 62 deletions
Download patch file Download diff file Expand all files Collapse all files

7
llama_stack/apis/reward_scoring/__init__.py
View file

@ -1,7 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from .reward_scoring import * # noqa: F401 F403

55
llama_stack/apis/reward_scoring/reward_scoring.py
View file

@ -1,55 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import List, Protocol, Union
from llama_models.schema_utils import json_schema_type, webmethod
from pydantic import BaseModel
from llama_models.llama3.api.datatypes import * # noqa: F403
@json_schema_type
class ScoredMessage(BaseModel):
message: Message
score: float
@json_schema_type
class DialogGenerations(BaseModel):
dialog: List[Message]
sampled_generations: List[Message]
@json_schema_type
class ScoredDialogGenerations(BaseModel):
dialog: List[Message]
scored_generations: List[ScoredMessage]
@json_schema_type
class RewardScoringRequest(BaseModel):
"""Request to score a reward function. A list of prompts and a list of responses per prompt."""
dialog_generations: List[DialogGenerations]
model: str
@json_schema_type
class RewardScoringResponse(BaseModel):
"""Response from the reward scoring. Batch of (prompt, response, score) tuples that pass the threshold."""
scored_generations: List[ScoredDialogGenerations]
class RewardScoring(Protocol):
@webmethod(route="/reward_scoring/score")
def reward_score(
self,
dialog_generations: List[DialogGenerations],
model: str,
) -> Union[RewardScoringResponse]: ...