sam.qwenTraining / trainer /input_router.py
AAJerry's picture
Rebuild SAMS v4 bounded selector and strict 40-case gate
ceb51d3
Raw History Blame Contribute Delete
2.41 kB
from __future__ import annotations
from dataclasses import dataclass
import re
from typing import Any
LOW_CONFIDENCE_THRESHOLD = 0.78
@dataclass(frozen=True)
class RouteResult:
call_model: bool
reason: str
speech: str | None = None
next_question_id: str | None = None
_TARGETED_CONFIRMATIONS: tuple[tuple[re.Pattern[str], str, str], ...] = (
(
re.compile(r"\b(?:dizz|busy|spinn|balance)\w*\b", re.I),
"The audio may have mentioned dizziness or balance. Are you feeling dizzy—yes or no?",
"symptom.dizziness.present",
),
(
re.compile(r"\b(?:breath|breathe|air|chest)\w*\b", re.I),
"The audio may have mentioned breathing or your chest. Are you having unusual difficulty breathing—yes or no?",
"symptom.breathing_difficulty.present",
),
(
re.compile(r"\b(?:watch|wash|wearable|sensor|connect)\w*\b", re.I),
"The audio may have mentioned your wearable connection. Is your wearable connected right now?",
"device.connection_status",
),
(
re.compile(r"\b(?:head|headache|pain)\w*\b", re.I),
"The audio may have mentioned head pain. Do you have a headache—yes or no?",
"symptom.headache.present",
),
)
def route_request(request: dict[str, Any]) -> RouteResult:
"""Bypass the LLM when the speech signal is not reliable enough to interpret."""
if request.get("task_type") != "climber_checkin":
return RouteResult(True, "non_speech_or_robot_request")
transcript = request.get("transcript")
if not isinstance(transcript, dict):
return RouteResult(True, "no_transcript_gate")
confidence = transcript.get("confidence")
text = str(transcript.get("text", "")).strip()
polarity_ambiguous = bool(transcript.get("polarity_ambiguous", False))
if isinstance(confidence, (int, float)) and confidence >= LOW_CONFIDENCE_THRESHOLD and not polarity_ambiguous:
return RouteResult(True, "speech_confidence_passed")
for pattern, speech, question_id in _TARGETED_CONFIRMATIONS:
if pattern.search(text):
return RouteResult(False, "low_confidence_targeted_confirmation", speech, question_id)
return RouteResult(
False,
"low_confidence_short_repeat",
"The audio was unclear. Please repeat only your current symptom or status in one short sentence.",
None,
)