feat(v1): 架构与算法优化二轮——共享 LLM 客户端去重、语义缓存 O(1) 提升/淘汰、分类器热路径清理
- 新增 router_system/llm_client.py：OpenAICompatClient 统一 experts/judge/fallback
三处复制的懒建 AsyncClient + /chat/completions + choices/usage 解析（~60 行去重）；
密钥解析统一走 config.get_api_key（激活原死代码，顺带消除 experts 默认环境名不一致）
- 语义缓存 L2：条目容器 list→OrderedDict（提升/淘汰 O(n)→O(1)），按 query 天然去重；
n-gram 向量 lru_cache 复用（同一次 miss 的 get/put 免重复分词）；
A/B：淘汰路径 0.040→0.034s，miss→put 往返 9.41→8.57s（-9%）
- RuleJudge 覆盖度：response.lower() 提出逐词循环（原 O(terms×len) 重复复制）
- extract_content_terms 纯函数 lru_cache 化（专家与 Judge 对同一查询免重复分词），返回 tuple
- RuleClassifier：_score 去掉败者领域白建的命中词 list（胜出后单独收集）；
修复 code 规则 ("api",0.7) 重复登记（原命中计 1.4 分）
- difficulty：正则模块级预编译
- tests：恢复上一轮引入的乱码中文 docstring；网关测试输入串恢复为可判 code 的中文查询
This commit is contained in:
tzt
+36
-19
@@ -12,11 +12,17 @@
|
||||
- 每条语义缓存条目在写入时预计算并缓存向量范数，查询时免重复计算（原来每对比较都重算）
|
||||
- 语义查找单遍完成：扫描即跟踪最优条目与命中计数，命中后不再二次线性查找
|
||||
- 相似度达到 1.0（完全相同查询）时提前终止扫描（余弦相似度上界，不可能更优）
|
||||
- 语义条目容器为 OrderedDict：提升/淘汰均为 O(1)（原 list pop 是 O(n) 搬移），
|
||||
且按 query 天然去重（原实现同一 query 可重复追加，重复条目白白参与扫描）
|
||||
- n-gram 向量经 lru_cache 复用：同一次未命中请求里 get 与紧随的 put
|
||||
不再重复做两遍分词+计数
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
from collections import OrderedDict
|
||||
from dataclasses import dataclass
|
||||
from functools import lru_cache
|
||||
from typing import Any, Dict, List, Optional, Tuple
|
||||
|
||||
|
||||
@@ -52,6 +58,17 @@ def _dot(vec_a: Dict[str, float], vec_b: Dict[str, float]) -> float:
|
||||
return sum(v * vec_b.get(k, 0.0) for k, v in vec_a.items())
|
||||
|
||||
|
||||
@lru_cache(maxsize=1024)
|
||||
def _embed(text: str) -> Tuple[Dict[str, float], float]:
|
||||
"""n-gram 向量 + 范数（带缓存）。
|
||||
|
||||
纯函数且同一次未命中请求会在 get/put 各用一次，缓存可省一遍分词。
|
||||
契约：返回的 dict 仅供只读（内部无任何修改点），调用方不得改写。
|
||||
"""
|
||||
vec = _tf_vector(_ngrams(text))
|
||||
return vec, _norm(vec)
|
||||
|
||||
|
||||
class RouterCache:
|
||||
"""L1 精确缓存 + L2 语义缓存。"""
|
||||
|
||||
@@ -63,7 +80,8 @@ class RouterCache:
|
||||
self.max_exact = max_exact
|
||||
self.max_semantic = max_semantic
|
||||
self._exact: Dict[str, CacheEntry] = {}
|
||||
self._semantic: List[Tuple[str, CacheEntry]] = [] # (query, entry)
|
||||
# (query -> entry)，插入序即淘汰序；len 兼容原 list 断言
|
||||
self._semantic: "OrderedDict[str, CacheEntry]" = OrderedDict()
|
||||
self._sem_vecs: Dict[str, Dict[str, float]] = {}
|
||||
self._sem_norms: Dict[str, float] = {} # 预计算范数，避免查询期重算
|
||||
self.hits = {"exact": 0, "semantic": 0}
|
||||
@@ -78,57 +96,56 @@ class RouterCache:
|
||||
return ("exact", entry.result)
|
||||
|
||||
if self.semantic_enabled:
|
||||
q_vec = _tf_vector(_ngrams(query))
|
||||
q_norm = _norm(q_vec)
|
||||
q_vec, q_norm = _embed(query)
|
||||
best_sim = 0.0
|
||||
best_idx = -1
|
||||
best_q = ""
|
||||
if q_norm > 0.0:
|
||||
# 单遍扫描：同时跟踪最优相似度与条目位置
|
||||
for i, (q, _e) in enumerate(self._semantic):
|
||||
# 单遍扫描（插入序）：同时跟踪最优条目，sim=1.0 提前终止
|
||||
for q, _e in self._semantic.items():
|
||||
n_q = self._sem_norms.get(q, 0.0)
|
||||
if n_q <= 0.0:
|
||||
continue
|
||||
sim = _dot(q_vec, self._sem_vecs.get(q, {})) / (q_norm * n_q)
|
||||
if sim > best_sim:
|
||||
best_sim = sim
|
||||
best_idx = i
|
||||
best_q = q
|
||||
if sim >= 1.0:
|
||||
break # 余弦相似度上界：完全相同查询，提前终止
|
||||
if best_idx >= 0 and best_sim >= self.similarity_threshold:
|
||||
best_q, best_entry = self._semantic[best_idx]
|
||||
if best_q and best_sim >= self.similarity_threshold:
|
||||
best_entry = self._semantic[best_q]
|
||||
# 完全相同查询（相似度=1.0）计为 exact 命中
|
||||
is_exact = best_sim >= 0.999
|
||||
level = "exact" if is_exact else "semantic"
|
||||
self.hits[level] += 1
|
||||
self._bump_semantic(best_idx, best_q, best_entry)
|
||||
self._bump_semantic(best_q, best_entry)
|
||||
return (level, best_entry.result)
|
||||
|
||||
self.misses += 1
|
||||
return None
|
||||
|
||||
def _bump_semantic(self, idx: int, query: str, entry: CacheEntry):
|
||||
"""语义命中：累计命中次数，达到阈值提升为精确缓存（O(1)，无需二次查找）。"""
|
||||
def _bump_semantic(self, query: str, entry: CacheEntry):
|
||||
"""语义命中：累计命中次数，达到阈值提升为精确缓存（OrderedDict 删除 O(1)）。"""
|
||||
entry.hits += 1
|
||||
if entry.hits >= self.promote_frequency:
|
||||
self._exact[query] = entry
|
||||
self._semantic.pop(idx)
|
||||
del self._semantic[query]
|
||||
self._sem_vecs.pop(query, None)
|
||||
self._sem_norms.pop(query, None)
|
||||
|
||||
# ---- 写入 ----
|
||||
def put(self, query: str, result: Dict[str, Any]):
|
||||
if query in self._exact:
|
||||
return
|
||||
if query in self._exact or query in self._semantic:
|
||||
return # 已缓存（语义区按 query 天然去重，避免重复条目参与扫描）
|
||||
entry = CacheEntry(result=result)
|
||||
if self.semantic_enabled:
|
||||
if len(self._semantic) >= self.max_semantic:
|
||||
old_q, _ = self._semantic.pop(0)
|
||||
old_q, _old = self._semantic.popitem(last=False) # O(1) 淘汰最旧
|
||||
self._sem_vecs.pop(old_q, None)
|
||||
self._sem_norms.pop(old_q, None)
|
||||
self._semantic.append((query, entry))
|
||||
vec = _tf_vector(_ngrams(query))
|
||||
self._semantic[query] = entry
|
||||
vec, norm = _embed(query) # 与 get miss 共享同一份缓存向量
|
||||
self._sem_vecs[query] = vec
|
||||
self._sem_norms[query] = _norm(vec)
|
||||
self._sem_norms[query] = norm
|
||||
else:
|
||||
self._exact[query] = entry
|
||||
if len(self._exact) > self.max_exact:
|
||||
|
||||
+15
-10
@@ -33,7 +33,7 @@ DOMAIN_RULES: Dict[str, List[Tuple[str, float]]] = {
|
||||
("algorithm", 0.8), ("sort", 0.7), ("array", 0.7), ("regex", 0.8),
|
||||
("import", 0.7), ("loop", 0.7), ("recursion", 0.8), ("refactor", 0.8),
|
||||
("deploy", 0.8), ("docker", 0.8), ("kubernetes", 0.8),
|
||||
("async", 0.7), ("flask", 0.7), ("django", 0.7), ("api", 0.7),
|
||||
("async", 0.7), ("flask", 0.7), ("django", 0.7),
|
||||
("索引", 0.8), ("优化", 0.7), ("查询", 0.6),
|
||||
],
|
||||
"math": [
|
||||
@@ -98,24 +98,29 @@ class RuleClassifier(BaseClassifier):
|
||||
self.confidence_floor = confidence_floor
|
||||
self.rules = DOMAIN_RULES
|
||||
|
||||
def _score(self, query: str) -> Tuple[Dict[str, float], Dict[str, List[str]]]:
|
||||
q = query.lower()
|
||||
def _score(self, q: str) -> Dict[str, float]:
|
||||
"""对已 lowercase 的查询按领域规则打分，返回有命中的领域分数。
|
||||
|
||||
命中词列表只对最终胜出领域有意义（classify 的唯一消费点），
|
||||
故不在各领域上白建 list——胜出后由 _matched_rules 单独收集。
|
||||
"""
|
||||
scores: Dict[str, float] = {}
|
||||
matched: Dict[str, List[str]] = {}
|
||||
for domain, rules in self.rules.items():
|
||||
s = 0.0
|
||||
hits = []
|
||||
for kw, w in rules:
|
||||
if kw in q:
|
||||
s += w
|
||||
hits.append(kw)
|
||||
if s > 0:
|
||||
scores[domain] = s
|
||||
matched[domain] = hits
|
||||
return scores, matched
|
||||
return scores
|
||||
|
||||
def _matched_rules(self, q: str, domain: str) -> List[str]:
|
||||
"""收集指定领域命中的关键词（保持登记顺序）。"""
|
||||
return [kw for kw, _w in self.rules.get(domain, ()) if kw in q]
|
||||
|
||||
def classify(self, query: str) -> Classification:
|
||||
raw, matched = self._score(query)
|
||||
q = query.lower()
|
||||
raw = self._score(q)
|
||||
if not raw:
|
||||
# 完全无命中 -> general，低置信度
|
||||
diff, ds = estimate_difficulty(query)
|
||||
@@ -150,7 +155,7 @@ class RuleClassifier(BaseClassifier):
|
||||
difficulty=diff,
|
||||
difficulty_score=ds,
|
||||
raw_scores={k: round(v, 3) for k, v in raw.items()},
|
||||
matched_rules=matched.get(best_domain, []),
|
||||
matched_rules=self._matched_rules(q, best_domain),
|
||||
)
|
||||
|
||||
|
||||
|
||||
+56
-52
@@ -1,52 +1,56 @@
|
||||
"""查询难度估计器（启发式，纯标准库）。
|
||||
|
||||
依据：RouterArena 用 Bloom 分类法把问题分为 easy/medium/hard。
|
||||
这里用查询长度、指令动词、数学/推理标记做轻量估计。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
from typing import Tuple
|
||||
|
||||
# 触发 hard 的指令动词 / 推理标记
|
||||
_HARD_MARKERS = [
|
||||
"证明", "推导", "为什么", "如何", "对比", "比较", "分析", "评估", "设计", "优化",
|
||||
"复杂度", "时间复杂度", "空间复杂度", "原理", "机制", "优缺点", "区别", "推论", "定理",
|
||||
"proof", "prove", "derive", "explain why", "why", "how", "compare", "contrast",
|
||||
"analy", "evaluate", "design", "optimize", "refactor", "architect",
|
||||
"implement", "debug", "review", "plan", "synthesize",
|
||||
"int", "sum", "sqrt", "lim", "log", "derivative", "integral",
|
||||
]
|
||||
# 触发 medium 的标记
|
||||
_MEDIUM_MARKERS = [
|
||||
"用", "写", "计算", "求解", "生成", "翻译", "总结", "解释",
|
||||
"注意", "建议", "是否", "实现", "步骤",
|
||||
"write", "code", "function", "script", "calculate", "solve", "summarize",
|
||||
"translate", "fix", "explain", "describe", "list",
|
||||
]
|
||||
|
||||
|
||||
def estimate_difficulty(query: str) -> Tuple[str, float]:
|
||||
"""返回 (difficulty, score)，score 属于 [0,1]。"""
|
||||
q = query.lower()
|
||||
hard_hits = sum(1 for m in _HARD_MARKERS if m in q)
|
||||
medium_hits = sum(1 for m in _MEDIUM_MARKERS if m in q)
|
||||
length = len(query)
|
||||
|
||||
score = 0.0
|
||||
score += min(0.30, length / 600.0) # 长度贡献
|
||||
score += min(0.55, hard_hits * 0.25) # 推理标记贡献
|
||||
score += min(0.30, medium_hits * 0.08) # 一般指令贡献
|
||||
|
||||
# 额外：代码 / 数学表达式（多步骤信号）
|
||||
if "```" in query or re.search(r"\b(def|class|function|import)\b", q):
|
||||
score += 0.15
|
||||
if re.search(r"[0-9]+\s*[+\-*/^=]\s*[0-9xya-z]", q):
|
||||
score += 0.15
|
||||
|
||||
score = max(0.0, min(1.0, score))
|
||||
if score >= 0.55:
|
||||
return "hard", score
|
||||
if score >= 0.25:
|
||||
return "medium", score
|
||||
return "easy", score␍
|
||||
"""查询难度估计器（启发式，纯标准库）。
|
||||
|
||||
依据：RouterArena 用 Bloom 分类法把问题分为 easy/medium/hard。
|
||||
这里用查询长度、指令动词、数学/推理标记做轻量估计。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
from typing import Tuple
|
||||
|
||||
# 触发 hard 的指令动词 / 推理标记
|
||||
_HARD_MARKERS = [
|
||||
"证明", "推导", "为什么", "如何", "对比", "比较", "分析", "评估", "设计", "优化",
|
||||
"复杂度", "时间复杂度", "空间复杂度", "原理", "机制", "优缺点", "区别", "推论", "定理",
|
||||
"proof", "prove", "derive", "explain why", "why", "how", "compare", "contrast",
|
||||
"analy", "evaluate", "design", "optimize", "refactor", "architect",
|
||||
"implement", "debug", "review", "plan", "synthesize",
|
||||
"int", "sum", "sqrt", "lim", "log", "derivative", "integral",
|
||||
]
|
||||
# 触发 medium 的标记
|
||||
_MEDIUM_MARKERS = [
|
||||
"用", "写", "计算", "求解", "生成", "翻译", "总结", "解释",
|
||||
"注意", "建议", "是否", "实现", "步骤",
|
||||
"write", "code", "function", "script", "calculate", "solve", "summarize",
|
||||
"translate", "fix", "explain", "describe", "list",
|
||||
]
|
||||
|
||||
# 预编译正则（模块级一次，避免每次调用走 re 内部缓存查找）
|
||||
_RE_CODE_EXPR = re.compile(r"\b(def|class|function|import)\b")
|
||||
_RE_ARITH_EXPR = re.compile(r"[0-9]+\s*[+\-*/^=]\s*[0-9xya-z]")
|
||||
|
||||
|
||||
def estimate_difficulty(query: str) -> Tuple[str, float]:
|
||||
"""返回 (difficulty, score)，score 属于 [0,1]。"""
|
||||
q = query.lower()
|
||||
hard_hits = sum(1 for m in _HARD_MARKERS if m in q)
|
||||
medium_hits = sum(1 for m in _MEDIUM_MARKERS if m in q)
|
||||
length = len(query)
|
||||
|
||||
score = 0.0
|
||||
score += min(0.30, length / 600.0) # 长度贡献
|
||||
score += min(0.55, hard_hits * 0.25) # 推理标记贡献
|
||||
score += min(0.30, medium_hits * 0.08) # 一般指令贡献
|
||||
|
||||
# 额外：代码 / 数学表达式（多步骤信号）
|
||||
if "```" in query or _RE_CODE_EXPR.search(q):
|
||||
score += 0.15
|
||||
if _RE_ARITH_EXPR.search(q):
|
||||
score += 0.15
|
||||
|
||||
score = max(0.0, min(1.0, score))
|
||||
if score >= 0.55:
|
||||
return "hard", score
|
||||
if score >= 0.25:
|
||||
return "medium", score
|
||||
return "easy", score
|
||||
|
||||
+261
-270
@@ -1,270 +1,261 @@
|
||||
"""专家模型池：统一 Expert 接口，支持三种后端。
|
||||
|
||||
- MockExpert ：确定性模板输出（零依赖，离线可跑，便于测试与演示）
|
||||
- HFExpert ：HuggingFace transformers 真实小模型（可选，需 ML 依赖）
|
||||
- APIExpert ：OpenAI 兼容 API（可选，需 API Key，如 DeepSeek）
|
||||
|
||||
成本估计：cost_est 按参数量粗估（美元/百万 token 的近似比例）。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import re
|
||||
from typing import Dict, List, Optional
|
||||
|
||||
from .models import ExpertResponse
|
||||
|
||||
# 按模型规模粗估的相对成本（$ / 1M output tokens，近似）
|
||||
MODEL_COST_EST = {
|
||||
"mock": 0.0,
|
||||
"0.5b": 0.02,
|
||||
"1b": 0.05,
|
||||
"1.7b": 0.08,
|
||||
"3b": 0.12,
|
||||
"4b": 0.15,
|
||||
"7b": 0.25,
|
||||
"70b": 2.50,
|
||||
"api": 1.00,
|
||||
}
|
||||
|
||||
|
||||
def _cost_for(model_name: str, default: str = "1b") -> float:
|
||||
mn = model_name.lower()
|
||||
for key in ("0.5b", "1.7b", "3b", "4b", "7b", "70b"):
|
||||
if key in mn:
|
||||
return MODEL_COST_EST[key]
|
||||
if "api" in mn or mn in ("deepseek-chat", "gpt-4o-mini", "claude"):
|
||||
return MODEL_COST_EST["api"]
|
||||
return MODEL_COST_EST.get(default, 0.1)
|
||||
|
||||
|
||||
def extract_content_terms(query: str) -> List[str]:
|
||||
"""抽取查询中的"内容词"（中文词/英文单词），用于 Judge 覆盖度与 Mock 回显。"""
|
||||
q = query.lower()
|
||||
terms: List[str] = []
|
||||
# 英文单词（>=2 字符）
|
||||
for w in re.findall(r"[a-z][a-z0-9_]{1,}", q):
|
||||
if w not in _STOPWORDS_EN and w not in terms:
|
||||
terms.append(w)
|
||||
# 中文：按 2-4 字窗口切分，保留含中文字符的片段
|
||||
cn = re.findall(r"[\u4e00-\u9fff]{2,8}", q)
|
||||
for c in cn:
|
||||
terms.append(c)
|
||||
return terms
|
||||
|
||||
|
||||
_STOPWORDS_EN = {
|
||||
"the", "a", "an", "is", "are", "to", "of", "in", "on", "for", "with", "and",
|
||||
"or", "do", "does", "can", "could", "would", "should", "please", "me", "my",
|
||||
"this", "that", "it", "be", "was", "were", "have", "has", "had", "will",
|
||||
"not", "no", "yes", "i", "you", "he", "she", "we", "they",
|
||||
}
|
||||
|
||||
|
||||
class Expert:
|
||||
name: str = "expert"
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class MockExpert(Expert):
|
||||
"""确定性模板专家：零依赖，离线可跑。
|
||||
|
||||
输出会回显查询中的内容词以提高 Judge 覆盖度，并带领域结构，
|
||||
使端到端管线（分类 -> 专家 -> Judge -> 缓存）可被稳定测试与演示。
|
||||
"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str = "mock"):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
await asyncio.sleep(0.001) # 模拟极短推理延迟
|
||||
terms = extract_content_terms(query)
|
||||
body = self._template(query, terms, difficulty)
|
||||
# 预估 token 数：中文约 1.5 字符/token，英文约 4 字符/token
|
||||
tokens = max(8, int(len(body) / 2.2))
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=1.0,
|
||||
tokens=tokens,
|
||||
cost_est=_cost_for(self.model) * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
def _template(self, query: str, terms: List[str], difficulty: str) -> str:
|
||||
kw = "、".join(terms[:6]) if terms else "该主题"
|
||||
if self.domain == "code":
|
||||
return (
|
||||
f"（mock 代码专家）针对「{query}」的实现思路如下：\n\n"
|
||||
f"```python\n"
|
||||
f"def solve() -> None:\n"
|
||||
f" # 关键点：{kw}\n"
|
||||
f" # 1. 明确输入输出约束\n"
|
||||
f" # 2. 选择合适数据结构\n"
|
||||
f" # 3. 处理边界条件（空输入、极端值）\n"
|
||||
f" # 4. 补充单元测试\n"
|
||||
f" pass\n"
|
||||
f"```\n\n"
|
||||
f"复杂度：平均 O(n)。请按上述步骤补充具体实现。"
|
||||
)
|
||||
if self.domain == "math":
|
||||
return (
|
||||
f"（mock 数学专家）求解「{query}」的步骤：\n\n"
|
||||
f"1. 明确已知条件与目标：{kw}\n"
|
||||
f"2. 选择合适的方法（代数变形 / 积分 / 归纳等）\n"
|
||||
f"3. 逐步推导并验证中间结果\n"
|
||||
f"4. 检查边界与特殊情况\n\n"
|
||||
f"结论：在标准假设下，结果可化简为闭合形式。完整推导见正式解答。"
|
||||
)
|
||||
if self.domain == "legal":
|
||||
return (
|
||||
f"（mock 法律专家）关于「{query}」的初步法律分析：\n\n"
|
||||
f"相关要点：{kw}\n"
|
||||
f"1. 适用法规：请以现行有效法条为准（建议核对最新修订版）\n"
|
||||
f"2. 合同/合规风险点识别\n"
|
||||
f"3. 责任划分与救济途径\n\n"
|
||||
f"⚠️ 提示：以上为一般性分析，不构成正式法律意见，个案请咨询执业律师。"
|
||||
)
|
||||
if self.domain == "medical":
|
||||
return (
|
||||
f"（mock 医学专家）关于「{query}」的科普性说明：\n\n"
|
||||
f"相关关键词：{kw}\n"
|
||||
f"1. 常见表现与可能原因\n"
|
||||
f"2. 一般处理建议与注意事项\n"
|
||||
f"3. 何时需要就医（警示信号）\n\n"
|
||||
f"⚠️ 提示：内容仅供健康科普，不能替代医生诊断；如有不适请及时就医。"
|
||||
)
|
||||
return (
|
||||
f"（mock 通用专家）关于「{query}」的回答：\n\n"
|
||||
f"核心要点：{kw}\n"
|
||||
f"1. 背景与定义\n"
|
||||
f"2. 主要分类/维度\n"
|
||||
f"3. 实际应用与注意事项\n\n"
|
||||
f"如需更深入的分析，可以补充更多上下文。"
|
||||
)
|
||||
|
||||
|
||||
class HFExpert(Expert):
|
||||
"""可选：HuggingFace 真实小模型（需 requirements-ml.txt）。"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
self._loaded = False
|
||||
self._model = None
|
||||
self._tokenizer = None
|
||||
|
||||
def _ensure_loaded(self):
|
||||
if self._loaded:
|
||||
return
|
||||
try:
|
||||
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||||
except ImportError as e:
|
||||
raise RuntimeError("HFExpert 需要安装 ML 依赖：pip install -r requirements-ml.txt") from e
|
||||
self._tokenizer = AutoTokenizer.from_pretrained(self.model)
|
||||
self._model = AutoModelForCausalLM.from_pretrained(
|
||||
self.model, device_map="auto", torch_dtype="auto"
|
||||
)
|
||||
self._loaded = True
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
self._ensure_loaded()
|
||||
return await asyncio.to_thread(self._generate_sync, query)
|
||||
|
||||
def _generate_sync(self, query: str) -> ExpertResponse:
|
||||
messages = [{"role": "user", "content": query}]
|
||||
text = self._tokenizer.apply_chat_template(messages, tokenize=False)
|
||||
inputs = self._tokenizer(text, return_tensors="pt").to(self._model.device)
|
||||
outputs = self._model.generate(
|
||||
**inputs,
|
||||
max_new_tokens=512,
|
||||
temperature=0.2,
|
||||
do_sample=True,
|
||||
)
|
||||
body = self._tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=512,
|
||||
cost_est=_cost_for(self.model) * 512 / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
class APIExpert(Expert):
|
||||
"""可选：OpenAI 兼容 Chat Completions（DeepSeek / OpenAI / 本地 vLLM）。"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str, base_url: str, api_key: str):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
self.base_url = base_url.rstrip("/")
|
||||
self.api_key = api_key
|
||||
self._client = None
|
||||
|
||||
def _get_client(self):
|
||||
if self._client is None:
|
||||
import httpx
|
||||
self._client = httpx.AsyncClient(timeout=60.0)
|
||||
return self._client
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
client = self._get_client()
|
||||
resp = await client.post(
|
||||
f"{self.base_url}/chat/completions",
|
||||
headers={"Authorization": f"Bearer {self.api_key}"},
|
||||
json={
|
||||
"model": self.model,
|
||||
"messages": [{"role": "user", "content": query}],
|
||||
"temperature": 0.2,
|
||||
"max_tokens": 1024,
|
||||
},
|
||||
)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
body = data["choices"][0]["message"]["content"]
|
||||
usage = data.get("usage", {})
|
||||
tokens = usage.get("completion_tokens", int(len(body) / 2.2))
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=tokens,
|
||||
cost_est=_cost_for(self.model) * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
def build_expert(domain: str, cfg: Dict) -> Expert:
|
||||
"""根据配置构建领域专家。cfg 为 experts.<domain> 段配置。"""
|
||||
etype = cfg.get("type", "mock")
|
||||
model = cfg.get("model", "mock")
|
||||
name = f"expert-{domain}"
|
||||
if etype == "mock":
|
||||
return MockExpert(name, domain, model)
|
||||
if etype == "hf":
|
||||
return HFExpert(name, domain, model)
|
||||
if etype == "api":
|
||||
base_url = cfg.get("base_url", "https://api.deepseek.com/v1")
|
||||
api_key = cfg.get("api_key") or _env(cfg.get("api_key_env", ""))
|
||||
if not api_key:
|
||||
raise RuntimeError(f"APIExpert({domain}) 缺少 API Key（env: {cfg.get('api_key_env')}）")
|
||||
return APIExpert(name, domain, model, base_url, api_key)
|
||||
raise ValueError(f"未知专家后端类型: {etype}（支持 mock | hf | api）")
|
||||
|
||||
|
||||
def _env(name: str) -> Optional[str]:
|
||||
import os
|
||||
return os.environ.get(name) if name else None
|
||||
|
||||
|
||||
def build_expert_pool(experts_cfg: Dict[str, Dict], domains: List[str]) -> Dict[str, Expert]:
|
||||
"""构建完整专家池。"""
|
||||
pool: Dict[str, Expert] = {}
|
||||
for domain in domains:
|
||||
cfg = experts_cfg.get(domain, {"type": "mock", "model": "mock"})
|
||||
pool[domain] = build_expert(domain, cfg)
|
||||
return pool␍
|
||||
"""专家模型池：统一 Expert 接口，支持三种后端。
|
||||
|
||||
- MockExpert ：确定性模板输出（零依赖，离线可跑，便于测试与演示）
|
||||
- HFExpert ：HuggingFace transformers 真实小模型（可选，需 ML 依赖）
|
||||
- APIExpert ：OpenAI 兼容 API（可选，需 API Key，如 DeepSeek）
|
||||
|
||||
成本估计：cost_est 按参数量粗估（美元/百万 token 的近似比例）。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import re
|
||||
from functools import lru_cache
|
||||
from typing import Dict, List, Optional, Tuple
|
||||
|
||||
from .config import get_api_key
|
||||
from .llm_client import OpenAICompatClient
|
||||
from .models import ExpertResponse
|
||||
|
||||
# 按模型规模粗估的相对成本（$ / 1M output tokens，近似）
|
||||
MODEL_COST_EST = {
|
||||
"mock": 0.0,
|
||||
"0.5b": 0.02,
|
||||
"1b": 0.05,
|
||||
"1.7b": 0.08,
|
||||
"3b": 0.12,
|
||||
"4b": 0.15,
|
||||
"7b": 0.25,
|
||||
"70b": 2.50,
|
||||
"api": 1.00,
|
||||
}
|
||||
|
||||
|
||||
def _cost_for(model_name: str, default: str = "1b") -> float:
|
||||
mn = model_name.lower()
|
||||
for key in ("0.5b", "1.7b", "3b", "4b", "7b", "70b"):
|
||||
if key in mn:
|
||||
return MODEL_COST_EST[key]
|
||||
if "api" in mn or mn in ("deepseek-chat", "gpt-4o-mini", "claude"):
|
||||
return MODEL_COST_EST["api"]
|
||||
return MODEL_COST_EST.get(default, 0.1)
|
||||
|
||||
|
||||
@lru_cache(maxsize=2048)
|
||||
def _extract_content_terms_cached(query: str) -> Tuple[str, ...]:
|
||||
q = query.lower()
|
||||
terms: List[str] = []
|
||||
# 英文单词（>=2 字符）
|
||||
for w in re.findall(r"[a-z][a-z0-9_]{1,}", q):
|
||||
if w not in _STOPWORDS_EN and w not in terms:
|
||||
terms.append(w)
|
||||
# 中文：按 2-4 字窗口切分，保留含中文字符的片段
|
||||
cn = re.findall(r"[\u4e00-\u9fff]{2,8}", q)
|
||||
for c in cn:
|
||||
terms.append(c)
|
||||
return tuple(terms)
|
||||
|
||||
|
||||
def extract_content_terms(query: str) -> Tuple[str, ...]:
|
||||
"""抽取查询中的"内容词"（中文词/英文单词），用于 Judge 覆盖度与 Mock 回显。
|
||||
|
||||
纯函数，但 Router 主链路中专家生成与 Judge 评估会对同一查询各调一次，
|
||||
故以 lru_cache 去重；返回不可变 tuple（调用方只读）。
|
||||
"""
|
||||
return _extract_content_terms_cached(query)
|
||||
|
||||
|
||||
_STOPWORDS_EN = {
|
||||
"the", "a", "an", "is", "are", "to", "of", "in", "on", "for", "with", "and",
|
||||
"or", "do", "does", "can", "could", "would", "should", "please", "me", "my",
|
||||
"this", "that", "it", "be", "was", "were", "have", "has", "had", "will",
|
||||
"not", "no", "yes", "i", "you", "he", "she", "we", "they",
|
||||
}
|
||||
|
||||
|
||||
class Expert:
|
||||
name: str = "expert"
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class MockExpert(Expert):
|
||||
"""确定性模板专家：零依赖，离线可跑。
|
||||
|
||||
输出会回显查询中的内容词以提高 Judge 覆盖度，并带领域结构，
|
||||
使端到端管线（分类 -> 专家 -> Judge -> 缓存）可被稳定测试与演示。
|
||||
"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str = "mock"):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
await asyncio.sleep(0.001) # 模拟极短推理延迟
|
||||
terms = extract_content_terms(query)
|
||||
body = self._template(query, terms, difficulty)
|
||||
# 预估 token 数：中文约 1.5 字符/token，英文约 4 字符/token
|
||||
tokens = max(8, int(len(body) / 2.2))
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=1.0,
|
||||
tokens=tokens,
|
||||
cost_est=_cost_for(self.model) * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
def _template(self, query: str, terms: List[str], difficulty: str) -> str:
|
||||
kw = "、".join(terms[:6]) if terms else "该主题"
|
||||
if self.domain == "code":
|
||||
return (
|
||||
f"（mock 代码专家）针对「{query}」的实现思路如下：\n\n"
|
||||
f"```python\n"
|
||||
f"def solve() -> None:\n"
|
||||
f" # 关键点：{kw}\n"
|
||||
f" # 1. 明确输入输出约束\n"
|
||||
f" # 2. 选择合适数据结构\n"
|
||||
f" # 3. 处理边界条件（空输入、极端值）\n"
|
||||
f" # 4. 补充单元测试\n"
|
||||
f" pass\n"
|
||||
f"```\n\n"
|
||||
f"复杂度：平均 O(n)。请按上述步骤补充具体实现。"
|
||||
)
|
||||
if self.domain == "math":
|
||||
return (
|
||||
f"（mock 数学专家）求解「{query}」的步骤：\n\n"
|
||||
f"1. 明确已知条件与目标：{kw}\n"
|
||||
f"2. 选择合适的方法（代数变形 / 积分 / 归纳等）\n"
|
||||
f"3. 逐步推导并验证中间结果\n"
|
||||
f"4. 检查边界与特殊情况\n\n"
|
||||
f"结论：在标准假设下，结果可化简为闭合形式。完整推导见正式解答。"
|
||||
)
|
||||
if self.domain == "legal":
|
||||
return (
|
||||
f"（mock 法律专家）关于「{query}」的初步法律分析：\n\n"
|
||||
f"相关要点：{kw}\n"
|
||||
f"1. 适用法规：请以现行有效法条为准（建议核对最新修订版）\n"
|
||||
f"2. 合同/合规风险点识别\n"
|
||||
f"3. 责任划分与救济途径\n\n"
|
||||
f"⚠️ 提示：以上为一般性分析，不构成正式法律意见，个案请咨询执业律师。"
|
||||
)
|
||||
if self.domain == "medical":
|
||||
return (
|
||||
f"（mock 医学专家）关于「{query}」的科普性说明：\n\n"
|
||||
f"相关关键词：{kw}\n"
|
||||
f"1. 常见表现与可能原因\n"
|
||||
f"2. 一般处理建议与注意事项\n"
|
||||
f"3. 何时需要就医（警示信号）\n\n"
|
||||
f"⚠️ 提示：内容仅供健康科普，不能替代医生诊断；如有不适请及时就医。"
|
||||
)
|
||||
return (
|
||||
f"（mock 通用专家）关于「{query}」的回答：\n\n"
|
||||
f"核心要点：{kw}\n"
|
||||
f"1. 背景与定义\n"
|
||||
f"2. 主要分类/维度\n"
|
||||
f"3. 实际应用与注意事项\n\n"
|
||||
f"如需更深入的分析，可以补充更多上下文。"
|
||||
)
|
||||
|
||||
|
||||
class HFExpert(Expert):
|
||||
"""可选：HuggingFace 真实小模型（需 requirements-ml.txt）。"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
self._loaded = False
|
||||
self._model = None
|
||||
self._tokenizer = None
|
||||
|
||||
def _ensure_loaded(self):
|
||||
if self._loaded:
|
||||
return
|
||||
try:
|
||||
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||||
except ImportError as e:
|
||||
raise RuntimeError("HFExpert 需要安装 ML 依赖：pip install -r requirements-ml.txt") from e
|
||||
self._tokenizer = AutoTokenizer.from_pretrained(self.model)
|
||||
self._model = AutoModelForCausalLM.from_pretrained(
|
||||
self.model, device_map="auto", torch_dtype="auto"
|
||||
)
|
||||
self._loaded = True
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
self._ensure_loaded()
|
||||
return await asyncio.to_thread(self._generate_sync, query)
|
||||
|
||||
def _generate_sync(self, query: str) -> ExpertResponse:
|
||||
messages = [{"role": "user", "content": query}]
|
||||
text = self._tokenizer.apply_chat_template(messages, tokenize=False)
|
||||
inputs = self._tokenizer(text, return_tensors="pt").to(self._model.device)
|
||||
outputs = self._model.generate(
|
||||
**inputs,
|
||||
max_new_tokens=512,
|
||||
temperature=0.2,
|
||||
do_sample=True,
|
||||
)
|
||||
body = self._tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=512,
|
||||
cost_est=_cost_for(self.model) * 512 / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
class APIExpert(Expert):
|
||||
"""可选：OpenAI 兼容 Chat Completions（DeepSeek / OpenAI / 本地 vLLM）。"""
|
||||
|
||||
def __init__(self, name: str, domain: str, model: str, base_url: str, api_key: str):
|
||||
self.name = name
|
||||
self.domain = domain
|
||||
self.model = model
|
||||
self._client = OpenAICompatClient(base_url, api_key, timeout=60.0)
|
||||
|
||||
async def generate(self, query: str, difficulty: str) -> ExpertResponse:
|
||||
data = await self._client.chat(
|
||||
self.model,
|
||||
[{"role": "user", "content": query}],
|
||||
temperature=0.2,
|
||||
max_tokens=1024,
|
||||
)
|
||||
body = self._client.completion_text(data)
|
||||
tokens = self._client.completion_tokens(data, body)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=tokens,
|
||||
cost_est=_cost_for(self.model) * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
def build_expert(domain: str, cfg: Dict) -> Expert:
|
||||
"""根据配置构建领域专家。cfg 为 experts.<domain> 段配置。"""
|
||||
etype = cfg.get("type", "mock")
|
||||
model = cfg.get("model", "mock")
|
||||
name = f"expert-{domain}"
|
||||
if etype == "mock":
|
||||
return MockExpert(name, domain, model)
|
||||
if etype == "hf":
|
||||
return HFExpert(name, domain, model)
|
||||
if etype == "api":
|
||||
base_url = cfg.get("base_url", "https://api.deepseek.com/v1")
|
||||
api_key = cfg.get("api_key") or get_api_key(cfg)
|
||||
if not api_key:
|
||||
raise RuntimeError(f"APIExpert({domain}) 缺少 API Key（env: {cfg.get('api_key_env')}）")
|
||||
return APIExpert(name, domain, model, base_url, api_key)
|
||||
raise ValueError(f"未知专家后端类型: {etype}（支持 mock | hf | api）")
|
||||
|
||||
|
||||
def build_expert_pool(experts_cfg: Dict[str, Dict], domains: List[str]) -> Dict[str, Expert]:
|
||||
"""构建完整专家池。"""
|
||||
pool: Dict[str, Expert] = {}
|
||||
for domain in domains:
|
||||
cfg = experts_cfg.get(domain, {"type": "mock", "model": "mock"})
|
||||
pool[domain] = build_expert(domain, cfg)
|
||||
return pool
|
||||
|
||||
+84
-99
@@ -1,99 +1,84 @@
|
||||
"""大模型回退层：Mock 与 OpenAI 兼容 API 两种后端。"""
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
from typing import Dict, Optional
|
||||
|
||||
from .models import ExpertResponse
|
||||
|
||||
|
||||
class FallbackProvider:
|
||||
name: str = "fallback"
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class MockFallback(FallbackProvider):
|
||||
"""确定性 mock 大模型：标识为 fallback，便于测试升级路径。"""
|
||||
|
||||
def __init__(self, model: str = "mock-large"):
|
||||
self.model = model
|
||||
self.name = f"fallback-{model}"
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
await asyncio.sleep(0.002)
|
||||
body = (
|
||||
f"（大模型回退）「{query}」\n\n"
|
||||
"这是一条来自大模型回退路径的完整回答。\n"
|
||||
"要点：\n"
|
||||
"1. 对复杂/跨域任务给出综合推理\n"
|
||||
"2. 补充领域专家未覆盖的上下文\n"
|
||||
"3. 给出可执行的后续建议\n"
|
||||
)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=2.0,
|
||||
tokens=120,
|
||||
cost_est=2.0 * 120 / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
class APIFallback(FallbackProvider):
|
||||
"""OpenAI 兼容大模型 API（如 DeepSeek / OpenAI / 本地 vLLM）。"""
|
||||
|
||||
def __init__(self, model: str, base_url: str, api_key: str):
|
||||
self.model = model
|
||||
self.base_url = base_url.rstrip("/")
|
||||
self.api_key = api_key
|
||||
self.name = f"fallback-{model}"
|
||||
self._client = None
|
||||
|
||||
def _get_client(self):
|
||||
if self._client is None:
|
||||
import httpx
|
||||
self._client = httpx.AsyncClient(timeout=90.0)
|
||||
return self._client
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
client = self._get_client()
|
||||
resp = await client.post(
|
||||
f"{self.base_url}/chat/completions",
|
||||
headers={"Authorization": f"Bearer {self.api_key}"},
|
||||
json={
|
||||
"model": self.model,
|
||||
"messages": [{"role": "user", "content": query}],
|
||||
"temperature": 0.3,
|
||||
"max_tokens": 2048,
|
||||
},
|
||||
)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
body = data["choices"][0]["message"]["content"]
|
||||
usage = data.get("usage", {})
|
||||
tokens = usage.get("completion_tokens", int(len(body) / 2.2))
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=tokens,
|
||||
cost_est=2.0 * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
def build_fallback(cfg: Dict) -> FallbackProvider:
|
||||
"""cfg 为 fallback 段配置。"""
|
||||
ftype = cfg.get("type", "mock")
|
||||
model = cfg.get("model", "deepseek-chat")
|
||||
if ftype == "mock":
|
||||
return MockFallback(model=model)
|
||||
if ftype == "api":
|
||||
import os
|
||||
api_key = cfg.get("api_key") or os.environ.get(cfg.get("api_key_env", "API_KEY"))
|
||||
if not api_key:
|
||||
raise RuntimeError(
|
||||
f"APIFallback 缺少 API Key：请设置环境变量 {cfg.get('api_key_env')} 或配置 api_key"
|
||||
)
|
||||
return APIFallback(model, cfg.get("base_url", "https://api.deepseek.com/v1"), api_key)
|
||||
raise ValueError(f"未知 fallback 类型: {ftype}（支持 mock | api）")␍
|
||||
"""大模型回退层：Mock 与 OpenAI 兼容 API 两种后端。"""
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
from typing import Dict
|
||||
|
||||
from .config import get_api_key
|
||||
from .llm_client import OpenAICompatClient
|
||||
from .models import ExpertResponse
|
||||
|
||||
|
||||
class FallbackProvider:
|
||||
name: str = "fallback"
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class MockFallback(FallbackProvider):
|
||||
"""确定性 mock 大模型：标识为 fallback，便于测试升级路径。"""
|
||||
|
||||
def __init__(self, model: str = "mock-large"):
|
||||
self.model = model
|
||||
self.name = f"fallback-{model}"
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
await asyncio.sleep(0.002)
|
||||
body = (
|
||||
f"（大模型回退）「{query}」\n\n"
|
||||
"这是一条来自大模型回退路径的完整回答。\n"
|
||||
"要点：\n"
|
||||
"1. 对复杂/跨域任务给出综合推理\n"
|
||||
"2. 补充领域专家未覆盖的上下文\n"
|
||||
"3. 给出可执行的后续建议\n"
|
||||
)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=2.0,
|
||||
tokens=120,
|
||||
cost_est=2.0 * 120 / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
class APIFallback(FallbackProvider):
|
||||
"""OpenAI 兼容大模型 API（如 DeepSeek / OpenAI / 本地 vLLM）。"""
|
||||
|
||||
def __init__(self, model: str, base_url: str, api_key: str):
|
||||
self.model = model
|
||||
self.name = f"fallback-{model}"
|
||||
self._client = OpenAICompatClient(base_url, api_key, timeout=90.0)
|
||||
|
||||
async def generate(self, query: str) -> ExpertResponse:
|
||||
data = await self._client.chat(
|
||||
self.model,
|
||||
[{"role": "user", "content": query}],
|
||||
temperature=0.3,
|
||||
max_tokens=2048,
|
||||
)
|
||||
body = self._client.completion_text(data)
|
||||
tokens = self._client.completion_tokens(data, body)
|
||||
return ExpertResponse(
|
||||
text=body,
|
||||
model_used=self.model,
|
||||
latency_ms=0.0,
|
||||
tokens=tokens,
|
||||
cost_est=2.0 * tokens / 1_000_000,
|
||||
)
|
||||
|
||||
|
||||
def build_fallback(cfg: Dict) -> FallbackProvider:
|
||||
"""cfg 为 fallback 段配置。"""
|
||||
ftype = cfg.get("type", "mock")
|
||||
model = cfg.get("model", "deepseek-chat")
|
||||
if ftype == "mock":
|
||||
return MockFallback(model=model)
|
||||
if ftype == "api":
|
||||
api_key = cfg.get("api_key") or get_api_key(cfg)
|
||||
if not api_key:
|
||||
raise RuntimeError(
|
||||
f"APIFallback 缺少 API Key：请设置环境变量 {cfg.get('api_key_env')} 或配置 api_key"
|
||||
)
|
||||
return APIFallback(model, cfg.get("base_url", "https://api.deepseek.com/v1"), api_key)
|
||||
raise ValueError(f"未知 fallback 类型: {ftype}（支持 mock | api）")
|
||||
|
||||
+168
-174
@@ -1,174 +1,168 @@
|
||||
"""质量控制器（Judge）：评估专家输出，决定是否升级大模型。
|
||||
|
||||
- RuleJudge：零依赖启发式（内容覆盖度 / 长度充分性 / 领域格式 / 安全提示），
|
||||
稳定可测，适合 MVP 与离线演示。
|
||||
- LLMJudge：可选，基于 transformers 小模型或 API 的 LLM-as-Judge。
|
||||
|
||||
设计对齐实现方案：overall_score < judge_fallback_threshold -> 升级大模型。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
from dataclasses import dataclass, field
|
||||
from typing import Dict, List
|
||||
|
||||
from .experts import extract_content_terms
|
||||
|
||||
# 各领域期望的响应长度范围（字符数）
|
||||
_EXPECTED_LEN = {
|
||||
"code": (60, 2000),
|
||||
"math": (60, 2000),
|
||||
"legal": (80, 3000),
|
||||
"medical": (80, 3000),
|
||||
"general": (40, 2000),
|
||||
}
|
||||
|
||||
# 领域格式检查：响应应包含的标记
|
||||
_DOMAIN_FORMAT_HINTS = {
|
||||
"code": ["```", "def ", "function", "class "],
|
||||
"math": ["步骤", "推导", "=", "解", "step"],
|
||||
"legal": ["⚠", "法律", "意见", "合规", "contract", "law"],
|
||||
"medical": ["⚠", "就医", "医生", "症状", "诊断", "symptom"],
|
||||
"general": [],
|
||||
}
|
||||
|
||||
|
||||
@dataclass
|
||||
class QualityEvaluation:
|
||||
overall_score: float
|
||||
scores: Dict[str, float] = field(default_factory=dict)
|
||||
needs_fallback: bool = False
|
||||
reasons: List[str] = field(default_factory=list)
|
||||
|
||||
def to_dict(self) -> Dict:
|
||||
return {
|
||||
"overall_score": round(self.overall_score, 4),
|
||||
"scores": {k: round(v, 4) for k, v in self.scores.items()},
|
||||
"needs_fallback": self.needs_fallback,
|
||||
"reasons": self.reasons,
|
||||
}
|
||||
|
||||
|
||||
class BaseJudge:
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class RuleJudge(BaseJudge):
|
||||
"""启发式质量评估（零依赖）。"""
|
||||
|
||||
def __init__(self, fallback_threshold: float = 0.70):
|
||||
self.fallback_threshold = fallback_threshold
|
||||
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
scores: Dict[str, float] = {}
|
||||
reasons: List[str] = []
|
||||
|
||||
# 1) 内容覆盖度：查询中的内容词有多少出现在响应里
|
||||
terms = extract_content_terms(query)
|
||||
if terms:
|
||||
hit = sum(1 for t in terms if t in response.lower())
|
||||
coverage = hit / len(terms)
|
||||
scores["coverage"] = coverage
|
||||
if coverage < 0.4:
|
||||
reasons.append(f"内容覆盖度低 ({coverage:.0%})")
|
||||
else:
|
||||
scores["coverage"] = 1.0
|
||||
|
||||
# 2) 长度充分性
|
||||
lo, hi = _EXPECTED_LEN.get(domain, (40, 2000))
|
||||
n = len(response)
|
||||
if n < lo:
|
||||
scores["length"] = max(0.0, n / lo)
|
||||
reasons.append(f"响应过短 ({n} 字符)")
|
||||
elif n > hi:
|
||||
scores["length"] = 0.8
|
||||
reasons.append(f"响应过长 ({n} 字符)")
|
||||
else:
|
||||
scores["length"] = 1.0
|
||||
|
||||
# 3) 领域格式检查
|
||||
hints = _DOMAIN_FORMAT_HINTS.get(domain, [])
|
||||
if hints:
|
||||
hit_hints = sum(1 for h in hints if h in response)
|
||||
scores["format"] = min(1.0, 0.4 + 0.2 * hit_hints)
|
||||
if hit_hints == 0:
|
||||
reasons.append("缺少领域格式特征")
|
||||
else:
|
||||
scores["format"] = 1.0
|
||||
|
||||
# 4) 安全/免责提示（法律、医疗领域应有警示语）
|
||||
if domain in ("legal", "medical") and ("⚠" not in response and "提示" not in response):
|
||||
scores["safety"] = 0.6
|
||||
reasons.append("缺少免责提示")
|
||||
else:
|
||||
scores["safety"] = 1.0
|
||||
|
||||
weights = {"coverage": 0.4, "length": 0.2, "format": 0.2, "safety": 0.2}
|
||||
overall = sum(scores.get(k, 0.0) * w for k, w in weights.items())
|
||||
needs = overall < self.fallback_threshold
|
||||
if needs:
|
||||
reasons.append("质量分低于阈值，建议升级大模型")
|
||||
return QualityEvaluation(
|
||||
overall_score=round(overall, 4),
|
||||
scores=scores,
|
||||
needs_fallback=needs,
|
||||
reasons=reasons,
|
||||
)
|
||||
|
||||
|
||||
class LLMJudge(BaseJudge):
|
||||
"""可选：LLM-as-Judge（API 后端）。"""
|
||||
|
||||
def __init__(self, model: str, base_url: str, api_key: str, fallback_threshold: float = 0.70):
|
||||
self.model = model
|
||||
self.base_url = base_url.rstrip("/")
|
||||
self.api_key = api_key
|
||||
self.fallback_threshold = fallback_threshold
|
||||
self._client = None
|
||||
|
||||
def _get_client(self):
|
||||
if self._client is None:
|
||||
import httpx
|
||||
self._client = httpx.AsyncClient(timeout=60.0)
|
||||
return self._client
|
||||
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
client = self._get_client()
|
||||
prompt = (
|
||||
f"你是质量评审员。评估以下回答对查询的满足程度，输出 0-1 分（相关性/正确性/完整性）。\n"
|
||||
f"查询: {query}\n领域: {domain}\n回答: {response[:2000]}\n"
|
||||
f"只输出一个 0 到 1 之间的数字。"
|
||||
)
|
||||
resp = await client.post(
|
||||
f"{self.base_url}/chat/completions",
|
||||
headers={"Authorization": f"Bearer {self.api_key}"},
|
||||
json={"model": self.model, "messages": [{"role": "user", "content": prompt}]},
|
||||
)
|
||||
resp.raise_for_status()
|
||||
try:
|
||||
score = float(resp.json()["choices"][0]["message"]["content"].strip())
|
||||
score = max(0.0, min(1.0, score))
|
||||
except Exception:
|
||||
score = 0.5
|
||||
return QualityEvaluation(
|
||||
overall_score=score,
|
||||
scores={"llm_judge": score},
|
||||
needs_fallback=score < self.fallback_threshold,
|
||||
)
|
||||
|
||||
|
||||
def build_judge(cfg: Dict, fallback_threshold: float = 0.70) -> BaseJudge:
|
||||
"""cfg 为 judge 段配置。"""
|
||||
jtype = cfg.get("type", "rule")
|
||||
if jtype == "rule":
|
||||
return RuleJudge(fallback_threshold=fallback_threshold)
|
||||
if jtype == "llm":
|
||||
import os
|
||||
api_key = cfg.get("api_key") or os.environ.get(cfg.get("api_key_env", "API_KEY"))
|
||||
return LLMJudge(
|
||||
cfg.get("model", "deepseek-chat"),
|
||||
cfg.get("base_url", "https://api.deepseek.com/v1"),
|
||||
api_key or "",
|
||||
fallback_threshold=fallback_threshold,
|
||||
)
|
||||
raise ValueError(f"未知 judge 类型: {jtype}（支持 rule | llm）")␍
|
||||
"""质量控制器（Judge）：评估专家输出，决定是否升级大模型。
|
||||
|
||||
- RuleJudge：零依赖启发式（内容覆盖度 / 长度充分性 / 领域格式 / 安全提示），
|
||||
稳定可测，适合 MVP 与离线演示。
|
||||
- LLMJudge：可选，基于 transformers 小模型或 API 的 LLM-as-Judge。
|
||||
|
||||
设计对齐实现方案：overall_score < judge_fallback_threshold -> 升级大模型。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
from dataclasses import dataclass, field
|
||||
from typing import Dict, List
|
||||
|
||||
from .experts import extract_content_terms
|
||||
|
||||
# 各领域期望的响应长度范围（字符数）
|
||||
_EXPECTED_LEN = {
|
||||
"code": (60, 2000),
|
||||
"math": (60, 2000),
|
||||
"legal": (80, 3000),
|
||||
"medical": (80, 3000),
|
||||
"general": (40, 2000),
|
||||
}
|
||||
|
||||
# 领域格式检查：响应应包含的标记
|
||||
_DOMAIN_FORMAT_HINTS = {
|
||||
"code": ["```", "def ", "function", "class "],
|
||||
"math": ["步骤", "推导", "=", "解", "step"],
|
||||
"legal": ["⚠", "法律", "意见", "合规", "contract", "law"],
|
||||
"medical": ["⚠", "就医", "医生", "症状", "诊断", "symptom"],
|
||||
"general": [],
|
||||
}
|
||||
|
||||
|
||||
@dataclass
|
||||
class QualityEvaluation:
|
||||
overall_score: float
|
||||
scores: Dict[str, float] = field(default_factory=dict)
|
||||
needs_fallback: bool = False
|
||||
reasons: List[str] = field(default_factory=list)
|
||||
|
||||
def to_dict(self) -> Dict:
|
||||
return {
|
||||
"overall_score": round(self.overall_score, 4),
|
||||
"scores": {k: round(v, 4) for k, v in self.scores.items()},
|
||||
"needs_fallback": self.needs_fallback,
|
||||
"reasons": self.reasons,
|
||||
}
|
||||
|
||||
|
||||
class BaseJudge:
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class RuleJudge(BaseJudge):
|
||||
"""启发式质量评估（零依赖）。"""
|
||||
|
||||
def __init__(self, fallback_threshold: float = 0.70):
|
||||
self.fallback_threshold = fallback_threshold
|
||||
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
scores: Dict[str, float] = {}
|
||||
reasons: List[str] = []
|
||||
|
||||
# 1) 内容覆盖度：查询中的内容词有多少出现在响应里
|
||||
terms = extract_content_terms(query)
|
||||
if terms:
|
||||
lowered = response.lower() # lowercase 一次，避免逐词重复复制长响应
|
||||
hit = sum(1 for t in terms if t in lowered)
|
||||
coverage = hit / len(terms)
|
||||
scores["coverage"] = coverage
|
||||
if coverage < 0.4:
|
||||
reasons.append(f"内容覆盖度低 ({coverage:.0%})")
|
||||
else:
|
||||
scores["coverage"] = 1.0
|
||||
|
||||
# 2) 长度充分性
|
||||
lo, hi = _EXPECTED_LEN.get(domain, (40, 2000))
|
||||
n = len(response)
|
||||
if n < lo:
|
||||
scores["length"] = max(0.0, n / lo)
|
||||
reasons.append(f"响应过短 ({n} 字符)")
|
||||
elif n > hi:
|
||||
scores["length"] = 0.8
|
||||
reasons.append(f"响应过长 ({n} 字符)")
|
||||
else:
|
||||
scores["length"] = 1.0
|
||||
|
||||
# 3) 领域格式检查
|
||||
hints = _DOMAIN_FORMAT_HINTS.get(domain, [])
|
||||
if hints:
|
||||
hit_hints = sum(1 for h in hints if h in response)
|
||||
scores["format"] = min(1.0, 0.4 + 0.2 * hit_hints)
|
||||
if hit_hints == 0:
|
||||
reasons.append("缺少领域格式特征")
|
||||
else:
|
||||
scores["format"] = 1.0
|
||||
|
||||
# 4) 安全/免责提示（法律、医疗领域应有警示语）
|
||||
if domain in ("legal", "medical") and ("⚠" not in response and "提示" not in response):
|
||||
scores["safety"] = 0.6
|
||||
reasons.append("缺少免责提示")
|
||||
else:
|
||||
scores["safety"] = 1.0
|
||||
|
||||
weights = {"coverage": 0.4, "length": 0.2, "format": 0.2, "safety": 0.2}
|
||||
overall = sum(scores.get(k, 0.0) * w for k, w in weights.items())
|
||||
needs = overall < self.fallback_threshold
|
||||
if needs:
|
||||
reasons.append("质量分低于阈值，建议升级大模型")
|
||||
return QualityEvaluation(
|
||||
overall_score=round(overall, 4),
|
||||
scores=scores,
|
||||
needs_fallback=needs,
|
||||
reasons=reasons,
|
||||
)
|
||||
|
||||
|
||||
class LLMJudge(BaseJudge):
|
||||
"""可选：LLM-as-Judge（API 后端）。"""
|
||||
|
||||
def __init__(self, model: str, base_url: str, api_key: str, fallback_threshold: float = 0.70):
|
||||
from .llm_client import OpenAICompatClient
|
||||
self.model = model
|
||||
self.fallback_threshold = fallback_threshold
|
||||
# 请求体与旧实现一致：不下发 temperature / max_tokens
|
||||
self._client = OpenAICompatClient(base_url, api_key, timeout=60.0)
|
||||
|
||||
async def evaluate(self, query: str, response: str, domain: str) -> QualityEvaluation:
|
||||
prompt = (
|
||||
f"你是质量评审员。评估以下回答对查询的满足程度，输出 0-1 分（相关性/正确性/完整性）。\n"
|
||||
f"查询: {query}\n领域: {domain}\n回答: {response[:2000]}\n"
|
||||
f"只输出一个 0 到 1 之间的数字。"
|
||||
)
|
||||
data = await self._client.chat(
|
||||
self.model,
|
||||
[{"role": "user", "content": prompt}],
|
||||
temperature=None,
|
||||
max_tokens=None,
|
||||
)
|
||||
try:
|
||||
score = float(self._client.completion_text(data).strip())
|
||||
score = max(0.0, min(1.0, score))
|
||||
except Exception:
|
||||
score = 0.5
|
||||
return QualityEvaluation(
|
||||
overall_score=score,
|
||||
scores={"llm_judge": score},
|
||||
needs_fallback=score < self.fallback_threshold,
|
||||
)
|
||||
|
||||
|
||||
def build_judge(cfg: Dict, fallback_threshold: float = 0.70) -> BaseJudge:
|
||||
"""cfg 为 judge 段配置。"""
|
||||
jtype = cfg.get("type", "rule")
|
||||
if jtype == "rule":
|
||||
return RuleJudge(fallback_threshold=fallback_threshold)
|
||||
if jtype == "llm":
|
||||
from .config import get_api_key
|
||||
api_key = cfg.get("api_key") or get_api_key(cfg)
|
||||
return LLMJudge(
|
||||
cfg.get("model", "deepseek-chat"),
|
||||
cfg.get("base_url", "https://api.deepseek.com/v1"),
|
||||
api_key or "",
|
||||
fallback_threshold=fallback_threshold,
|
||||
)
|
||||
raise ValueError(f"未知 judge 类型: {jtype}（支持 rule | llm）")
|
||||
|
||||
@@ -0,0 +1,69 @@
|
||||
"""OpenAI 兼容 Chat Completions 共享客户端（架构去重）。
|
||||
|
||||
此前 experts.APIExpert / judge.LLMJudge / fallback.APIFallback 各自复制一份
|
||||
"懒建 AsyncClient + POST /chat/completions + choices/usage 解析"的等价逻辑
|
||||
（超时、温度、token 上限各自维护）。本模块将其统一为单一组件：
|
||||
|
||||
- 懒建 httpx.AsyncClient 并长期复用（与仓库内其他客户端一致）
|
||||
- temperature / max_tokens 传 None 时不出现在请求体（保持各后端原请求形状）
|
||||
- 密钥解析统一走 config.get_api_key（此前 experts 的默认环境名与另两处不一致）
|
||||
|
||||
依赖说明：httpx 仅 api 类型后端需要，import 延迟到连接创建时，
|
||||
router_system 核心保持零第三方依赖约束不变。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
|
||||
class OpenAICompatClient:
|
||||
"""OpenAI 兼容 /chat/completions 客户端（懒建连接、跨调用复用）。"""
|
||||
|
||||
def __init__(self, base_url: str, api_key: str, timeout: float = 60.0):
|
||||
self.base_url = base_url.rstrip("/")
|
||||
self.api_key = api_key
|
||||
self._timeout = timeout
|
||||
self._client = None
|
||||
|
||||
def _get_client(self):
|
||||
if self._client is None:
|
||||
import httpx
|
||||
self._client = httpx.AsyncClient(timeout=self._timeout)
|
||||
return self._client
|
||||
|
||||
async def chat(
|
||||
self,
|
||||
model: str,
|
||||
messages: List[Dict[str, str]],
|
||||
temperature: Optional[float] = 0.2,
|
||||
max_tokens: Optional[int] = 1024,
|
||||
) -> Dict[str, Any]:
|
||||
"""调用 /chat/completions，返回完整响应 JSON（含 usage）。
|
||||
|
||||
HTTP 非 2xx 时抛 httpx.HTTPStatusError；temperature/max_tokens
|
||||
为 None 时对应字段不下发（与旧实现中 LLMJudge 的请求体一致）。
|
||||
"""
|
||||
payload: Dict[str, Any] = {"model": model, "messages": messages}
|
||||
if temperature is not None:
|
||||
payload["temperature"] = temperature
|
||||
if max_tokens is not None:
|
||||
payload["max_tokens"] = max_tokens
|
||||
client = self._get_client()
|
||||
resp = await client.post(
|
||||
f"{self.base_url}/chat/completions",
|
||||
headers={"Authorization": f"Bearer {self.api_key}"},
|
||||
json=payload,
|
||||
)
|
||||
resp.raise_for_status()
|
||||
return resp.json()
|
||||
|
||||
@staticmethod
|
||||
def completion_text(data: Dict[str, Any]) -> str:
|
||||
"""提取首条回复文本。"""
|
||||
return data["choices"][0]["message"]["content"]
|
||||
|
||||
@staticmethod
|
||||
def completion_tokens(data: Dict[str, Any], text: str) -> int:
|
||||
"""提取 completion token 数；usage 缺失时按 ~2.2 字符/token 估算。"""
|
||||
usage = data.get("usage", {})
|
||||
return usage.get("completion_tokens", int(len(text) / 2.2))
|
||||
Reference in New Issue
Block a user