fix(ai_router): C1 修復 — AIProviderEnum 對齊 Registry 實際 Provider 名稱
Some checks failed
CD Pipeline / build-and-deploy (push) Failing after 37s
Some checks failed
CD Pipeline / build-and-deploy (push) Failing after 37s
問題: AIProviderEnum.NVIDIA = "nvidia" 在 Registry 無對應 Provider
OpenClawNemoProvider.name = "openclaw_nemo"
NemotronProvider.name = "nemotron"
→ 高複雜度/Tool Calling 路由永遠 skip，靜默 fallback 到 Gemini/Ollama
修復:
- 新增 OPENCLAW_NEMO = "openclaw_nemo" (一般推理, via .188 → NVIDIA NIM)
- 新增 NEMOTRON = "nemotron" (Tool Calling, direct NVIDIA NIM)
- 移除 NVIDIA = "nvidia" (Registry 無對應)
- 規則 4 (複雜度>=4/HIGH風險): NVIDIA → OPENCLAW_NEMO
- route_tool_calling: NVIDIA → NEMOTRON
- Rate Limiter check: "nvidia" → "openclaw_nemo"
- _full_fallback_chain: OPENCLAW_NEMO 首位
- _tool_calling_fallback_chain: NEMOTRON 首位
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
OG T
@@ -71,8 +71,12 @@ class AIProviderEnum(Enum):
|
||||
OLLAMA = "ollama"
|
||||
GEMINI = "gemini"
|
||||
CLAUDE = "claude"
|
||||
# 2026-03-29 ogt: ADR-036 Nemotron Tool Calling (83.3% 精準度)
|
||||
NVIDIA = "nvidia"
|
||||
# 2026-04-02 ogt: C1 修復 — 對齊 Registry 實際名稱
|
||||
# OpenClawNemoProvider.name = "openclaw_nemo" (一般推理, via .188)
|
||||
# NemotronProvider.name = "nemotron" (Tool Calling, direct NVIDIA NIM)
|
||||
# 舊版 NVIDIA = "nvidia" 已移除: Registry 無此 Provider
|
||||
OPENCLAW_NEMO = "openclaw_nemo"
|
||||
NEMOTRON = "nemotron"
|
||||
|
||||
|
||||
# Provider 對應延遲預算 (ms)
|
||||
@@ -80,8 +84,9 @@ PROVIDER_LATENCY_BUDGET: dict[AIProviderEnum, int] = {
|
||||
AIProviderEnum.OLLAMA: 60000, # 本地，允許較長處理時間
|
||||
AIProviderEnum.GEMINI: 30000, # 雲端，較低延遲
|
||||
AIProviderEnum.CLAUDE: 30000, # 雲端，較低延遲
|
||||
# 2026-03-29 ogt: ADR-036 Nemotron Tool Calling (延遲 11-45s)
|
||||
AIProviderEnum.NVIDIA: 60000, # Tool Calling 專用，允許較長時間
|
||||
# 2026-04-02 ogt: C1 修復 — 對齊 Registry 名稱
|
||||
AIProviderEnum.OPENCLAW_NEMO: 60000, # via .188 → NVIDIA NIM，允許較長時間
|
||||
AIProviderEnum.NEMOTRON: 60000, # Tool Calling 專用，允許較長時間
|
||||
}
|
||||
|
||||
|
||||
@@ -210,21 +215,25 @@ class AIRouter:
|
||||
self._ollama_summary = self._model_registry.get_model("ollama", "summary")
|
||||
self._gemini_default = self._model_registry.get_model("gemini", "default")
|
||||
self._claude_default = self._model_registry.get_model("claude", "default")
|
||||
# 2026-03-29 ogt: ADR-036 Nemotron Tool Calling
|
||||
self._nvidia_default = self._model_registry.get_model("nvidia", "default")
|
||||
# 2026-04-02 ogt: C1 修復 — openclaw_nemo (一般推理) + nemotron (Tool Calling)
|
||||
self._openclaw_nemo_default = self._model_registry.get_model("nvidia", "default")
|
||||
self._nemotron_default = self._model_registry.get_model("nvidia", "default")
|
||||
# 向後相容別名
|
||||
self._nvidia_default = self._openclaw_nemo_default
|
||||
|
||||
# Provider 對應模型映射
|
||||
self._provider_models: dict[AIProviderEnum, str] = {
|
||||
AIProviderEnum.OLLAMA: self._ollama_default,
|
||||
AIProviderEnum.GEMINI: self._gemini_default,
|
||||
AIProviderEnum.CLAUDE: self._claude_default,
|
||||
AIProviderEnum.NVIDIA: self._nvidia_default, # ADR-036
|
||||
AIProviderEnum.OPENCLAW_NEMO: self._openclaw_nemo_default,
|
||||
AIProviderEnum.NEMOTRON: self._nemotron_default,
|
||||
}
|
||||
|
||||
# 完整 Fallback 鏈 (Provider, Model)
|
||||
# 2026-03-30 ogt: NVIDIA 成為首選仲裁，加入 Fallback 鏈首位
|
||||
# 2026-04-02 ogt: C1 修復 — OPENCLAW_NEMO 首選仲裁
|
||||
self._full_fallback_chain: list[tuple[AIProviderEnum, str]] = [
|
||||
(AIProviderEnum.NVIDIA, self._nvidia_default),
|
||||
(AIProviderEnum.OPENCLAW_NEMO, self._openclaw_nemo_default),
|
||||
(AIProviderEnum.GEMINI, self._gemini_default),
|
||||
(AIProviderEnum.CLAUDE, self._claude_default),
|
||||
(AIProviderEnum.OLLAMA, self._ollama_default),
|
||||
@@ -232,7 +241,7 @@ class AIRouter:
|
||||
|
||||
# Tool Calling 專用 Fallback 鏈 (ADR-036)
|
||||
self._tool_calling_fallback_chain: list[tuple[AIProviderEnum, str]] = [
|
||||
(AIProviderEnum.NVIDIA, self._nvidia_default),
|
||||
(AIProviderEnum.NEMOTRON, self._nemotron_default),
|
||||
(AIProviderEnum.GEMINI, self._gemini_default),
|
||||
(AIProviderEnum.CLAUDE, self._claude_default),
|
||||
]
|
||||
@@ -394,12 +403,13 @@ class AIRouter:
|
||||
return provider, model, reason
|
||||
|
||||
# =======================================================================
|
||||
# 規則 4: 複雜度 4-5 或 HIGH 風險 → Nvidia Nemotron
|
||||
# 規則 4: 複雜度 4-5 或 HIGH 風險 → OpenClaw Nemo (via .188 → NVIDIA NIM)
|
||||
# 2026-04-02 ogt: C1 修復 — NVIDIA→OPENCLAW_NEMO 對齊 Registry 名稱
|
||||
# =======================================================================
|
||||
if score >= 4 or risk == RiskLevel.HIGH:
|
||||
provider = AIProviderEnum.NVIDIA
|
||||
model = self._nvidia_default
|
||||
reason = f"複雜度={score}/5, 風險={risk.value} → Nvidia (fallback Gemini)"
|
||||
provider = AIProviderEnum.OPENCLAW_NEMO
|
||||
model = self._openclaw_nemo_default
|
||||
reason = f"複雜度={score}/5, 風險={risk.value} → OpenClaw Nemo (fallback Gemini)"
|
||||
return provider, model, reason
|
||||
|
||||
# =======================================================================
|
||||
@@ -538,8 +548,9 @@ class AIRouter:
|
||||
Returns:
|
||||
(provider, model, fallback_chain)
|
||||
"""
|
||||
provider = AIProviderEnum.NVIDIA
|
||||
model = self._nvidia_default
|
||||
# 2026-04-02 ogt: C1 修復 — Tool Calling 使用 NEMOTRON (direct NIM)
|
||||
provider = AIProviderEnum.NEMOTRON
|
||||
model = self._nemotron_default
|
||||
fallback_chain = [
|
||||
(p, m) for p, m in self._tool_calling_fallback_chain if p != provider
|
||||
]
|
||||
@@ -599,8 +610,8 @@ class AIRouter:
|
||||
{
|
||||
"rule": 4,
|
||||
"condition": "complexity >= 4 OR HIGH risk",
|
||||
"provider": "nvidia",
|
||||
"reason": "高複雜度需要 Nvidia Nemotron 強大推理能力",
|
||||
"provider": "openclaw_nemo",
|
||||
"reason": "高複雜度需要 Nvidia Nemotron 強大推理能力 (via .188)",
|
||||
},
|
||||
{
|
||||
"rule": 5,
|
||||
@@ -852,8 +863,8 @@ class AIRouterExecutor:
|
||||
continue
|
||||
|
||||
# 閘門 2: Rate Limiter
|
||||
# 2026-04-02 Claude Code: Phase 24 B3 — 加入 nemotron Rate Limiter
|
||||
if provider_name in ("nvidia", "gemini", "claude", "nemotron"):
|
||||
# 2026-04-02 Claude Code: Phase 24 B3 + C1 修復 — Rate Limiter (含 openclaw_nemo)
|
||||
if provider_name in ("openclaw_nemo", "nemotron", "gemini", "claude"):
|
||||
try:
|
||||
from src.services.ai_rate_limiter import get_ai_rate_limiter
|
||||
rate_limiter = get_ai_rate_limiter()
|
||||
|
||||
Reference in New Issue
Block a user