daarion-admin/microdao-daarion
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(fabric): decommission Swapper from critical path, NCS = source of truth

Browse Source 
- Node Worker: replace swapper_vision with ollama_vision (direct Ollama API)
- Node Worker: add NATS subjects for stt/tts/image (stubs ready)
- Node Worker: remove SWAPPER_URL dependency from config
- Router: vision calls go directly to Ollama /api/generate with images
- Router: local LLM calls go directly to Ollama /api/generate
- Router: add OLLAMA_URL and PREFER_NODE_WORKER=true feature flag
- Router: /v1/models now uses NCS global capabilities pool
- NCS: SWAPPER_URL="" -> skip Swapper probing (status=disabled)
- Swapper configs: remove all hardcoded model lists, keep only runtime
  URLs, timeouts, limits
- docker-compose.node1.yml: add OLLAMA_URL, PREFER_NODE_WORKER for router;
  SWAPPER_URL= for NCS; remove swapper-service from node-worker depends_on
- docker-compose.node2-sofiia.yml: same changes for NODA2

Swapper service still runs but is NOT in the critical inference path.
Source of truth for models is now NCS -> Ollama /api/tags.

Made-with: Cursor
This commit is contained in:
Apple
2026-02-27 04:16:16 -08:00
parent 90080c632a
commit 194c87f53c
11 changed files with 347 additions and 614 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
services/node-capabilities/main.py
View File

@@ -20,7 +20,7 @@ app = FastAPI(title="Node Capabilities Service", version="1.0.0")
NODE_ID = os.getenv("NODE_ID", "noda2")
OLLAMA_BASE_URL = os.getenv("OLLAMA_BASE_URL", "http://host.docker.internal:11434")
SWAPPER_URL = os.getenv("SWAPPER_URL", "http://swapper-service:8890")
SWAPPER_URL = os.getenv("SWAPPER_URL", "") # empty = skip Swapper probing
LLAMA_SERVER_URL = os.getenv("LLAMA_SERVER_URL", "")
_cache: Dict[str, Any] = {}
@@ -74,7 +74,10 @@ async def _collect_ollama() -> Dict[str, Any]:
async def _collect_swapper() -> Dict[str, Any]:
runtime: Dict[str, Any] = {"base_url": SWAPPER_URL, "status": "unknown", "models": [], "vision_models": [], "active_model": None}
runtime: Dict[str, Any] = {"base_url": SWAPPER_URL or "n/a", "status": "unknown", "models": [], "vision_models": [], "active_model": None}
if not SWAPPER_URL:
runtime["status"] = "disabled"
return runtime
try:
async with httpx.AsyncClient(timeout=5) as c:
h = await c.get(f"{SWAPPER_URL}/health")

1
services/node-worker/config.py
View File

@@ -4,7 +4,6 @@ import os
NODE_ID = os.getenv("NODE_ID", "noda2")
NATS_URL = os.getenv("NATS_URL", "nats://dagi-nats:4222")
OLLAMA_BASE_URL = os.getenv("OLLAMA_BASE_URL", "http://host.docker.internal:11434")
SWAPPER_URL = os.getenv("SWAPPER_URL", "http://swapper-service:8890")
DEFAULT_LLM = os.getenv("NODE_DEFAULT_LLM", "qwen3:14b")
DEFAULT_VISION = os.getenv("NODE_DEFAULT_VISION", "llava:13b")
MAX_CONCURRENCY = int(os.getenv("NODE_WORKER_MAX_CONCURRENCY", "2"))

49
services/node-worker/providers/ollama_vision.py Normal file
View File

@@ -0,0 +1,49 @@
"""Ollama vision provider — direct Ollama API with images, no Swapper dependency."""
import logging
from typing import Any, Dict, List, Optional
import httpx
from config import OLLAMA_BASE_URL, DEFAULT_VISION
logger = logging.getLogger("provider.ollama_vision")
async def infer(
images: Optional[List[str]] = None,
prompt: str = "",
model: str = "",
system: str = "",
max_tokens: int = 1024,
temperature: float = 0.2,
timeout_s: float = 60.0,
) -> Dict[str, Any]:
model = model or DEFAULT_VISION
payload: Dict[str, Any] = {
"model": model,
"prompt": prompt or "Describe this image.",
"stream": False,
"options": {"num_predict": max_tokens, "temperature": temperature},
}
if images:
clean = []
for img in images:
if "," in img and img.startswith("data:"):
clean.append(img.split(",", 1)[1])
else:
clean.append(img)
payload["images"] = clean
if system:
payload["system"] = system
async with httpx.AsyncClient(timeout=timeout_s) as c:
resp = await c.post(f"{OLLAMA_BASE_URL}/api/generate", json=payload)
resp.raise_for_status()
data = resp.json()
return {
"text": data.get("response", ""),
"model": model,
"provider": "ollama_vision",
"eval_count": data.get("eval_count", 0),
}

23
services/node-worker/worker.py
View File

@@ -9,7 +9,7 @@ from typing import Any, Dict
import config
from models import JobRequest, JobResponse, JobError
from idempotency import IdempotencyStore
from providers import ollama, swapper_vision
from providers import ollama, ollama_vision
import fabric_metrics as fm
logger = logging.getLogger("node-worker")
@@ -27,9 +27,13 @@ async def start(nats_client):
global _nats_client
_nats_client = nats_client
nid = config.NODE_ID.lower()
subjects = [
f"node.{config.NODE_ID.lower()}.llm.request",
f"node.{config.NODE_ID.lower()}.vision.request",
f"node.{nid}.llm.request",
f"node.{nid}.vision.request",
f"node.{nid}.stt.request",
f"node.{nid}.tts.request",
f"node.{nid}.image.request",
]
for subj in subjects:
await nats_client.subscribe(subj, cb=_handle_request)
@@ -160,7 +164,7 @@ async def _execute(job: JobRequest, remaining_ms: int) -> JobResponse:
)
elif job.required_type == "vision":
result = await asyncio.wait_for(
swapper_vision.infer(
ollama_vision.infer(
images=payload.get("images"),
prompt=payload.get("prompt", ""),
model=model,
@@ -171,11 +175,20 @@ async def _execute(job: JobRequest, remaining_ms: int) -> JobResponse:
),
timeout=timeout_s,
)
elif job.required_type in ("stt", "tts", "image"):
return JobResponse(
job_id=job.job_id, trace_id=job.trace_id, node_id=config.NODE_ID,
status="error",
error=JobError(
code="NOT_YET_IMPLEMENTED",
message=f"{job.required_type} adapter coming soon; use direct runtime API for now",
),
)
else:
return JobResponse(
job_id=job.job_id, trace_id=job.trace_id, node_id=config.NODE_ID,
status="error",
error=JobError(code="UNSUPPORTED_TYPE", message=f"{job.required_type} not implemented"),
error=JobError(code="UNSUPPORTED_TYPE", message=f"{job.required_type} not supported"),
)
logger.info(

298
services/router/main.py
View File

@@ -877,12 +877,13 @@ app = FastAPI(title="DAARION Router", version="2.0.0")
# Configuration
NATS_URL = os.getenv("NATS_URL", "nats://nats:4222")
SWAPPER_URL = os.getenv("SWAPPER_URL", "http://swapper-service:8890")
# All multimodal services now through Swapper
STT_URL = os.getenv("STT_URL", "http://swapper-service:8890") # Swapper /stt endpoint
TTS_URL = os.getenv("TTS_URL", "http://swapper-service:8890") # Swapper /tts endpoint
VISION_URL = os.getenv("VISION_URL", "http://172.18.0.1:11434") # Host Ollama
OCR_URL = os.getenv("OCR_URL", "http://swapper-service:8890") # Swapper /ocr endpoint
DOCUMENT_URL = os.getenv("DOCUMENT_URL", "http://swapper-service:8890") # Swapper /document endpoint
OLLAMA_URL = os.getenv("OLLAMA_URL", "http://host.docker.internal:11434")
PREFER_NODE_WORKER = os.getenv("PREFER_NODE_WORKER", "true").lower() in ("true", "1")
STT_URL = os.getenv("STT_URL", "http://swapper-service:8890")
TTS_URL = os.getenv("TTS_URL", "http://swapper-service:8890")
VISION_URL = os.getenv("VISION_URL", "http://host.docker.internal:11434")
OCR_URL = os.getenv("OCR_URL", "http://swapper-service:8890")
DOCUMENT_URL = os.getenv("DOCUMENT_URL", "http://swapper-service:8890")
CITY_SERVICE_URL = os.getenv("CITY_SERVICE_URL", "http://daarion-city-service:7001")
# CrewAI Routing Configuration
@@ -1083,8 +1084,8 @@ async def startup_event():
runtime_guard_engine = None
# Log backend URLs
logger.info(f"📡 Swapper URL: {SWAPPER_URL}")
logger.info(f"📡 STT URL: {STT_URL}")
logger.info(f"📡 Ollama URL: {OLLAMA_URL} (prefer_node_worker={PREFER_NODE_WORKER})")
logger.info(f"📡 Swapper URL: {SWAPPER_URL} (legacy, being decommissioned)")
logger.info(f"📡 Vision URL: {VISION_URL}")
logger.info(f"📡 OCR URL: {OCR_URL}")
logger.info(f"📡 Neo4j URL: {NEO4J_URI}")
@@ -2388,33 +2389,39 @@ async def agent_infer(agent_id: str, request: InferRequest):
logger.warning(f"⚠️ Deterministic AgroMatrix plant flow failed, fallback to generic vision: {e}")
try:
# Use Swapper's /vision endpoint (manages model loading)
vision_payload = {
"model": "qwen3-vl-8b",
"prompt": request.prompt,
"images": request.images, # Swapper handles data URL conversion
"max_tokens": request.max_tokens or 1024,
"temperature": request.temperature or 0.7
}
# Add system prompt if available
if system_prompt:
if memory_brief_text:
vision_payload["system"] = f"{system_prompt}\n\n[INTERNAL MEMORY - do NOT repeat to user]\n{memory_brief_text}"
vision_model = "qwen3-vl-8b"
vision_system = system_prompt or ""
if vision_system and memory_brief_text:
vision_system = f"{vision_system}\n\n[INTERNAL MEMORY - do NOT repeat to user]\n{memory_brief_text}"
clean_images = []
for img in (request.images or []):
if "," in img and img.startswith("data:"):
clean_images.append(img.split(",", 1)[1])
else:
vision_payload["system"] = system_prompt
logger.info(f"🖼️ Sending to Swapper /vision: {SWAPPER_URL}/vision")
clean_images.append(img)
logger.info(f"🖼️ Vision inference: model={vision_model} images={len(clean_images)} prefer_nw={PREFER_NODE_WORKER}")
vision_resp = await http_client.post(
f"{SWAPPER_URL}/vision",
json=vision_payload,
timeout=120.0
f"{OLLAMA_URL}/api/generate",
json={
"model": vision_model.replace("-", "-vl:").replace("qwen3-vl:", "qwen3-vl:") if ":" not in vision_model else vision_model,
"prompt": request.prompt,
"images": clean_images,
"system": vision_system,
"stream": False,
"options": {
"num_predict": request.max_tokens or 1024,
"temperature": request.temperature or 0.7,
},
},
timeout=120.0,
)
if vision_resp.status_code == 200:
vision_data = vision_resp.json()
raw_response = vision_data.get("text", "")
raw_response = vision_data.get("response", vision_data.get("text", ""))
full_response = _sanitize_vision_text_for_user(raw_response)
vision_web_query = ""
vision_sources: List[Dict[str, str]] = []
@@ -2519,14 +2526,23 @@ async def agent_infer(agent_id: str, request: InferRequest):
"Відповідай українською 2-4 реченнями, без службових фраз. "
f"Запит користувача: {request.prompt}"
)
retry_ollama_payload = {
"model": vision_model.replace("-", "-vl:").replace("qwen3-vl:", "qwen3-vl:") if ":" not in vision_model else vision_model,
"prompt": retry_payload["prompt"],
"images": clean_images,
"stream": False,
"options": {"num_predict": request.max_tokens or 1024, "temperature": 0.7},
}
if retry_payload.get("system"):
retry_ollama_payload["system"] = retry_payload["system"]
retry_resp = await http_client.post(
f"{SWAPPER_URL}/vision",
json=retry_payload,
timeout=120.0
f"{OLLAMA_URL}/api/generate",
json=retry_ollama_payload,
timeout=120.0,
)
if retry_resp.status_code == 200:
retry_data = retry_resp.json()
retry_raw = retry_data.get("text", "")
retry_raw = retry_data.get("response", retry_data.get("text", ""))
retry_text = _sanitize_vision_text_for_user(retry_raw)
if retry_raw and not retry_text:
retry_text = _extract_vision_search_facts(retry_raw, max_chars=280)
@@ -2541,7 +2557,7 @@ async def agent_infer(agent_id: str, request: InferRequest):
elif request_agent_id in DETERMINISTIC_PLANT_POLICY_AGENTS and _vision_response_is_blurry(full_response):
full_response = _build_image_fallback_response(request_agent_id, request.prompt)
full_response = await _finalize_response_text(full_response, "swapper-vision")
full_response = await _finalize_response_text(full_response, "ollama-vision")
# Store vision message in agent-specific memory
if MEMORY_RETRIEVAL_AVAILABLE and memory_retrieval and chat_id and user_id and full_response:
@@ -2567,10 +2583,10 @@ async def agent_infer(agent_id: str, request: InferRequest):
response=full_response,
model="qwen3-vl-8b",
tokens_used=None,
backend="swapper-vision"
backend="ollama-vision"
)
else:
logger.error(f"❌ Swapper vision error: {vision_resp.status_code} - {vision_resp.text[:200]}")
logger.error(f"❌ Ollama vision error: {vision_resp.status_code} - {vision_resp.text[:200]}")
fallback_response = await _finalize_response_text(
_build_image_fallback_response(request_agent_id, request.prompt),
"swapper-vision-fallback",
@@ -2579,7 +2595,7 @@ async def agent_infer(agent_id: str, request: InferRequest):
response=fallback_response,
model="qwen3-vl-8b",
tokens_used=None,
backend="swapper-vision-fallback"
backend="vision-fallback"
)
except Exception as e:
@@ -2592,7 +2608,7 @@ async def agent_infer(agent_id: str, request: InferRequest):
response=fallback_response,
model="qwen3-vl-8b",
tokens_used=None,
backend="swapper-vision-fallback"
backend="vision-fallback"
)
# =========================================================================
@@ -3142,90 +3158,88 @@ async def agent_infer(agent_id: str, request: InferRequest):
logger.warning(f"⚠️ No local model in config, using hardcoded fallback: {local_model}")
try:
# Check if Swapper is available
health_resp = await http_client.get(f"{SWAPPER_URL}/health", timeout=5.0)
if health_resp.status_code == 200:
logger.info(f"📡 Calling Swapper with local model: {local_model}")
# Generate response via Swapper (which handles model loading)
generate_resp = await http_client.post(
f"{SWAPPER_URL}/generate",
json={
"model": local_model,
"prompt": request.prompt,
"system": system_prompt,
"max_tokens": request.max_tokens,
"temperature": request.temperature,
"stream": False
ollama_model = local_model.replace("-", ":") if ":" not in local_model else local_model
logger.info(f"📡 Calling Ollama direct: model={ollama_model}")
generate_resp = await http_client.post(
f"{OLLAMA_URL}/api/generate",
json={
"model": ollama_model,
"prompt": request.prompt,
"system": system_prompt,
"stream": False,
"options": {
"num_predict": request.max_tokens or 2048,
"temperature": request.temperature or 0.7,
},
timeout=300.0
)
if generate_resp.status_code == 200:
data = generate_resp.json()
local_response = _normalize_text_response(data.get("response", ""))
},
timeout=300.0,
)
if generate_resp.status_code == 200:
data = generate_resp.json()
local_response = _normalize_text_response(data.get("response", ""))
# Empty-answer gate for selected local top-level agents.
if request_agent_id in EMPTY_ANSWER_GUARD_AGENTS and _needs_empty_answer_recovery(local_response):
logger.warning(f"⚠️ Empty-answer gate triggered for {request_agent_id}, retrying local generate once")
retry_prompt = (
f"{request.prompt}\n\n"
"Відповідай коротко і конкретно (2-5 речень), без службових або мета-фраз."
)
retry_resp = await http_client.post(
f"{SWAPPER_URL}/generate",
json={
"model": local_model,
"prompt": retry_prompt,
"system": system_prompt,
"max_tokens": request.max_tokens,
"temperature": request.temperature,
"stream": False
},
timeout=300.0
)
if retry_resp.status_code == 200:
retry_data = retry_resp.json()
retry_text = _normalize_text_response(retry_data.get("response", ""))
if retry_text and not _needs_empty_answer_recovery(retry_text):
local_response = retry_text
if _needs_empty_answer_recovery(local_response):
local_response = (
"Я не отримав корисну відповідь з першої спроби. "
"Сформулюй запит коротко ще раз, і я відповім конкретно."
)
local_response = await _finalize_response_text(local_response, "swapper+ollama")
# Store in agent-specific memory
if MEMORY_RETRIEVAL_AVAILABLE and memory_retrieval and chat_id and user_id and local_response:
asyncio.create_task(
memory_retrieval.store_message(
agent_id=request_agent_id,
user_id=user_id,
username=username,
message_text=request.prompt,
response_text=local_response,
chat_id=chat_id
)
)
return InferResponse(
response=local_response,
model=local_model,
tokens_used=data.get("eval_count", 0),
backend="swapper+ollama"
if request_agent_id in EMPTY_ANSWER_GUARD_AGENTS and _needs_empty_answer_recovery(local_response):
logger.warning(f"⚠️ Empty-answer gate triggered for {request_agent_id}, retrying local generate once")
retry_prompt = (
f"{request.prompt}\n\n"
"Відповідай коротко і конкретно (2-5 речень), без службових або мета-фраз."
)
else:
logger.error(f"❌ Swapper error: {generate_resp.status_code} - {generate_resp.text}")
retry_resp = await http_client.post(
f"{OLLAMA_URL}/api/generate",
json={
"model": ollama_model,
"prompt": retry_prompt,
"system": system_prompt,
"stream": False,
"options": {
"num_predict": request.max_tokens or 2048,
"temperature": request.temperature or 0.7,
},
},
timeout=300.0,
)
if retry_resp.status_code == 200:
retry_data = retry_resp.json()
retry_text = _normalize_text_response(retry_data.get("response", ""))
if retry_text and not _needs_empty_answer_recovery(retry_text):
local_response = retry_text
if _needs_empty_answer_recovery(local_response):
local_response = (
"Я не отримав корисну відповідь з першої спроби. "
"Сформулюй запит коротко ще раз, і я відповім конкретно."
)
local_response = await _finalize_response_text(local_response, "ollama-direct")
if MEMORY_RETRIEVAL_AVAILABLE and memory_retrieval and chat_id and user_id and local_response:
asyncio.create_task(
memory_retrieval.store_message(
agent_id=request_agent_id,
user_id=user_id,
username=username,
message_text=request.prompt,
response_text=local_response,
chat_id=chat_id
)
)
return InferResponse(
response=local_response,
model=local_model,
tokens_used=data.get("eval_count", 0),
backend="ollama-direct"
)
else:
logger.error(f"❌ Ollama generate error: {generate_resp.status_code} - {generate_resp.text[:200]}")
except Exception as e:
logger.error(f"❌ Swapper/Ollama error: {e}")
# Fallback to direct Ollama if Swapper fails
logger.error(f"❌ Ollama direct error: {e}")
try:
logger.info(f"🔄 Falling back to direct Ollama connection")
logger.info(f"🔄 Falling back to Ollama with hardcoded model")
generate_resp = await http_client.post(
f"{VISION_URL}/api/generate",
f"{OLLAMA_URL}/api/generate",
json={
"model": "qwen3:8b", # Use actual Ollama model name
"model": "qwen3:8b",
"prompt": request.prompt,
"system": system_prompt,
"stream": False,
@@ -3526,42 +3540,38 @@ async def documents_versions(doc_id: str, agent_id: str, limit: int = 20):
@app.get("/v1/models")
async def list_available_models():
"""List all available models across backends"""
"""List all available models from NCS (global capabilities pool)."""
models = []
# Get Swapper models
try:
resp = await http_client.get(f"{SWAPPER_URL}/models", timeout=5.0)
if resp.status_code == 200:
data = resp.json()
for m in data.get("models", []):
models.append({
"id": m.get("name"),
"backend": "swapper",
"size_gb": m.get("size_gb"),
"status": m.get("status", "available")
})
from global_capabilities_client import get_global_capabilities
pool = await get_global_capabilities()
for m in pool.get("served_models", []):
models.append({
"id": m.get("name"),
"backend": m.get("runtime", "unknown"),
"node": m.get("node", "?"),
"type": m.get("type", "llm"),
"size_gb": m.get("size_gb"),
"status": "served",
})
except Exception as e:
logger.warning(f"Cannot get Swapper models: {e}")
# Get Ollama models
try:
resp = await http_client.get(f"{VISION_URL}/api/tags", timeout=5.0)
if resp.status_code == 200:
data = resp.json()
for m in data.get("models", []):
# Avoid duplicates
model_name = m.get("name")
if not any(x.get("id") == model_name for x in models):
logger.warning(f"Cannot get NCS global models: {e}")
if not models:
try:
resp = await http_client.get(f"{OLLAMA_URL}/api/tags", timeout=5.0)
if resp.status_code == 200:
for m in resp.json().get("models", []):
models.append({
"id": model_name,
"id": m.get("name"),
"backend": "ollama",
"size_gb": round(m.get("size", 0) / 1e9, 1),
"status": "loaded"
"status": "loaded",
})
except Exception as e:
logger.warning(f"Cannot get Ollama models: {e}")
except Exception as e:
logger.warning(f"Cannot get Ollama models: {e}")
return {"models": models, "total": len(models)}

109
services/swapper-service/config/swapper_config.yaml
View File

@@ -1,90 +1,35 @@
# Swapper Configuration for Node #2 (Development Node)
# Single-active LLM scheduler
# MacBook Pro M4 Max - Apple Silicon (40-core GPU, 64GB RAM)
# Auto-generated configuration with available Ollama models
# Swapper Configuration — Default / Fallback
#
# NOTE: Swapper is now a runtime gateway / executor only.
# Source of truth for models is NCS (Node Capabilities Service).
# No hardcoded model lists — Swapper queries NCS or Ollama /api/tags at startup.
#
# Per-node overrides: swapper_config_node1.yaml, swapper_config_node2.yaml
swapper:
mode: single-active
max_concurrent_models: 1
node_id: default
runtimes:
ollama:
url: http://localhost:11434
timeout: 300
limits:
llm_concurrency: 2
vision_concurrency: 1
max_concurrent_models: 2
model_swap_timeout: 300
gpu_enabled: true
metal_acceleration: true # Apple Silicon GPU acceleration
# Модель для автоматичного завантаження при старті (опціонально)
# Якщо не вказано - моделі завантажуються тільки за запитом
# Рекомендовано: gpt-oss:latest (швидка модель) або phi3:latest (легка модель)
default_model: gpt-oss:latest # Модель активується автоматично при старті
models:
# Fast LLM - GPT-OSS 20B (High Priority) - Main model for general tasks
gpt-oss-latest:
path: ollama:gpt-oss:latest
type: llm
size_gb: 13.0
priority: high
description: "Fast LLM for general tasks and conversations (20.9B params)"
# Lightweight LLM - Phi3 3.8B (High Priority) - Fast responses
phi3-latest:
path: ollama:phi3:latest
type: llm
size_gb: 2.2
priority: high
description: "Lightweight LLM for fast responses (3.8B params)"
# Code Specialist - StarCoder2 3B (Medium Priority) - Code engineering
starcoder2-3b:
path: ollama:starcoder2:3b
type: code
size_gb: 1.7
priority: medium
description: "Code specialist model for code engineering (3B params)"
# Reasoning Model - Mistral Nemo 12.2B (High Priority) - Advanced reasoning
mistral-nemo-12b:
path: ollama:mistral-nemo:12b
type: llm
size_gb: 7.1
priority: high
description: "Advanced reasoning model for complex tasks (12.2B params)"
# Reasoning Model - Gemma2 27B (Medium Priority) - Strategic reasoning
gemma2-27b:
path: ollama:gemma2:27b
type: llm
size_gb: 15.0
priority: medium
description: "Reasoning model for strategic tasks (27.2B params)"
# Code Specialist - DeepSeek Coder 33B (High Priority) - Advanced code tasks
deepseek-coder-33b:
path: ollama:deepseek-coder:33b
type: code
size_gb: 18.0
priority: high
description: "Advanced code specialist model (33B params)"
# Code Specialist - Qwen2.5 Coder 32B (High Priority) - Advanced code tasks
qwen2.5-coder-32b:
path: ollama:qwen2.5-coder:32b
type: code
size_gb: 19.0
priority: high
description: "Advanced code specialist model (32.8B params)"
# Reasoning Model - DeepSeek R1 70B (High Priority) - Strategic reasoning (large model)
deepseek-r1-70b:
path: ollama:deepseek-r1:70b
type: llm
size_gb: 42.0
priority: high
description: "Strategic reasoning model (70.6B params, quantized)"
timeouts:
llm_ms: 120000
vision_ms: 180000
stt_ms: 60000
tts_ms: 60000
gpu:
enabled: false
metal_acceleration: false
storage:
models_dir: /app/models
cache_dir: /app/cache
swap_dir: /app/swap
ollama:
url: http://localhost:11434 # Native Ollama on MacBook (via Pieces OS or brew)
timeout: 300

228
services/swapper-service/config/swapper_config_node1.yaml
View File

@@ -1,186 +1,37 @@
# Swapper Configuration for Node #1 (Production Server)
# Optimized Multimodal Stack: LLM + Vision + OCR + Document + Audio
# Hetzner GEX44 - NVIDIA RTX 4000 SFF Ada (20GB VRAM)
#
# ВАЖЛИВО: Ембедінги через зовнішні API:
# - Text: Cohere API (embed-multilingual-v3.0, 1024 dim)
# - Image: Vision Encoder (OpenCLIP ViT-L/14, 768 dim)
# НЕ використовуємо локальні embedding моделі!
# NOTE: Swapper is now a runtime gateway / executor only.
# Source of truth for models is NCS (Node Capabilities Service).
# No hardcoded model lists — Swapper queries NCS or Ollama /api/tags at startup.
swapper:
mode: multi-active
max_concurrent_models: 4 # LLM + OCR + STT + TTS (до 15GB)
node_id: noda1
runtimes:
ollama:
url: http://172.18.0.1:11434
timeout: 300
# comfyui:
# url: http://127.0.0.1:8188
limits:
llm_concurrency: 2
vision_concurrency: 1
max_concurrent_models: 4
model_swap_timeout: 300
gpu_enabled: true
timeouts:
llm_ms: 120000
vision_ms: 180000
stt_ms: 60000
tts_ms: 60000
image_gen_ms: 300000
gpu:
enabled: true
metal_acceleration: false
default_model: qwen3-8b
lazy_load_ocr: true
lazy_load_audio: true
# Автоматичне вивантаження при нестачі VRAM
auto_unload_on_oom: true
vram_threshold_gb: 18 # Починати вивантажувати при 18GB
models:
# ============================================
# LLM MODELS (Ollama) - тільки qwen3
# ============================================
# Primary LLM - Qwen3 8B (includes math, coding, reasoning)
qwen3-8b:
path: ollama:qwen3:8b
type: llm
size_gb: 5.2
priority: high
description: "Qwen3 8B - primary LLM with math, coding, reasoning capabilities"
capabilities:
- chat
- math
- coding
- reasoning
- multilingual
# ============================================
# VISION MODELS (Ollama)
# ============================================
# Vision Model - Qwen3-VL 8B
qwen3-vl-8b:
path: ollama:qwen3-vl:8b
type: vision
size_gb: 6.1
priority: high
description: "Qwen3-VL 8B for image understanding and visual reasoning"
capabilities:
- image_understanding
- visual_qa
- diagram_analysis
- ocr_basic
# ============================================
# OCR/DOCUMENT MODELS (HuggingFace)
# ============================================
# GOT-OCR2.0 - Best for documents, tables, formulas
got-ocr2:
path: huggingface:stepfun-ai/GOT-OCR2_0
type: ocr
size_gb: 7.0
priority: high
description: "Best OCR for documents, tables, formulas, handwriting"
capabilities:
- documents
- tables
- formulas
- handwriting
- multilingual
# Donut - Document Understanding (no external OCR, 91% CORD)
donut-base:
path: huggingface:naver-clova-ix/donut-base
type: ocr
size_gb: 3.0
priority: high
description: "Document parsing without OCR engine (91% CORD accuracy)"
capabilities:
- document_parsing
- receipts
- forms
- invoices
# Donut fine-tuned for receipts/invoices (CORD dataset)
donut-cord:
path: huggingface:naver-clova-ix/donut-base-finetuned-cord-v2
type: ocr
size_gb: 3.0
priority: medium
description: "Donut fine-tuned for receipts extraction"
capabilities:
- receipts
- invoices
- structured_extraction
# IBM Granite Docling - Document conversion with structure preservation
granite-docling:
path: huggingface:ds4sd/docling-ibm-granite-vision-1b
type: document
size_gb: 2.5
priority: high
description: "IBM Granite Docling for PDF/document structure extraction"
capabilities:
- pdf_conversion
- table_extraction
- formula_extraction
- layout_preservation
- doctags_format
# ============================================
# AUDIO MODELS - STT (Speech-to-Text)
# ============================================
# Faster Whisper Large-v3 - Best STT quality
faster-whisper-large:
path: huggingface:Systran/faster-whisper-large-v3
type: stt
size_gb: 3.0
priority: high
description: "Faster Whisper Large-v3 - best quality, 99 languages"
capabilities:
- speech_recognition
- transcription
- multilingual
- timestamps
- ukrainian
# Whisper Small - Fast/lightweight for quick transcription
whisper-small:
path: huggingface:openai/whisper-small
type: stt
size_gb: 0.5
priority: medium
description: "Whisper Small for fast transcription"
capabilities:
- speech_recognition
- transcription
# ============================================
# AUDIO MODELS - TTS (Text-to-Speech)
# ============================================
# Coqui XTTS-v2 - Best multilingual TTS with Ukrainian support
xtts-v2:
path: huggingface:coqui/XTTS-v2
type: tts
size_gb: 2.0
priority: high
description: "XTTS-v2 multilingual TTS with voice cloning, Ukrainian support"
capabilities:
- text_to_speech
- voice_cloning
- multilingual
- ukrainian
- 17_languages
# ============================================
# IMAGE GENERATION MODELS (HuggingFace/Diffusers)
# ============================================
# FLUX.2 Klein 4B - High quality image generation with lazy loading
flux-klein-4b:
path: huggingface:black-forest-labs/FLUX.2-klein-base-4B
type: image_generation
size_gb: 15.4
priority: medium
description: "FLUX.2 Klein 4B - high quality image generation, lazy loaded on demand"
capabilities:
- text_to_image
- high_quality
- 1024x1024
- artistic
default_params:
num_inference_steps: 50
guidance_scale: 4.0
width: 1024
height: 1024
vram_threshold_gb: 18
storage:
models_dir: /app/models
@@ -188,33 +39,8 @@ storage:
swap_dir: /app/swap
huggingface_cache: /root/.cache/huggingface
ollama:
url: http://172.18.0.1:11434
timeout: 300
huggingface:
device: cuda
torch_dtype: float16
trust_remote_code: true
low_cpu_mem_usage: true
# ============================================
# EMBEDDING SERVICES (External APIs)
# НЕ через Swapper - окремі сервіси!
# ============================================
#
# Text Embeddings:
# Service: Memory Service → Cohere API
# Model: embed-multilingual-v3.0
# Dimension: 1024
# Endpoint: Memory Service handles internally
#
# Image/Multimodal Embeddings:
# Service: Vision Encoder (port 8001)
# Model: OpenCLIP ViT-L/14
# Dimension: 768
# Endpoint: http://vision-encoder:8001/embed
#
# Vector Storage:
# Qdrant (port 6333) - separate collections for text vs image embeddings
# ВАЖЛИВО: НЕ змішувати embedding spaces в одній колекції!

142
services/swapper-service/config/swapper_config_node2.yaml
View File

@@ -1,126 +1,40 @@
# Swapper Configuration for Node #2 (Development Node)
# Single-active LLM scheduler
# MacBook Pro M4 Max - Apple Silicon (40-core GPU, 64GB RAM)
# Auto-generated configuration with available Ollama models
#
# NOTE: Swapper is now a runtime gateway / executor only.
# Source of truth for models is NCS (Node Capabilities Service).
# No hardcoded model lists — Swapper queries NCS or Ollama /api/tags at startup.
swapper:
mode: single-active
node_id: noda2
runtimes:
ollama:
url: http://host.docker.internal:11434
timeout: 300
# mlx:
# stt_model: whisper-large-v3-turbo
# tts_model: kokoro-82m
# comfyui:
# url: http://127.0.0.1:8188
limits:
llm_concurrency: 1
vision_concurrency: 1
max_concurrent_models: 1
model_swap_timeout: 300
gpu_enabled: true
metal_acceleration: true # Apple Silicon GPU acceleration
# Модель для автоматичного завантаження при старті (опціонально)
# Якщо не вказано - моделі завантажуються тільки за запитом
# Рекомендовано: gpt-oss:latest (швидка модель) або phi3:latest (легка модель)
# Стартова модель має бути реально встановлена в Ollama на NODA2
default_model: qwen3:14b # Модель активується автоматично при старті
models:
# Fast LLM - GPT-OSS 20B (High Priority) - Main model for general tasks
gpt-oss-latest:
path: ollama:gpt-oss:latest
type: llm
size_gb: 13.0
priority: high
description: "Fast LLM for general tasks and conversations (20.9B params)"
# Lightweight LLM - Phi3 3.8B (High Priority) - Fast responses
phi3-latest:
path: ollama:phi3:latest
type: llm
size_gb: 2.2
priority: high
description: "Lightweight LLM for fast responses (3.8B params)"
# General Reasoning - Qwen3 14B (High Priority)
qwen3-14b:
path: ollama:qwen3:14b
type: llm
size_gb: 9.3
priority: high
description: "Balanced local model for Sofiia and router fallback"
timeouts:
llm_ms: 120000
vision_ms: 180000
stt_ms: 60000
tts_ms: 60000
image_gen_ms: 300000
# Reasoning Model - Qwen3.5 35B A3B (High Priority)
qwen3.5-35b-a3b:
path: ollama:qwen3.5:35b-a3b
type: llm
size_gb: 22.0
priority: high
description: "Large reasoning model for complex Sofiia requests"
# Reasoning Model - GLM 4.7 Flash (High Priority) - Fast general model
glm-4.7-flash:
path: ollama:glm-4.7-flash:32k
type: llm
size_gb: 19.0
priority: high
description: "Multi-purpose reasoning model (fast context)"
# Reasoning Model - Gemma2 27B (Medium Priority) - Strategic reasoning
gemma2-27b:
path: ollama:gemma2:27b
type: llm
size_gb: 15.0
priority: medium
description: "Reasoning model for strategic tasks (27.2B params)"
# Code Specialist - DeepSeek Coder 33B (High Priority) - Advanced code tasks
deepseek-coder-33b:
path: ollama:deepseek-coder:33b
type: code
size_gb: 18.0
priority: high
description: "Advanced code specialist model (33B params)"
# Code Specialist - Qwen2.5 Coder 32B (High Priority) - Advanced code tasks
qwen2.5-coder-32b:
path: ollama:qwen2.5-coder:32b
type: code
size_gb: 19.0
priority: high
description: "Advanced code specialist model (32.8B params)"
# Reasoning Model - DeepSeek R1 70B (High Priority) - Strategic reasoning (large model)
deepseek-r1-70b:
path: ollama:deepseek-r1:70b
type: llm
size_gb: 42.0
priority: high
description: "Strategic reasoning model (70.6B params, quantized)"
# Vision Model - LLaVA 13B (P0 Fix: NODA2 fallback vision)
# Available in Ollama on NODA2 — used until qwen3-vl:8b is installed
llava-13b:
path: ollama:llava:13b
type: vision
size_gb: 8.0
priority: high
description: "LLaVA 13B vision model (multimodal CLIP+LLM). P0 fallback until qwen3-vl:8b."
vision: true
ollama_model: "llava:13b"
# Vision Model - Qwen3-VL 8B (RECOMMENDED: install with: ollama pull qwen3-vl:8b)
# Better quality than llava:13b. Enable once installed.
# qwen3-vl-8b:
# path: ollama:qwen3-vl:8b
# type: vision
# size_gb: 5.5
# priority: high
# description: "Qwen3-VL 8B — modern vision-language model (recommended)"
# vision: true
# ollama_model: "qwen3-vl:8b"
gpu:
enabled: true
metal_acceleration: true
storage:
models_dir: /app/models
cache_dir: /app/cache
swap_dir: /app/swap
ollama:
url: http://host.docker.internal:11434 # host.docker.internal → native Ollama on MacBook (NODA2 P1 fix)
timeout: 300
# Vision endpoint configuration
# /vision/models returns all models where vision: true
vision:
default_model: llava-13b
ollama_base_url: http://host.docker.internal:11434

82
services/swapper-service/config/swapper_config_node3.yaml
View File

@@ -1,63 +1,37 @@
# Swapper Configuration for Node #3 (AI/ML Workstation)
# Single-active LLM scheduler
# Threadripper PRO + RTX 3090 24GB - GPU-intensive workloads
# Threadripper PRO + RTX 3090 24GB — GPU-intensive workloads
#
# NOTE: Swapper is now a runtime gateway / executor only.
# Source of truth for models is NCS (Node Capabilities Service).
# No hardcoded model lists.
swapper:
mode: single-active
max_concurrent_models: 1
node_id: noda3
runtimes:
ollama:
url: http://localhost:11434
timeout: 300
comfyui:
url: http://127.0.0.1:8188
limits:
llm_concurrency: 2
vision_concurrency: 1
max_concurrent_models: 2
model_swap_timeout: 300
gpu_enabled: true
metal_acceleration: false # NVIDIA GPU, not Apple Silicon
# Модель для автоматичного завантаження при старті
# qwen3-8b - основна модель (4.87 GB), швидка відповідь на перший запит
default_model: qwen3-8b
models:
# Primary LLM - Qwen3 8B (High Priority) - Main model from INFRASTRUCTURE.md
qwen3-8b:
path: ollama:qwen3:8b
type: llm
size_gb: 4.87
priority: high
description: "Primary LLM for general tasks and conversations"
# Vision Model - Qwen3-VL 8B (High Priority) - For image processing
qwen3-vl-8b:
path: ollama:qwen3-vl:8b
type: vision
size_gb: 5.72
priority: high
description: "Vision model for image understanding and processing"
# Qwen2.5 7B Instruct (High Priority)
qwen2.5-7b-instruct:
path: ollama:qwen2.5:7b-instruct-q4_K_M
type: llm
size_gb: 4.36
priority: high
description: "Qwen2.5 7B Instruct model"
# Lightweight LLM - Qwen2.5 3B Instruct (Medium Priority)
qwen2.5-3b-instruct:
path: ollama:qwen2.5:3b-instruct-q4_K_M
type: llm
size_gb: 1.80
priority: medium
description: "Lightweight LLM for faster responses"
# Math Specialist - Qwen2 Math 7B (High Priority)
qwen2-math-7b:
path: ollama:qwen2-math:7b
type: math
size_gb: 4.13
priority: high
description: "Specialized model for mathematical tasks"
timeouts:
llm_ms: 120000
vision_ms: 180000
image_gen_ms: 600000
gpu:
enabled: true
metal_acceleration: false
auto_unload_on_oom: true
vram_threshold_gb: 22
storage:
models_dir: /app/models
cache_dir: /app/cache
swap_dir: /app/swap
ollama:
url: http://ollama:11434 # From Docker container to Ollama service
timeout: 300