File size: 11,531 Bytes
59e6760 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 |
from __future__ import annotations
import os
import json
import logging
import time
from models import OptimizeRequest, QARequest, AutotuneRequest
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
import uvicorn
try:
from ragmint.autotuner import AutoRAGTuner
from ragmint.qa_generator import generate_validation_qa
from ragmint.explainer import explain_results
from ragmint.leaderboard import Leaderboard
from ragmint.tuner import RAGMint
except Exception as e:
AutoRAGTuner = None
generate_validation_qa = None
explain_results = None
Leaderboard = None
RAGMint = None
_import_error = e
else:
_import_error = None
from dotenv import load_dotenv
load_dotenv()
# Logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger("ragmint_mcp_server")
# FastAPI
app = FastAPI(title="Ragmint MCP Server", version="0.1.0")
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
DEFAULT_DATA_DIR = "../data/docs"
LEADERBOARD_STORAGE = "experiments/leaderboard.jsonl"
os.makedirs("../experiments", exist_ok=True)
@app.get("/health")
def health():
return {
"status": "ok",
"ragmint_imported": _import_error is None,
"import_error": str(_import_error) if _import_error else None,
}
@app.post("/optimize_rag")
def optimize_rag(req: OptimizeRequest):
logger.info("Received optimize_rag request: %s", req.json())
if RAGMint is None:
raise HTTPException(
status_code=500,
detail=f"Ragmint imports failed or RAGMint unavailable: {_import_error}"
)
docs_path = req.docs_path or DEFAULT_DATA_DIR
if not os.path.isdir(docs_path):
raise HTTPException(status_code=400, detail=f"docs_path does not exist: {docs_path}")
try:
# Build RAGMint exactly from request
rag = RAGMint(
docs_path=docs_path,
retrievers=req.retriever,
embeddings=req.embedding_model,
rerankers=(req.rerankers or ["mmr"]),
chunk_sizes=req.chunk_sizes,
overlaps=req.overlaps,
strategies=req.strategy,
)
# Validation selection
validation_set = None
validation_choice = (req.validation_choice or "").strip()
default_val_path = os.path.join(docs_path, "validation_qa.json")
# Auto
if not validation_choice:
if os.path.exists(default_val_path):
validation_set = default_val_path
logger.info("Using default validation set: %s", validation_set)
else:
logger.warning("No validation_choice provided and no default found.")
validation_set = None
# Remote HF dataset
elif "/" in validation_choice and not os.path.exists(validation_choice):
validation_set = validation_choice
logger.info("Using Hugging Face validation dataset: %s", validation_set)
# Local file
elif os.path.exists(validation_choice):
validation_set = validation_choice
logger.info("Using local validation dataset: %s", validation_set)
# Generate
elif validation_choice.lower() == "generate":
try:
gen_path = os.path.join(docs_path, "validation_qa.json")
generate_validation_qa(
docs_path=docs_path,
output_path=gen_path,
llm_model=req.llm_model if hasattr(req, "llm_model") else "gemini-2.5-flash-lite"
)
validation_set = gen_path
logger.info("Generated new validation QA set at: %s", validation_set)
except Exception as e:
logger.exception("Failed to generate validation QA dataset: %s", e)
raise HTTPException(status_code=500, detail=f"Failed to generate validation QA dataset: {e}")
# Optimize
start_time = time.time()
best, results = rag.optimize(
validation_set=validation_set,
metric=req.metric,
trials=req.trials,
search_type=req.search_type
)
elapsed = time.time() - start_time
run_id = f"opt_{int(time.time())}"
# Corpus stats
try:
corpus_stats = {
"num_docs": len(rag.documents),
"avg_len": sum(len(d.split()) for d in rag.documents) / max(1, len(rag.documents)),
"corpus_size": sum(len(d) for d in rag.documents),
}
except Exception:
corpus_stats = None
# Leaderboard
try:
if Leaderboard:
lb = Leaderboard()
lb.upload(
run_id=run_id,
best_config=best,
best_score=best.get("faithfulness", best.get("score", 0.0)),
all_results=results,
documents=os.listdir(docs_path),
model=best.get("embedding_model", req.embedding_model),
corpus_stats=corpus_stats,
)
except Exception:
logger.exception("Leaderboard persistence failed for optimize_rag")
return {
"status": "finished",
"run_id": run_id,
"elapsed_seconds": elapsed,
"best_config": best,
"results": results,
"corpus_stats": corpus_stats,
}
except Exception as exc:
logger.exception("optimize_rag failed")
raise HTTPException(status_code=500, detail=str(exc))
@app.post("/autotune_rag")
def autotune_rag(req: AutotuneRequest):
logger.info("Received autotune_rag request: %s", req.json())
if AutoRAGTuner is None or RAGMint is None:
raise HTTPException(
status_code=500,
detail=f"Ragmint autotuner/RAGMint imports failed: {_import_error}"
)
docs_path = req.docs_path or DEFAULT_DATA_DIR
if not os.path.isdir(docs_path):
raise HTTPException(status_code=400, detail=f"docs_path does not exist: {docs_path}")
try:
start_time = time.time()
tuner = AutoRAGTuner(docs_path=docs_path)
rec = tuner.recommend(
embedding_model=req.embedding_model,
num_chunk_pairs=req.num_chunk_pairs
)
chunk_candidates = tuner.suggest_chunk_sizes(
model_name=rec.get("embedding_model"),
num_pairs=int(req.num_chunk_pairs),
step=20
)
chunk_sizes = sorted({c for c, _ in chunk_candidates})
overlaps = sorted({o for _, o in chunk_candidates})
rag = RAGMint(
docs_path=docs_path,
retrievers=[rec["retriever"]],
embeddings=[rec["embedding_model"]],
rerankers=["mmr"],
chunk_sizes=chunk_sizes,
overlaps=overlaps,
strategies=[rec["strategy"]],
)
# Validation selection
validation_set = None
validation_choice = (req.validation_choice or "").strip()
default_val_path = os.path.join(docs_path, "validation_qa.jsonl")
if not validation_choice:
if os.path.exists(default_val_path):
validation_set = default_val_path
logger.info("Using default validation set: %s", validation_set)
else:
logger.warning("No validation_choice provided and no default found.")
validation_set = None
elif "/" in validation_choice and not os.path.exists(validation_choice):
validation_set = validation_choice
elif os.path.exists(validation_choice):
validation_set = validation_choice
elif validation_choice.lower() == "generate":
try:
gen_path = os.path.join(docs_path, "validation_qa.json")
generate_validation_qa(
docs_path=docs_path,
output_path=gen_path,
llm_model=req.llm_model if hasattr(req, "llm_model") else "gemini-2.5-flash-lite",
)
validation_set = gen_path
except Exception as e:
logger.exception("Failed to generate validation QA dataset: %s", e)
raise HTTPException(status_code=500, detail=f"Failed to generate validation QA dataset: {e}")
# Full optimize
best, results = rag.optimize(
validation_set=validation_set,
metric=req.metric,
search_type=req.search_type,
trials=req.trials,
)
elapsed = time.time() - start_time
run_id = f"autotune_{int(time.time())}"
# Corpus stats
try:
corpus_stats = {
"num_docs": len(rag.documents),
"avg_len": sum(len(d.split()) for d in rag.documents) / max(1, len(rag.documents)),
"corpus_size": sum(len(d) for d in rag.documents),
}
except Exception:
corpus_stats = None
# Leaderboard
try:
if Leaderboard:
lb = Leaderboard()
lb.upload(
run_id=run_id,
best_config=best,
best_score=best.get("faithfulness", best.get("score", 0.0)),
all_results=results,
documents=os.listdir(docs_path),
model=best.get("embedding_model", rec.get("embedding_model")),
corpus_stats=corpus_stats,
)
except Exception:
logger.exception("Leaderboard persistence failed for autotune_rag")
return {
"status": "finished",
"run_id": run_id,
"elapsed_seconds": elapsed,
"recommendation": rec,
"chunk_candidates": chunk_candidates,
"best_config": best,
"results": results,
"corpus_stats": corpus_stats,
}
except Exception as exc:
logger.exception("autotune_rag failed")
raise HTTPException(status_code=500, detail=str(exc))
@app.post("/generate_validation_qa")
def generate_qa(req: QARequest):
logger.info("Received generate_validation_qa request: %s", req.json())
if generate_validation_qa is None:
raise HTTPException(status_code=500, detail=f"Ragmint imports failed: {_import_error}")
try:
out_path = f"data/docs/validation_qa.json"
os.makedirs(os.path.dirname(out_path), exist_ok=True)
generate_validation_qa(
docs_path=req.docs_path,
output_path=out_path,
llm_model=req.llm_model,
batch_size=req.batch_size,
min_q=req.min_q,
max_q=req.max_q,
)
with open(out_path, "r", encoding="utf-8") as f:
data = json.load(f)
return {
"status": "finished",
"output_path": out_path,
"preview_count": len(data),
"sample": data[:5],
}
except Exception as exc:
logger.exception("generate_validation_qa failed")
raise HTTPException(status_code=500, detail=str(exc))
# -----------------------
# FastAPI launch
# -----------------------
def main():
uvicorn.run(app, host="0.0.0.0", port=8000, log_level="info")
if __name__ == "__main__":
main()
|