File size: 6,331 Bytes
ceb778d |
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 |
#!/usr/bin/env python3
"""
FastAPI server for serving Mistral 7B fine-tuned models
"""
import os
import sys
from typing import Optional, Dict, Any
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
import uvicorn
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent.parent))
from inference.inference_mistral7b import load_local_model, generate_with_local_model, get_device_info
import torch
# Configuration - Resolve model path relative to msp root
_MODEL_BASE = Path(__file__).parent.parent / "mistral7b-finetuned-ahb2apb"
DEFAULT_MODEL_PATH = str(_MODEL_BASE)
# Global model and tokenizer (loaded once at startup)
model = None
tokenizer = None
device_info = None
app = FastAPI(
title="Mistral 7B AHB2APB API",
description="API for serving the fine-tuned Mistral 7B model for AHB2APB conversion",
version="1.0.0"
)
# Enable CORS
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Request/Response models
class GenerateRequest(BaseModel):
prompt: str
max_length: Optional[int] = 512
temperature: Optional[float] = 0.7
class GenerateResponse(BaseModel):
response: str
model: str
max_length: int
temperature: float
class HealthResponse(BaseModel):
status: str
model_loaded: bool
device: str
model_path: str
@app.on_event("startup")
async def load_model():
"""Load the model when the server starts"""
global model, tokenizer, device_info
model_path = os.environ.get("MODEL_PATH", DEFAULT_MODEL_PATH)
print(f"Loading model from: {model_path}")
print("=" * 70)
try:
device_info = get_device_info()
model, tokenizer = load_local_model(model_path)
print(f"\n✓ Model loaded successfully on {device_info['device']}!")
print(f"✓ Server ready to accept requests")
print("=" * 70)
except Exception as e:
print(f"\n✗ Error loading model: {e}")
print("=" * 70)
sys.exit(1)
@app.get("/health", response_model=HealthResponse)
async def health_check():
"""Health check endpoint"""
return HealthResponse(
status="healthy" if model is not None else "error",
model_loaded=model is not None,
device=device_info["device"] if device_info else "unknown",
model_path=os.environ.get("MODEL_PATH", DEFAULT_MODEL_PATH)
)
@app.get("/")
async def root():
"""Root endpoint with API information"""
return {
"name": "Mistral 7B AHB2APB API",
"version": "1.0.0",
"status": "running",
"model": os.environ.get("MODEL_PATH", DEFAULT_MODEL_PATH),
"endpoints": {
"health": "/health",
"generate": "/api/generate",
"docs": "/docs"
}
}
@app.post("/api/generate", response_model=GenerateResponse)
async def generate(request: GenerateRequest):
"""
Generate text from a prompt using the fine-tuned model
"""
if model is None or tokenizer is None:
raise HTTPException(status_code=503, detail="Model not loaded")
try:
response = generate_with_local_model(
model=model,
tokenizer=tokenizer,
prompt=request.prompt,
max_length=request.max_length or 512,
temperature=request.temperature or 0.7
)
return GenerateResponse(
response=response,
model=os.environ.get("MODEL_PATH", DEFAULT_MODEL_PATH),
max_length=request.max_length or 512,
temperature=request.temperature or 0.7
)
except Exception as e:
raise HTTPException(status_code=500, detail=f"Generation error: {str(e)}")
@app.post("/api/generate/batch")
async def generate_batch(requests: list[GenerateRequest]):
"""
Generate text from multiple prompts (batch processing)
"""
if model is None or tokenizer is None:
raise HTTPException(status_code=503, detail="Model not loaded")
try:
responses = []
for req in requests:
response = generate_with_local_model(
model=model,
tokenizer=tokenizer,
prompt=req.prompt,
max_length=req.max_length or 512,
temperature=req.temperature or 0.7
)
responses.append({
"response": response,
"prompt": req.prompt
})
return {"results": responses}
except Exception as e:
raise HTTPException(status_code=500, detail=f"Batch generation error: {str(e)}")
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser(description="Start Mistral 7B API server")
parser.add_argument(
"--model-path",
type=str,
default=DEFAULT_MODEL_PATH,
help=f"Path to fine-tuned model (default: {DEFAULT_MODEL_PATH})"
)
parser.add_argument(
"--host",
type=str,
default="0.0.0.0",
help="Host to bind to (default: 0.0.0.0)"
)
parser.add_argument(
"--port",
type=int,
default=8000,
help="Port to bind to (default: 8000)"
)
parser.add_argument(
"--reload",
action="store_true",
help="Enable auto-reload (for development)"
)
parser.add_argument(
"--workers",
type=int,
default=1,
help="Number of worker processes (default: 1)"
)
args = parser.parse_args()
# Set model path as environment variable for the startup event
os.environ["MODEL_PATH"] = args.model_path
print(f"\n🚀 Starting Mistral 7B AHB2APB API Server")
print(f" Model: {args.model_path}")
print(f" Host: {args.host}")
print(f" Port: {args.port}")
print(f" Workers: {args.workers}")
print(f" Reload: {args.reload}\n")
# Change to api directory for proper module resolution
import os
os.chdir(os.path.dirname(os.path.abspath(__file__)))
uvicorn.run(
"api_server:app",
host=args.host,
port=args.port,
reload=args.reload,
workers=1 if args.reload else args.workers
)
|