Spaces:
Configuration error
Configuration error
File size: 12,106 Bytes
d4398e6 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 | """
FastAPI Deployment Server
==========================
One-click deployment bridge for fine-tuned models.
"""
import os
from pathlib import Path
from typing import Optional, List, Dict, Any
from dataclasses import dataclass
from datetime import datetime
from rich.console import Console
console = Console()
@dataclass
class GenerationRequest:
"""Request model for text generation."""
prompt: str
system_prompt: Optional[str] = None
max_tokens: int = 512
temperature: float = 0.7
top_p: float = 0.9
stream: bool = False
@dataclass
class GenerationResponse:
"""Response model for text generation."""
generated_text: str
prompt: str
model: str
tokens_generated: int
generation_time: float
class DeploymentServer:
"""
FastAPI-based deployment server for fine-tuned models.
Features:
- RESTful API for inference
- Health check endpoint
- Batch generation support
- Automatic model loading
"""
def __init__(
self,
model_path: str,
host: str = "0.0.0.0",
port: int = 8000,
max_seq_length: int = 2048
):
"""
Initialize the deployment server.
Args:
model_path: Path to the fine-tuned model
host: Server host
port: Server port
max_seq_length: Maximum sequence length
"""
self.model_path = model_path
self.host = host
self.port = port
self.max_seq_length = max_seq_length
self.model = None
self.tokenizer = None
self.app = None
def load_model(self):
"""Load the fine-tuned model."""
console.print(f"\n[bold blue]📂 Loading model from:[/] {self.model_path}")
try:
from unsloth import FastLanguageModel
self.model, self.tokenizer = FastLanguageModel.from_pretrained(
model_name=self.model_path,
max_seq_length=self.max_seq_length,
dtype=None,
load_in_4bit=True,
)
FastLanguageModel.for_inference(self.model)
console.print("[green]✓ Model loaded successfully[/]")
except ImportError:
console.print("[yellow]⚠️ Unsloth not available, trying transformers...[/]")
from transformers import AutoModelForCausalLM, AutoTokenizer
self.tokenizer = AutoTokenizer.from_pretrained(self.model_path)
self.model = AutoModelForCausalLM.from_pretrained(
self.model_path,
device_map="auto",
torch_dtype="auto"
)
console.print("[green]✓ Model loaded with transformers[/]")
def generate(
self,
prompt: str,
system_prompt: Optional[str] = None,
max_tokens: int = 512,
temperature: float = 0.7,
top_p: float = 0.9
) -> GenerationResponse:
"""
Generate text from the model.
Args:
prompt: User prompt
system_prompt: Optional system prompt
max_tokens: Maximum tokens to generate
temperature: Sampling temperature
top_p: Top-p sampling parameter
Returns:
GenerationResponse with generated text
"""
if self.model is None:
raise RuntimeError("Model not loaded. Call load_model() first.")
start_time = datetime.now()
# Format prompt with Alpaca template
if system_prompt:
formatted_prompt = f"""{system_prompt}
### Instruction:
{prompt}
### Response:
"""
else:
formatted_prompt = f"""Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
"""
# Tokenize
inputs = self.tokenizer(
formatted_prompt,
return_tensors="pt"
).to(self.model.device)
# Generate
outputs = self.model.generate(
**inputs,
max_new_tokens=max_tokens,
temperature=temperature,
top_p=top_p,
do_sample=True,
pad_token_id=self.tokenizer.eos_token_id
)
# Decode
full_response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
# Extract just the generated part
if "### Response:" in full_response:
generated_text = full_response.split("### Response:")[-1].strip()
else:
generated_text = full_response[len(formatted_prompt):].strip()
generation_time = (datetime.now() - start_time).total_seconds()
tokens_generated = len(self.tokenizer.encode(generated_text))
return GenerationResponse(
generated_text=generated_text,
prompt=prompt,
model=self.model_path,
tokens_generated=tokens_generated,
generation_time=generation_time
)
def create_app(self):
"""Create the FastAPI application."""
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
from typing import List, Optional
app = FastAPI(
title="Auto-FineTune-Ops Inference API",
description="API for serving fine-tuned LLM models",
version="1.0.0"
)
# CORS middleware
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Pydantic models for API
class GenerateRequest(BaseModel):
prompt: str
system_prompt: Optional[str] = None
max_tokens: int = 512
temperature: float = 0.7
top_p: float = 0.9
class GenerateResponse(BaseModel):
generated_text: str
prompt: str
model: str
tokens_generated: int
generation_time: float
class BatchGenerateRequest(BaseModel):
prompts: List[str]
system_prompt: Optional[str] = None
max_tokens: int = 512
temperature: float = 0.7
top_p: float = 0.9
class HealthResponse(BaseModel):
status: str
model: str
model_loaded: bool
@app.get("/health", response_model=HealthResponse)
async def health_check():
"""Health check endpoint."""
return HealthResponse(
status="healthy",
model=self.model_path,
model_loaded=self.model is not None
)
@app.post("/generate", response_model=GenerateResponse)
async def generate_text(request: GenerateRequest):
"""Generate text from a single prompt."""
if self.model is None:
raise HTTPException(status_code=503, detail="Model not loaded")
try:
result = self.generate(
prompt=request.prompt,
system_prompt=request.system_prompt,
max_tokens=request.max_tokens,
temperature=request.temperature,
top_p=request.top_p
)
return GenerateResponse(
generated_text=result.generated_text,
prompt=result.prompt,
model=result.model,
tokens_generated=result.tokens_generated,
generation_time=result.generation_time
)
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
@app.post("/generate/batch", response_model=List[GenerateResponse])
async def batch_generate(request: BatchGenerateRequest):
"""Generate text from multiple prompts."""
if self.model is None:
raise HTTPException(status_code=503, detail="Model not loaded")
results = []
for prompt in request.prompts:
try:
result = self.generate(
prompt=prompt,
system_prompt=request.system_prompt,
max_tokens=request.max_tokens,
temperature=request.temperature,
top_p=request.top_p
)
results.append(GenerateResponse(
generated_text=result.generated_text,
prompt=result.prompt,
model=result.model,
tokens_generated=result.tokens_generated,
generation_time=result.generation_time
))
except Exception as e:
results.append(GenerateResponse(
generated_text=f"Error: {str(e)}",
prompt=prompt,
model=self.model_path,
tokens_generated=0,
generation_time=0.0
))
return results
@app.get("/")
async def root():
"""Root endpoint with API info."""
return {
"name": "Auto-FineTune-Ops Inference API",
"version": "1.0.0",
"model": self.model_path,
"endpoints": {
"/health": "Health check",
"/generate": "Generate text (POST)",
"/generate/batch": "Batch generation (POST)"
}
}
self.app = app
return app
def run(self, reload: bool = False):
"""
Start the FastAPI server.
Args:
reload: Enable auto-reload for development
"""
import uvicorn
console.print("\n" + "="*60)
console.print("[bold magenta]🚀 DEPLOYMENT SERVER[/]")
console.print("="*60)
# Load model if not already loaded
if self.model is None:
self.load_model()
# Create app if not already created
if self.app is None:
self.create_app()
console.print(f"\n[bold green]Starting server at http://{self.host}:{self.port}[/]")
console.print("[dim]Press Ctrl+C to stop[/]\n")
uvicorn.run(
self.app,
host=self.host,
port=self.port,
reload=reload
)
def main():
"""CLI entry point for deployment."""
import argparse
parser = argparse.ArgumentParser(description="Deploy fine-tuned model as API")
parser.add_argument("--model", required=True, help="Path to fine-tuned model")
parser.add_argument("--host", default="0.0.0.0", help="Server host")
parser.add_argument("--port", type=int, default=8000, help="Server port")
parser.add_argument("--reload", action="store_true", help="Enable auto-reload")
args = parser.parse_args()
server = DeploymentServer(
model_path=args.model,
host=args.host,
port=args.port
)
server.run(reload=args.reload)
if __name__ == "__main__":
main()
|