File size: 5,334 Bytes
8f6910f | 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 | """FastAPI application for Myanmar Ghost model."""
import logging
from pathlib import Path
from typing import Any, Dict, List, Optional
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel, Field
import torch
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
app = FastAPI(
title="Myanmar Ghost API",
description="Advanced Myanmar Language Understanding Model",
version="1.0.0",
)
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Global model reference
model = None
tokenizer = None
class TextInput(BaseModel):
text: str = Field(..., description="Myanmar text to analyze")
include_prosody: bool = Field(False, description="Include prosody features")
class SentimentResponse(BaseModel):
text: str
sentiment: str
confidence: float
probabilities: Dict[str, float]
class BatchTextInput(BaseModel):
texts: List[str] = Field(..., description="List of Myanmar texts")
class BatchSentimentResponse(BaseModel):
results: List[SentimentResponse]
@app.on_event("startup")
async def startup_event():
"""Load model on startup."""
global model, tokenizer
logger.info("Loading Myanmar Ghost model...")
try:
from transformers import AutoModelForSequenceClassification, AutoTokenizer
model_name = "amkyawdev/Myanmar-Ghost-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)
model.eval()
logger.info(f"Model loaded: {model_name}")
except Exception as e:
logger.warning(f"Could not load model from HuggingFace: {e}")
logger.info("Using placeholder for demonstration")
@app.get("/")
async def root():
"""Root endpoint."""
return {
"name": "Myanmar Ghost API",
"version": "1.0.0",
"status": "online",
}
@app.get("/health")
async def health():
"""Health check endpoint."""
return {
"status": "healthy",
"model_loaded": model is not None,
}
@app.post("/predict", response_model=SentimentResponse)
async def predict(input_data: TextInput) -> SentimentResponse:
"""Predict sentiment for a single text."""
if model is None or tokenizer is None:
raise HTTPException(status_code=503, detail="Model not loaded")
try:
# Tokenize
inputs = tokenizer(
input_data.text,
return_tensors="pt",
truncation=True,
max_length=512,
)
# Predict
with torch.no_grad():
outputs = model(**inputs)
probs = torch.softmax(outputs.logits, dim=-1)[0]
# Get prediction
sentiment_idx = probs.argmax().item()
confidence = probs[sentiment_idx].item()
sentiment_labels = ["negative", "neutral", "positive", "sarcastic"]
sentiment = sentiment_labels[sentiment_idx]
probabilities = {
label: probs[i].item()
for i, label in enumerate(sentiment_labels)
}
return SentimentResponse(
text=input_data.text,
sentiment=sentiment,
confidence=confidence,
probabilities=probabilities,
)
except Exception as e:
logger.error(f"Prediction error: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/predict_batch", response_model=BatchSentimentResponse)
async def predict_batch(input_data: BatchTextInput) -> BatchSentimentResponse:
"""Predict sentiment for multiple texts."""
if model is None or tokenizer is None:
raise HTTPException(status_code=503, detail="Model not loaded")
results = []
try:
for text in input_data.texts:
# Tokenize
inputs = tokenizer(
text,
return_tensors="pt",
truncation=True,
max_length=512,
)
# Predict
with torch.no_grad():
outputs = model(**inputs)
probs = torch.softmax(outputs.logits, dim=-1)[0]
# Get prediction
sentiment_idx = probs.argmax().item()
confidence = probs[sentiment_idx].item()
sentiment_labels = ["negative", "neutral", "positive", "sarcastic"]
sentiment = sentiment_labels[sentiment_idx]
probabilities = {
label: probs[i].item()
for i, label in enumerate(sentiment_labels)
}
results.append(SentimentResponse(
text=text,
sentiment=sentiment,
confidence=confidence,
probabilities=probabilities,
))
return BatchSentimentResponse(results=results)
except Exception as e:
logger.error(f"Batch prediction error: {e}")
raise HTTPException(status_code=500, detail=str(e))
if __name__ == "__main__":
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=8000)
|