File size: 1,494 Bytes
3466d4c
 
bc6423e
 
3466d4c
 
 
 
 
 
 
 
 
bc6423e
 
 
 
 
 
 
 
3466d4c
 
 
 
bc6423e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

app = FastAPI()

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)
# ===== 載入模型 =====
# 換成你自己的模型 repo 名稱
MODEL_REPO = "raxhel/Malvec"

# 如果是私有模型,會自動從 HF_TOKEN 抓取權限
tokenizer = AutoTokenizer.from_pretrained(MODEL_REPO, use_auth_token=True)
model = AutoModelForSequenceClassification.from_pretrained(MODEL_REPO, use_auth_token=True)
model.eval()

@app.get("/")
def home():
    return {"message": "FastAPI Space is running!"}

@app.post("/test")
def test_model(data: dict):
    text = data.get("text", "")
    if not text:
        return {"error": "請提供 text"}

    # Tokenize
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)

    # 模型推論
    with torch.no_grad():
        outputs = model(**inputs)

    # 取 logits
    logits = outputs.logits
    probs = torch.softmax(logits, dim=-1)
    pred_label_id = torch.argmax(probs, dim=-1).item()
    confidence = probs[0][pred_label_id].item()

    # 若 config 裡有 id2label
    label = model.config.id2label.get(pred_label_id, str(pred_label_id))

    return {
        "input_text": text,
        "predicted_label": label,
        "confidence": round(confidence, 4),
        "logits": logits.tolist()
    }