app.py

import spaces
import gradio
import json
import torch
from transformers import AutoTokenizer
from transformers import pipeline
from optimum.onnxruntime import ORTModelForSequenceClassification
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware

# CORS Config
app = FastAPI()

app.add_middleware(
    CORSMiddleware,
    allow_origins=["https://statosphere-3704059fdd7e.c5v4v4jx6pq5.win","https://crunchatize-77a78ffcc6a6.c5v4v4jx6pq5.win","https://crunchatize-2-2b4f5b1479a6.c5v4v4jx6pq5.win","https://tamabotchi-2dba63df3bf1.c5v4v4jx6pq5.win","https://lord-raven.github.io"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

print(f"Is CUDA available: {torch.cuda.is_available()}")
print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")

# "xenova/mobilebert-uncased-mnli" "typeform/mobilebert-uncased-mnli" Fast but small--same as bundled in Statosphere
# "xenova/deberta-v3-base-tasksource-nli" Not impressed
# "Xenova/bart-large-mnli" A bit slow
# "Xenova/distilbert-base-uncased-mnli" "typeform/distilbert-base-uncased-mnli" Bad answers
# "Xenova/deBERTa-v3-base-mnli" "MoritzLaurer/DeBERTa-v3-base-mnli" Still a bit slow and not great answers
# "xenova/nli-deberta-v3-small" "cross-encoder/nli-deberta-v3-small" Was using this for a good while and it was...okay
# model_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
# file_name = "onnx/model.onnx"
# tokenizer_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
# model = ORTModelForSequenceClassification.from_pretrained(model_name, file_name=file_name, provider="CUDAExecutionProvider")
# tokenizer = AutoTokenizer.from_pretrained(tokenizer_name, model_max_length=512)

model = ORTModelForSequenceClassification.from_pretrained(
    "philschmid/tiny-bert-sst2-distilled",
    export=True,
    provider="CUDAExecutionProvider",
)

tokenizer = AutoTokenizer.from_pretrained("philschmid/tiny-bert-sst2-distilled")

classifier = pipeline(task="zero-shot-classification", model=model, tokenizer=tokenizer, device="cuda:0")

def classify(data_string, request: gradio.Request):
    if request:
        if request.headers["origin"] not in ["https://statosphere-3704059fdd7e.c5v4v4jx6pq5.win", "https://crunchatize-77a78ffcc6a6.c5v4v4jx6pq5.win", "https://crunchatize-2-2b4f5b1479a6.c5v4v4jx6pq5.win", "https://tamabotchi-2dba63df3bf1.c5v4v4jx6pq5.win", "https://ravenok-statosphere-backend.hf.space", "https://lord-raven.github.io"]:
            return "{}"
    data = json.loads(data_string)
    # if 'task' in data and data['task'] == 'few_shot_classification':
    #     return few_shot_classification(data)
    # else:
    return zero_shot_classification(data)

@spaces.GPU
def zero_shot_classification(data):
    results = classifier(data['sequence'], candidate_labels=data['candidate_labels'], hypothesis_template=data['hypothesis_template'], multi_label=data['multi_label'])
    response_string = json.dumps(results)
    return response_string

def create_sequences(data):
    # return ['###Given:\n' + data['sequence'] + '\n###End Given\n###Hypothesis:\n' + data['hypothesis_template'].format(label) + "\n###End Hypothesis" for label in data['candidate_labels']]
    return [data['sequence'] + '\n' + data['hypothesis_template'].format(label) for label in data['candidate_labels']]

# def few_shot_classification(data):
#     sequences = create_sequences(data)
#     print(sequences)
#     # results = onnx_few_shot_model(sequences)
#     probs = onnx_few_shot_model.predict_proba(sequences)
#     scores = [true[0] for true in probs]

#     composite = list(zip(scores, data['candidate_labels']))
#     composite = sorted(composite, key=lambda x: x[0], reverse=True)

#     labels, scores = zip(*composite)

#     response_dict = {'scores': scores, 'labels': labels}
#     print(response_dict)
#     response_string = json.dumps(response_dict)
#     return response_string

gradio_interface = gradio.Interface(
    fn = classify,
    inputs = gradio.Textbox(label="JSON Input"),
    outputs = gradio.Textbox()
)
gradio_interface.launch()