입력:  "Thank you <X> me to your party <Y> week"
타겟:  "<X> for inviting <Y> last <Z>"

# T5의 텍스트-투-텍스트 입력 형식

tasks = {
    "번역":     "translate English to German: That is good.",
    "요약":     "summarize: State authorities dispatched teams Tuesday...",
    "분류":     "sst2 sentence: This movie was absolutely wonderful.",
    "유사도":   "stsb sentence1: The cat sat on the mat. sentence2: A cat is on the mat.",
    "QA":       "question: What is gravity? context: Gravity is a force...",
}

# 모든 태스크가 동일한 모델, 동일한 인터페이스
for task, input_text in tasks.items():
    output = model.generate(tokenizer(input_text, return_tensors="pt").input_ids)
    print(f"{task}: {tokenizer.decode(output[0], skip_special_tokens=True)}")

from transformers import MBartForConditionalGeneration, MBart50TokenizerFast

model = MBartForConditionalGeneration.from_pretrained(
    "facebook/mbart-large-50-many-to-many-mmt"
)
tokenizer = MBart50TokenizerFast.from_pretrained(
    "facebook/mbart-large-50-many-to-many-mmt"
)

# 한국어 → 영어 번역
tokenizer.src_lang = "ko_KR"
text = "인공지능은 인간의 학습, 추론, 지각 능력을 컴퓨터로 구현하는 기술입니다."
inputs = tokenizer(text, return_tensors="pt")

generated = model.generate(
    **inputs,
    forced_bos_token_id=tokenizer.lang_code_to_id["en_XX"],
    max_new_tokens=128,
)

translation = tokenizer.decode(generated[0], skip_special_tokens=True)
print(f"원문: {text}")
print(f"번역: {translation}")

from transformers import T5ForConditionalGeneration, T5Tokenizer

# T5-Base 모델 로드
model = T5ForConditionalGeneration.from_pretrained("t5-base")
tokenizer = T5Tokenizer.from_pretrained("t5-base")

# 요약할 텍스트 (프리픽스 "summarize:" 추가)
article = """
summarize: The tower is 324 metres (1,063 ft) tall, about the same height as an 81-storey building,
and the tallest structure in Paris. Its base is square, measuring 125 metres (410 ft) on each side.
During its construction, the Eiffel Tower surpassed the Washington Monument to become the tallest
man-made structure in the world, a title it held for 41 years until the Chrysler Building in New York
City was finished in 1930. It was the first structure to reach a height of 300 metres.
"""

inputs = tokenizer(article, return_tensors="pt", max_length=512, truncation=True)

# 요약 생성
summary_ids = model.generate(
    inputs.input_ids,
    max_new_tokens=100,
    num_beams=4,           # 빔 서치
    length_penalty=2.0,    # 길이 패널티
    early_stopping=True,
)

summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
print(f"요약: {summary}")

from transformers import BartForConditionalGeneration, BartTokenizer

# BART-Large CNN 모델 (요약 Fine-tuned)
model = BartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
tokenizer = BartTokenizer.from_pretrained("facebook/bart-large-cnn")

# 뉴스 기사 요약
article = """
New York (CNN) -- More than 80 million Americans are under a winter storm warning as a major
storm system moves across the eastern United States. The storm is expected to bring heavy snow,
ice, and freezing rain to a wide area from the Midwest to the Northeast. Schools, government
offices, and businesses have closed in several states. Airlines have cancelled thousands of
flights. The National Weather Service warned of dangerous travel conditions and potential power
outages.
"""

inputs = tokenizer(article, return_tensors="pt", max_length=1024, truncation=True)

summary_ids = model.generate(
    inputs.input_ids,
    max_new_tokens=150,
    num_beams=4,
    length_penalty=2.0,
)

print(tokenizer.decode(summary_ids[0], skip_special_tokens=True))

from transformers import T5ForConditionalGeneration, T5Tokenizer

model = T5ForConditionalGeneration.from_pretrained("t5-base")
tokenizer = T5Tokenizer.from_pretrained("t5-base")

def text_to_text(input_text):
    """T5의 텍스트-투-텍스트 인터페이스로 다양한 태스크를 수행합니다."""
    inputs = tokenizer(input_text, return_tensors="pt", max_length=512, truncation=True)
    outputs = model.generate(inputs.input_ids, max_new_tokens=100)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# 다양한 태스크를 동일한 인터페이스로 수행
tasks = [
    "translate English to German: How are you?",
    "summarize: The quick brown fox jumps over the lazy dog. This is a famous pangram.",
    "sst2 sentence: This movie was really enjoyable and fun to watch.",
    "cola sentence: The cat sat on the mat.",
]

for task in tasks:
    result = text_to_text(task)
    print(f"입력: {task[:60]}...")
    print(f"출력: {result}\n")