app.py

import torch
import json
import os
from transformers import T5Tokenizer, T5ForConditionalGeneration, Trainer, TrainingArguments, DataCollatorForSeq2Seq
from datasets import Dataset

# ✅ 1. Ielādē treniņa datus
with open("train.json", "r", encoding="utf-8") as f:
    train_data = json.load(f)

# ✅ 2. Pārveido datus Hugging Face dataset formātā
dataset = Dataset.from_list(
    [{"input_text": d["jautājums"], "target_text": d["atbilde"]} for d in train_data]
)

# ✅ 3. Sadala datus treniņam un testam (80% treniņam, 20% testam)
train_test_split = dataset.train_test_split(test_size=0.2)
train_dataset = train_test_split["train"]
eval_dataset = train_test_split["test"]

# ✅ 4. Izvēlas modeli (`mT5-small` vai citu)
model_name = "google/mt5-small"
device = "cuda" if torch.cuda.is_available() else "cpu"

tokenizer = T5Tokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name).to(device)

# ✅ 5. Sagatavo datus treniņam
def preprocess_data(examples):
    model_inputs = tokenizer(examples["input_text"], max_length=128, truncation=True)
    labels = tokenizer(examples["target_text"], max_length=128, truncation=True)
    model_inputs["labels"] = labels["input_ids"]
    return model_inputs

train_dataset = train_dataset.map(preprocess_data, batched=True)
eval_dataset = eval_dataset.map(preprocess_data, batched=True)

# ✅ 6. Definē treniņa parametrus
training_args = TrainingArguments(
    output_dir="./trained_model",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    logging_strategy="epoch",
    learning_rate=5e-5,
    per_device_train_batch_size=4,
    per_device_eval_batch_size=4,
    num_train_epochs=3,
    weight_decay=0.01,
    push_to_hub=False,  # ✅ Pārvērties True, ja vēlies augšupielādēt Hugging Face Hub
)

data_collator = DataCollatorForSeq2Seq(tokenizer, model=model)

# ✅ 7. Izveidot `Trainer` klasi
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    tokenizer=tokenizer,
    data_collator=data_collator,
)

# ✅ 8. Sāc AI apmācību
trainer.train()

# ✅ 9. Saglabā trenēto modeli
model.save_pretrained("./trained_model")
tokenizer.save_pretrained("./trained_model")

# ✅ 10. Novērtē AI precizitāti pēc apmācības
results = trainer.evaluate()
print("📊 Testēšanas rezultāti:", results)

# ✅ 11. Ja vēlies augšupielādēt AI uz Hugging Face Hub
# (Pirms tam jāpieslēdzas Hugging Face ar `notebook_login()`)
# trainer.push_to_hub("lietotajsvards/elektro-ai")