Rut5 Base Sum Gazeta

🚀 RuT5SumGazeta

This is a model for abstractive summarization of Russian texts, built upon rut5-base. It offers an effective solution for summarizing Russian documents, helping users quickly obtain key information from large - scale text data.

🚀 Quick Start

Colab Link

You can quickly start using this model through Colab: link

Code Example

from transformers import AutoTokenizer, T5ForConditionalGeneration

model_name = "IlyaGusev/rut5_base_sum_gazeta"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)

article_text = "..."

input_ids = tokenizer(
    [article_text],
    max_length=600,
    add_special_tokens=True,
    padding="max_length",
    truncation=True,
    return_tensors="pt"
)["input_ids"]

output_ids = model.generate(
    input_ids=input_ids,
    no_repeat_ngram_size=4
)[0]

summary = tokenizer.decode(output_ids, skip_special_tokens=True)
print(summary)

✨ Features

• Based on rut5 - base: Leveraging the powerful features of [rut5 - base](https://huggingface.co/cointegrated/rut5 - base), it has good performance in Russian text processing.
• Abstractive Summarization: Capable of generating abstractive summaries, extracting key information from the text and presenting it in a concise way.

📦 Installation

The installation steps are mainly reflected in the code example above. You need to install relevant Python libraries such as transformers and datasets. You can use the following command to install:

pip install transformers datasets

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, T5ForConditionalGeneration

model_name = "IlyaGusev/rut5_base_sum_gazeta"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)

article_text = "..."

input_ids = tokenizer(
    [article_text],
    max_length=600,
    add_special_tokens=True,
    padding="max_length",
    truncation=True,
    return_tensors="pt"
)["input_ids"]

output_ids = model.generate(
    input_ids=input_ids,
    no_repeat_ngram_size=4
)[0]

summary = tokenizer.decode(output_ids, skip_special_tokens=True)
print(summary)

Advanced Usage

import json
import torch
from transformers import AutoTokenizer, T5ForConditionalGeneration
from datasets import load_dataset


def gen_batch(inputs, batch_size):
    batch_start = 0
    while batch_start < len(inputs):
        yield inputs[batch_start: batch_start + batch_size]
        batch_start += batch_size


def predict(
    model_name,
    input_records,
    output_file,
    max_source_tokens_count=600,
    batch_size=8
):
    device = "cuda" if torch.cuda.is_available() else "cpu"

    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = T5ForConditionalGeneration.from_pretrained(model_name).to(device)
    
    predictions = []
    for batch in gen_batch(input_records, batch_size):
        texts = [r["text"] for r in batch]
        input_ids = tokenizer(
            texts,                                                                                                     
            add_special_tokens=True,
            max_length=max_source_tokens_count,
            padding="max_length",
            truncation=True,
            return_tensors="pt"
        )["input_ids"].to(device)
        
        output_ids = model.generate(
            input_ids=input_ids,
            no_repeat_ngram_size=4
        )
        summaries = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
        for s in summaries:
            print(s)
        predictions.extend(summaries)
    with open(output_file, "w") as w:
        for p in predictions:
            w.write(p.strip().replace("\n", " ") + "\n")

gazeta_test = load_dataset('IlyaGusev/gazeta', script_version="v1.0")["test"]
predict("IlyaGusev/rut5_base_sum_gazeta", list(gazeta_test), "t5_predictions.txt")

📚 Documentation

Training Data

• Dataset: Gazeta

Training Procedure

• Training script: train.py
• Config: t5_training_config.json

Eval Results

Gazeta v1

Gazeta v2

Evaluation Script

• Script: evaluate.py
• Flags: --language ru --tokenize - after --lower

📄 License

This project is licensed under the [apache - 2.0](https://www.apache.org/licenses/LICENSE - 2.0) license.