File size: 2,214 Bytes
249be7e c16db27 249be7e c16db27 249be7e |
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 57 58 59 60 61 62 |
---
language:
- cs
- cs
tags:
- abstractive summarization
- mbart-cc25
- Czech
license: apache-2.0
datasets:
- SumeCzech dataset news-based
metrics:
- rouge
- rougeraw
---
# mBART fine-tuned model for Czech abstractive summarization (HT2A-S)
This model is a fine-tuned checkpoint of [facebook/mbart-large-cc25](https://huggingface.co/facebook/mbart-large-cc25) on the Czech news dataset to produce Czech abstractive summaries.
## Task
The model deals with the task ``Headline + Text to Abstract`` (HT2A) which consists in generating a multi-sentence summary considered as an abstract from a Czech news text.
## Dataset
The model has been trained on the [SumeCzech](https://ufal.mff.cuni.cz/sumeczech) dataset. The dataset includes around 1M Czech news-based documents consisting of a Headline, Abstract, and Full-text sections. Truncation and padding were configured for 512 tokens for the encoder and 128 for the decoder.
## Training
The model has been trained on 1x NVIDIA Tesla A100 40GB for 20 hours, 1x NVIDIA Tesla V100 32GB for 40 hours, and 4x NVIDIA Tesla A100 40GB for 20 hours. During training, the model has seen 6928K documents corresponding to roughly 8 epochs.
# Use
Assuming you are using the provided Summarizer.ipynb file.
```python
def summ_config():
cfg = OrderedDict([
# summarization model - checkpoint from website
("model_name", "krotima1/mbart-ht2a-s"),
("inference_cfg", OrderedDict([
("num_beams", 4),
("top_k", 40),
("top_p", 0.92),
("do_sample", True),
("temperature", 0.89),
("repetition_penalty", 1.2),
("no_repeat_ngram_size", None),
("early_stopping", True),
("max_length", 128),
("min_length", 10),
])),
#texts to summarize
("text",
[
"Input your Czech text",
]
),
])
return cfg
cfg = summ_config()
#load model
model = AutoModelForSeq2SeqLM.from_pretrained(cfg["model_name"])
tokenizer = AutoTokenizer.from_pretrained(cfg["model_name"])
# init summarizer
summarize = Summarizer(model, tokenizer, cfg["inference_cfg"])
summarize(cfg["text"])
``` |