|
--- |
|
license: apache-2.0 |
|
--- |
|
|
|
Base Model: `google/t5-small` |
|
|
|
A seq2seq event triggers and entities tagger trained on the dataset: `ahmeshaf/ecb_plus_ed` |
|
|
|
## Usage |
|
|
|
Input: |
|
```shell |
|
triggers: I like this model and hate this sentence |
|
``` |
|
|
|
Output: |
|
```shell |
|
like | hate |
|
``` |
|
|
|
- Python |
|
|
|
### Using .generate() |
|
```python |
|
from transformers import GenerationConfig, T5ForConditionalGeneration, T5Tokenizer |
|
|
|
model_name = "ahmeshaf/ecb_tagger_seq2seq" |
|
|
|
model = T5ForConditionalGeneration.from_pretrained(model_name) |
|
tokenizer = T5Tokenizer.from_pretrained(model_name) |
|
generation_config = GenerationConfig.from_pretrained(model_name) |
|
|
|
tokenized_inputs = tokenizer(["I like this model and hate this sentence ."], return_tensors="pt") |
|
outputs = model.generate(**tokenized_inputs, generation_config=generation_config) |
|
|
|
print(tokenizer.batch_decode(outputs, skip_special_tokens=True)) |
|
|
|
# ['like | hate'] |
|
``` |
|
|
|
### Using pipeline |
|
|
|
```python |
|
from transformers import pipeline |
|
srl = pipeline("text2text-generation", "ahmeshaf/ecb_tagger_seq2seq") |
|
print(srl(["I like this model and hate this sentence ."])) |
|
|
|
# [{'generated_text': 'like | hate'}] |
|
``` |
|
|
