Flan-T5 Base — ONNX
ONNX export of Flan-T5 Base (250M parameters) with encoder-decoder architecture and KV cache support.
Converted for use with inference4j, an inference-only AI library for Java.
Original Source
- Repository: google/flan-t5-base
- License: Apache 2.0
Usage with inference4j
// Summarization
try (var gen = FlanT5TextGenerator.flanT5Base().build()) {
System.out.println(gen.summarize("Long article text..."));
}
// Translation
try (var gen = FlanT5TextGenerator.flanT5Base().build()) {
System.out.println(gen.translate("Hello!", Language.EN, Language.FR));
}
Model Details
| Property | Value |
|---|---|
| Architecture | T5 encoder-decoder (250M parameters) |
| Tasks | Summarization, translation, grammar correction, text-to-SQL |
| Tokenizer | SentencePiece (32,128 tokens) |
| Original framework | PyTorch (transformers) |
| Export method | Hugging Face Optimum (encoder-decoder with KV cache) |
License
This model is licensed under the Apache License 2.0. Original model by Google.