madlad400-3b-mt-onnx
Overview
This repository contains quantized ONNX encoder and decoder files for MADLAD-400-3B-MT, a multilingual machine translation model based on the T5 architecture.
Source
| Field | Value |
|---|
| Upstream model | google/madlad400-3b-mt |
| ONNX artifact source | ISoloist1/madlad400-3b-mt-onnx |
| Upstream source revision | fa184c675da0b5c9e1c8694fccd4e12e2d422094 |
| ONNX artifact source revision | 0dfc0b48e063b55508704ff75e21416c8679f663 |
| Packaging source revision | 67037ad42f58d6c0fc3dafaa45f3ec97a46e7eb9 |
| Export tool/script | ONNX Runtime quantization pipeline (mirrored from ISoloist1 ONNX bundle) |
| Quantization recipe | ONNX Runtime INT8 encoder/decoder graphs |
Precision and Packaging
Export tooling, precision, and quantization are recorded in the Source table above. This packaging mirror does not publish independent parity benchmarks; validate on your target execution provider before production use.
Files
| File | Description |
|---|
encoder_model_quantized.onnx | Quantized ONNX encoder model |
decoder_model_quantized.onnx | Quantized ONNX decoder model |
config.json | T5 model configuration |
tokenizer_config.json | T5 tokenizer configuration |
spiece.model | SentencePiece tokenizer model |
Intended Use
Use these ONNX model files for multilingual machine translation experiments or integration work.
Training Data
The upstream model card lists allenai/MADLAD-400 and describes training on large-scale public multilingual data, with additional parallel data for the machine translation model.
Evaluation
The upstream model card reports evaluation in the research paper using WMT, NTREX, FLORES-200, and Gatones. This repository does not document ONNX-specific evaluation or parity results.
Limitations
- Translation quality varies by language and domain; validate on your target language pairs and content.
License
Apache-2.0, matching the upstream source model.
