Open-weight model
mbart-large-50-one-to-many-mmt
by AI at Meta facebook/mbart-large-50-one-to-many-mmt
This model is a fine-tuned checkpoint of mBART-large-50. mbart-large-50-one-to-many-mmt is fine-tuned for multilingual machine translation. It was introduced in Multilingual Translation with Extensible Multilingual Pretraining and Finetuning paper.
Model Card
This model is a fine-tuned checkpoint of mBART-large-50. mbart-large-50-one-to-many-mmt is fine-tuned for multilingual machine translation. It was introduced in Multilingual Translation with Extensible Multilingual Pretraining and Finetuning paper. The model can translate English to other 49 languages mentioned below. To translate into a target language, the target language id is forced as the first generated token. To force the target language id as the first generated token, pass the forcedbostokenid parameter to the generate method. See the model hub to look for more fine-tuned versions. Arabic (arAR), Czech (csCZ), German (deDE), English (enXX), Spanish (esXX), Estonian (etEE), Finnish…
Excerpt from the card by AI at Meta.
Configuration
- Architecture
- MBartForConditionalGeneration
- Context length (tokens)
- 1,024
- Layers
- 12
- Vocabulary size
- 250,054
- Model type
- mbart
Identity and Version
- Repository
- facebook/mbart-large-50-one-to-many-mmt
- Publisher
- AI at Meta
- Task
- Not stated by the source
- Modality
- Other
- Library
- transformers
- Parameters
- Not stated by the source
- Languages
- ar, cs, de, en, es, et, fi, fr
- Revision
- 9dde161a836123acc97aa194fe507d3ac1e2a46e
- First published
- 2022-03-02
- Last updated
- 2023-03-28
Files and Weights
10 files, 7.3 GB in total. The weights are 3 files totalling 7.3 GB in bin, h5, msgpack.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| flax_model.msgpack | Weights | 2.4 GB | 0b7365f3d85f |
| pytorch_model.bin | Weights | 2.4 GB | 51fbf92efa74 |
| tf_model.h5 | Weights | 2.4 GB | 006794f8dab3 |
| config.json | Configuration | 1.4 KB | — |
| generation_config.json | Configuration | 261 B | — |
| special_tokens_map.json | Configuration | 717 B | — |
| README.md | Documentation | 3.6 KB | — |
| sentencepiece.bpe.model | Other | 5.1 MB | cfc8146abe2a |
| .gitattributes | Repository | 690 B | — |
| tokenizer_config.json | Tokenizer | 528 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 7.3 GB
Released by AI at Meta through its official repository on Hugging Face.
Built From
- Described by arXiv:2008.00401
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 7.3 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About mbart-large-50-one-to-many-mmt
What is mbart-large-50-one-to-many-mmt's context length?
1,024 tokens, from the maximum position embeddings in its published configuration.