SalamandraTA 7B v3
SalamandraTA 7B v3 is BSC's expanded translation model with terminology and structured-document support.
Model and training
The current card covers 40 languages and three varieties. It adds terminology-aware translation, HTML and XML handling, post-editing and named-entity tasks. BSC states that training-data conditions require distribution under GPL 3.0. This is a material license change from the Apache-licensed v1 and v2 branches. The official GGUF package belongs to this release.SalamandraTA 7B v3: official model card
Practical use
Practical guidance. Evaluate tag preservation and supplied terminology separately from translation quality. Keep markup validation in the application, since a language model can still produce malformed output. Review the v3 license for the intended distribution and service setup; a previously approved v2 workflow does not automatically establish the same terms for v3.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | salamandraTA-7b-instructsalamandraTA-7B-instruct-GGUF |
| Context window | 8,192 tokens Maximum sequence length documented for these weights; the prompt and generated text share the window. SalamandraTA 7B v3: official model card |
| License | GNU General Public License 3.0; earlier versioned branches have different terms.SalamandraTA 7B v3: official model card |
| Model size | 7,768,117,248 parameters.SalamandraTA 7B v3: official model card |
Related models
- SalamandraTA 2B
Barcelona Supercomputing Center · SalamandraTA
SalamandraTA 2B is a small translation model derived from BSC's Salamandra 2B base.
Launch date not confirmedDownloadable - SalamandraTA 7B v2
Barcelona Supercomputing Center · SalamandraTA
SalamandraTA 7B v2 expands BSC's translation model beyond its original European-language coverage.
Launch date not confirmedDownloadable