> For the complete documentation index, see [llms.txt](https://langtech-bsc.gitbook.io/aina-kit/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://langtech-bsc.gitbook.io/aina-kit/testing/via-docker.md).

# Via Docker

També s’han preparat alguns models que requereixen poca capacitat de còmput en fer l’inferència com a “dockers” auto-continguts als quals es pot accedir mitjançant una RestFUL API.

* RestFUL API i interfície web per servir models de **síntesi de la parla**
  * Accés al Docker: <https://hub.docker.com/r/projecteaina/tts-api>
  * Accés a GitHub: <https://github.com/projecte-aina/tts-api>
* RestFUL API per servir models de **traducció automàtica**
  * Accés al Docker: <https://hub.docker.com/r/projecteaina/mt-api>
  * Accés a GitHub: <https://github.com/projecte-aina/mt-api>

Si teniu interès en que posem a disposició altres models en format "Docker", feu-nos arribar la [petició a través del formulari de contacte](https://form.typeform.com/to/KcjhThot?typeform-source=projecteaina.cat).

### Requisits mínims/recomanats per als models disponibles

<table data-view="cards"><thead><tr><th>Model</th><th>Type</th><th>MINIMUM GPU vRAM MEMORY</th><th>RECOMMENDED HARDWARE</th><th>Works on CPU?</th></tr></thead><tbody><tr><td><a href="https://huggingface.co/BSC-LT/salamandra-7b-instruct">BSC-LT/salamandra-7b-instruct</a></td><td>Text Generation</td><td>24 GB</td><td><p><strong>Minimum/Suggested:</strong></p><p>NVIDIA A10G 24GB<br>NVIDIA L4 24GB<br><br><strong>Next level:</strong><br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></p></td><td>No</td></tr><tr><td><a href="https://huggingface.co/BSC-LT/salamandra-7b">BSC-LT/salamandra-7b</a></td><td>Text Generation</td><td>24 GB</td><td><p><strong>Minimum/Suggested:</strong></p><p>NVIDIA A10G 24GB<br>NVIDIA L4 24GB<br><br><strong>Next level:</strong><br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></p></td><td>No</td></tr><tr><td><a href="https://huggingface.co/BSC-LT/salamandra-2b-instruct">BSC-LT/salamandra-2b-instruct</a></td><td>Text Generation</td><td>-</td><td><strong>Minimum:</strong><br>CPU with 4 cores and 8GB of RAM<br><br><strong>Suggested:</strong><br>NVIDIA T4 16GB<br><br><strong>Next level:</strong><br>NVIDIA A10G 24GB<br>NVIDIA L4 24GB<br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></td><td>Yes, but response times increase considerably.</td></tr><tr><td><a href="https://huggingface.co/BSC-LT/salamandra-2b">BSC-LT/salamandra-2b</a></td><td>Text Generation</td><td>-</td><td><strong>Minimum:</strong><br><br>CPU with 4 cores and 8GB of RAM<br><br><strong>Suggested:</strong><br>NVIDIA T4 16GB<br><br><strong>Next level:</strong><br><br>NVIDIA A10G 24GB<br>NVIDIA L4 24GB<br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></td><td>Yes, but response times increase considerably.</td></tr><tr><td><a href="https://huggingface.co/projecte-aina/whisper-large-v3-ca-3catparla">projecte-aina/whisper-large-v3-ca-3catparla</a></td><td>Speech Recognition</td><td>16GB</td><td><strong>Minimum/Suggested:</strong><br>NVIDIA T4 16GB<br>NVIDIA L4 24GB<br>NVIDIA A10G 24GB<br><br><strong>Next level:</strong><br><br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></td><td>No</td></tr><tr><td><a href="https://huggingface.co/projecte-aina/matxa-tts-cat-multiaccent">projecte-aina/matxa-tts-cat-multiaccent</a></td><td>Text to Speech</td><td>-</td><td><strong>Minimum:</strong><br><br>CPU with 4 cores and 8GB of RAM<br>NVIDIA GTX1060 (minimum tested by us, lower may work)<br><br><strong>Suggested:</strong><br>NVIDIA T4 16GB<br><br><strong>Next level:</strong><br><br>NVIDIA A10G 24GB<br>NVIDIA L4 24GB<br>NVIDIA T4 60GB (4 GPUs)<br>NVIDIA A10G 96GB (4 GPUs)<br>NVIDIA A100<br>NVIDIA H100<br></td><td>Yes</td></tr></tbody></table>