You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: articles/ai-services/translator/custom-translator/azure-ai-foundry/concepts/adaptive-custom-translation.md
+9-11Lines changed: 9 additions & 11 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -22,7 +22,7 @@ Adaptive custom translation (adaptCT) is a translation enhancement feature, desi
22
22
23
23
AdaptCT APIs for dynamic domain adaptations allow you to upload between 5 and 30,000 prealigned bilingual segments. With this capability, you can quickly build a custom language pair dataset index that's ready in minutes. This index can then be used with [Azure AI Translator 2025-05-01-preview APIs](/azure/ai-services/translator/text-translation/preview/overview). Unlike traditional custom models that require large training datasets and separate deployment, AdaptCT uses few-shot learning to dynamically select relevant sentence pairs from the index at runtime. This capability enables on-the-fly adaptation of the LLM's output to match domain-specific terminology, context, and style. `Available via Azure AI Foundry.`
24
24
25
-
Key Differences
25
+
### Key differences
26
26
***Custom translator**: Fine-tunes a dedicated translation model using your dataset; model is trained and deployed within ~48 hours.
27
27
***Adaptive custom translation**: No fine-tuning or deployment required; updates by rebuilding the dataset index, ready within minutes.
28
28
@@ -33,7 +33,7 @@ Key Differences
33
33
> - Make sure to test thoroughly before using in production environments.
34
34
> - Project and workspace are used interchangeably to mean a Foundry created project.
0 commit comments