Add dedicated website translation model

README.md

@@ -652,8 +652,10 @@ curated source files, with the Nahuatl home page intentionally biased toward as
 many Nahuatl words as possible while keeping technical terms understandable.
 Unsupported browser languages use the same-origin `/translate.php` endpoint,
 which calls Ollama server-side through `OLLAMA_HOST` and `OLLAMA_MODEL`; the
-browser never calls the private Ollama IP directly. Generated runtime language
-JSON is saved through `save_lang.php` on the website PVC.
+browser never calls the private Ollama IP directly. The default model is the
+custom `website-translator` Ollama model defined in
+`apps/website/ollama/Modelfile`. Generated runtime language JSON is saved
+through `save_lang.php` on the website PVC.
 
 The CV page has two client-side presentation modes:
 

apps/website/Dockerfile

@@ -87,7 +87,7 @@ RUN rm -f /var/www/localhost/htdocs/index.html && \
 
 ENV WEBSITE_LANG_WRITE_DIR=/var/www/localhost/htdocs/db/lang
 ENV OLLAMA_HOST=http://192.168.100.68:11434
-ENV OLLAMA_MODEL=llama3.2:3b
+ENV OLLAMA_MODEL=website-translator
 
 USER apache
 

apps/website/ollama/Modelfile

@@ -0,0 +1,15 @@
+FROM llama3.2:3b
+
+PARAMETER temperature 0
+PARAMETER top_p 0.2
+PARAMETER num_ctx 8192
+
+SYSTEM """
+You are a translation engine only.
+
+Translate every input string into the requested target language.
+Preserve the input order exactly.
+Preserve URLs, email addresses, product names, code identifiers, version numbers, and HTML entities.
+Return only a valid JSON array of translated strings.
+Do not return markdown, comments, explanations, or surrounding text.
+"""

apps/website/ollama/README.md

@@ -0,0 +1,13 @@
+# Website Translation Model
+
+This directory defines the Ollama model used by the website translation endpoint.
+
+Create or update the model on the Ollama host:
+
+```sh
+ollama create website-translator -f apps/website/ollama/Modelfile
+```
+
+The website uses `OLLAMA_MODEL=website-translator`. Recreate the model after
+changing the `Modelfile`, then restart the website pods so the cache key includes
+the active model name.

apps/website/web-app.yaml

@@ -71,7 +71,7 @@ spec:
             - name: OLLAMA_HOST
               value: http://192.168.100.68:11434
             - name: OLLAMA_MODEL
-              value: llama3.2:3b
+              value: website-translator
             - name: TRANSLATION_REDIS_HOST
               value: website-redis
             - name: TRANSLATION_REDIS_PORT