Add a single-model ollama image.

Yes, it's yet another huge image (26GB) :)

This image is useful to benchmark checkpoint/restore performance.
To fairly measure it, we need an image that has only one model
preloaded into it, as this mirrors the situation where someone might
want to use checkpoint/restore for an inference model where there is
only one model to serve.

This image uses mixtral as a medium-large model. The larger the model,
the more important the gains in checkpoint/restore performance, so
measuring with a large model provides a more representative and
significant result.

PiperOrigin-RevId: 613335037

images/gpu/ollama/singlemodel/Dockerfile

@@ -0,0 +1,15 @@
+# https://hub.docker.com/r/ollama/ollama
+FROM ollama/ollama:0.1.26
+
+ENV PATH=$PATH:/usr/local/nvidia/bin:/bin/nvidia/bin
+ENV OLLAMA_ORIGINS=*
+ENV OLLAMA_HOST=0.0.0.0:11434
+
+# Pre-install a single model.
+RUN bash -c '                                  \
+    ( ollama serve ) & serverpid="$!";         \
+    sleep 5;                                   \
+    ollama pull mixtral:instruct &&            \
+    kill "$serverpid" &&                       \
+    wait "$serverpid"                          \
+'