feat(docs): [SCv1] Automatically create and upload a custom HF model …

…to seldon-models in GCS on every version release (#5104)

* V2D-1258 Create and upload a custom HF model to seldon-models in GCS

* Remove left over link for kfserving storage initialiser

* Change GCS location to make it obvious it is a custom hf model

examples/upgrade-to-rclone/global-rclone-upgrade.ipynb

@@ -10,7 +10,7 @@
     "In this documentation page we provide an example upgrade path from kfserving-based to rclone-based storage initializer. This is required due to the fact that secret format expected by these two storage initializers is different. \n",
     "\n",
     "Storage initializers are used by Seldon's pre-packaged model servers to download models binaries. \n",
-    "As it is explained in the [SC 1.8 upgrading notes](https://docs.seldon.io/projects/seldon-core/en/latest/reference/upgrading.html#upgrading-to-1-8) the [seldonio/rclone-storage-initializer](https://github.com/SeldonIO/seldon-core/tree/master/components/rclone-storage-initializer) became default storage initializer in v1.8.0. However, it is still possible to run with kfserving-based Storage Initializer as documented [here](https://docs.seldon.io/projects/seldon-core/en/latest/servers/kfserving-storage-initializer.html).\n",
+    "As it is explained in the [SC 1.8 upgrading notes](https://docs.seldon.io/projects/seldon-core/en/latest/reference/upgrading.html#upgrading-to-1-8) the [seldonio/rclone-storage-initializer](https://github.com/SeldonIO/seldon-core/tree/master/components/rclone-storage-initializer) became default storage initializer in v1.8.0.\n",
     "\n",
     "In this tutorial we aim to provide an intuition of the steps you will have to carry to migrate to the new rclone-based Storage Initializer with the context that every cluster configuration will be different, so you should be able to see this as something you can build from.\n",
     "\n",

examples/upgrade-to-rclone/rclone-upgrade.ipynb

@@ -10,7 +10,7 @@
     "In this documentation page we provide an example upgrade path from kfserving-based to rclone-based storage initializer. This is required due to the fact that secret format expected by these two storage initializers is different. \n",
     "\n",
     "Storage initializers are used by Seldon's pre-packaged model servers to download models binaries. \n",
-    "As it is explained in the [SC 1.8 upgrading notes](https://docs.seldon.io/projects/seldon-core/en/latest/reference/upgrading.html#upgrading-to-1-8) the [seldonio/rclone-storage-initializer](https://github.com/SeldonIO/seldon-core/tree/master/components/rclone-storage-initializer) became default storage initializer in v1.8.0. However, it is still possible to run with kfserving-based Storage Initializer as documented [here](https://docs.seldon.io/projects/seldon-core/en/latest/servers/kfserving-storage-initializer.html).\n",
+    "As it is explained in the [SC 1.8 upgrading notes](https://docs.seldon.io/projects/seldon-core/en/latest/reference/upgrading.html#upgrading-to-1-8) the [seldonio/rclone-storage-initializer](https://github.com/SeldonIO/seldon-core/tree/master/components/rclone-storage-initializer) became default storage initializer in v1.8.0.\n",
     "\n",
     "In this tutorial we will show how to upgrade your configuration to new Storage Initializer with focus on getting the new format of a required secret right.\n",
     "\n",

servers/Makefile

@@ -1,4 +1,4 @@
-models: mlflowserver-models sklearnserver-models
+models: mlflowserver-models sklearnserver-models huggingface-models
 
 
 mlflowserver-models:
@@ -8,3 +8,7 @@ mlflowserver-models:
 sklearnserver-models:
 	make -C sklearnserver/models/iris
 	make -C sklearnserver/models/moviesentiment
+
+
+huggingface-models:
+	make -C huggingface/models/text-generation

servers/huggingface/models/text-generation/.gitignore

@@ -0,0 +1,2 @@
+.env/
+text-generation-model-artefacts/

servers/huggingface/models/text-generation/Makefile

@@ -0,0 +1,16 @@
+VERSION := $(shell cat ../../../../version.txt)
+
+all: env train upload
+
+model: env train
+
+env:
+	python3 -m venv .env
+	./.env/bin/pip install --upgrade pip setuptools
+	./.env/bin/pip install -r requirements.txt
+
+train:
+	.env/bin/python train.py
+
+upload:
+	gsutil cp -r text-generation-model-artefacts/* gs://seldon-models/v${VERSION}/huggingface/custom-text-generation/

servers/huggingface/models/text-generation/requirements.txt

@@ -0,0 +1,3 @@
+seldon_core
+transformers >=4.30,<4.32
+tensorflow == 2.12.0

servers/huggingface/models/text-generation/train.py

@@ -0,0 +1,19 @@
+from transformers import (
+    GPT2Tokenizer,
+    TFGPT2LMHeadModel,
+    pipeline,
+)
+
+
+def main() -> None:
+    tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
+    model = TFGPT2LMHeadModel.from_pretrained("gpt2")
+
+    p = pipeline(task="text-generation", model=model, tokenizer=tokenizer)
+
+    p.save_pretrained("text-generation-model-artefacts")
+
+
+if __name__ == "__main__":
+    print("Building a custom HuggingFace model...")
+    main()