mirror of
https://github.com/GoogleCloudPlatform/vertex-ai-samples.git
synced 2026-09-26 14:42:04 +00:00
feat: Getting tuned embeddings using text-embedding-005. (#3673)
* feat: Getting tuned embeddings using text-embedding-005. * feat: Getting tuned embeddings using text-embedding-005. * feat: Getting tuned embeddings using text-embedding-005.
This commit is contained in:
lee1premium
@@ -72,7 +72,7 @@
|
||||
"source": [
|
||||
"### Objective\n",
|
||||
"\n",
|
||||
"In this tutorial, you learn how to tune a text-embedding model, textembedding-gecko.\n",
|
||||
"In this tutorial, you learn how to tune a text-embedding model.\n",
|
||||
"\n",
|
||||
"This tutorial uses the following Google Cloud ML services and resources:\n",
|
||||
"\n",
|
||||
@@ -131,18 +131,18 @@
|
||||
"source": [
|
||||
"## Installation\n",
|
||||
"\n",
|
||||
"This tutorial requires you to install the `google-cloud-aiplatform` package."
|
||||
"This tutorial requires you to install the latest `google-cloud-aiplatform` package release."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"id": "ARIslNGX4IR3"
|
||||
"id": "Nn6Ze-c4IZyt"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"! pip3 install --upgrade --user --quiet google-cloud-aiplatform"
|
||||
"!pip3 install --quiet --upgrade google-cloud-aiplatform"
|
||||
]
|
||||
},
|
||||
{
|
||||
@@ -210,9 +210,9 @@
|
||||
"id": "0SG0ZnxmM1Ev"
|
||||
},
|
||||
"source": [
|
||||
"### Initialize Vertex AI Platform\n",
|
||||
"### Initialize Vertex AI SDK\n",
|
||||
"\n",
|
||||
"Import and Initialize AI platform for your project and region.\n",
|
||||
"Import and Initialize the SDK for your project and region.\n",
|
||||
"\n",
|
||||
"**If you don't know your project ID**, try the following:\n",
|
||||
"* Run `gcloud config list`.\n",
|
||||
@@ -226,7 +226,6 @@
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"cellView": "form",
|
||||
"id": "SWbf4r0tLUvA"
|
||||
},
|
||||
"outputs": [],
|
||||
@@ -328,7 +327,7 @@
|
||||
"\n",
|
||||
"(Optionally), to resume this tutorial from where you left off with your previous tuning session, set **`TUNING_JOB_ID`** accordingly. Or, clear **`TUNING_JOB_ID`** to start over a fresh tuning session.\n",
|
||||
"\n",
|
||||
"This tutorial creates a tuning job of a Vertex AI pipeline for tuning a text-embedding model within your project. See also [creating tuning jobs with parameters and defaults](https://cloud.google.com/vertex-ai/generative-ai/docs/models/tune-embeddings#create-embedding-tuning-job), the latest [text-embedding models and eligible tasks](https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#api_changes_to_models_released_on_or_after_august_2023)."
|
||||
"This tutorial creates a tuning job of a Vertex AI pipeline for tuning a text-embedding model within your project. See also [creating tuning jobs with parameters and defaults](https://cloud.google.com/vertex-ai/generative-ai/docs/models/tune-embeddings#create-embedding-tuning-job), and the latest [text-embedding models and eligible tasks](https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#api_changes_to_models_released_on_or_after_august_2023)."
|
||||
]
|
||||
},
|
||||
{
|
||||
@@ -348,7 +347,6 @@
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"cellView": "form",
|
||||
"id": "G1i__mAHNY9U"
|
||||
},
|
||||
"outputs": [],
|
||||
@@ -377,10 +375,10 @@
|
||||
"batch_size = 128 # @param {type: \"number\"}\n",
|
||||
"# @markdown ---\n",
|
||||
"# @markdown `base_model_version_id`: Optional. Use this to specify what text embedding model to tune. This must be a stable version, for example `text-embedding-004`. To learn more, see [Supported Models](https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#supported-models).\n",
|
||||
"base_model_version_id = \"text-embedding-004\" # @param [\"textembedding-gecko@003\", \"text-embedding-004\", \"textembedding-gecko-multilingual@001\", \"text-multilingual-embedding-002\"]\n",
|
||||
"base_model_version_id = \"text-embedding-005\" # @param [\"text-embedding-005\", \"text-embedding-004\", \"text-multilingual-embedding-002\"]\n",
|
||||
"# @markdown ---\n",
|
||||
"# @markdown `task_type`: Optional. The task type expected to be used during inference. See the [embedding API reference](https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text-embeddings-api#request_body) to learn more.\n",
|
||||
"task_type = \"DEFAULT\" # @param [\"DEFAULT\", \"RETRIEVAL_QUERY\", \"RETRIEVAL_DOCUMENT\", \"SEMANTIC_SIMILARITY\", \"CLASSIFICATION\", \"CLUSTERING\", \"QUESTION_ANSWERING\", \"FACT_VERIFICATION\"]\n",
|
||||
"task_type = \"DEFAULT\" # @param [\"DEFAULT\", \"RETRIEVAL_QUERY\", \"RETRIEVAL_DOCUMENT\", \"SEMANTIC_SIMILARITY\", \"CLASSIFICATION\", \"CLUSTERING\", \"QUESTION_ANSWERING\", \"FACT_VERIFICATION\", \"CODE_RETRIEVAL_QUERY\"]\n",
|
||||
"# @markdown ---\n",
|
||||
"# @markdown `learning_rate_multiplier`: Optional. A multiplier to apply to the recommended learning rate. To train with the recommended learning rate, use 1.0.\n",
|
||||
"learning_rate_multiplier = 1.0 # @param {type: \"number\"}\n",
|
||||
@@ -389,13 +387,26 @@
|
||||
"output_dimensionality = 768 # @param {type: \"number\"}\n",
|
||||
"\n",
|
||||
"# Synchronously validate some edge cases that will cause async validation to fail.\n",
|
||||
"if base_model_version_id not in [\"text-embedding-004\", \"text-multilingual-embedding-002\"]:\n",
|
||||
"if base_model_version_id not in [\"text-embedding-005\"]:\n",
|
||||
" if task_type in [\"CODE_RETRIEVAL_QUERY\"]:\n",
|
||||
" raise ValueError(\n",
|
||||
" f\"task_type '{task_type}' is not valid for model '{base_model_version_id}'.\"\n",
|
||||
" )\n",
|
||||
"if base_model_version_id not in [\n",
|
||||
" \"text-embedding-005\",\n",
|
||||
" \"text-embedding-004\",\n",
|
||||
" \"text-multilingual-embedding-002\",\n",
|
||||
"]:\n",
|
||||
" if task_type in [\"QUESTION_ANSWERING\", \"FACT_VERIFICATION\"]:\n",
|
||||
" raise ValueError(f\"task_type '{task_type}' is not valid for model '{base_model_version_id}'.\")\n",
|
||||
"\n",
|
||||
" raise ValueError(\n",
|
||||
" f\"task_type '{task_type}' is not valid for model '{base_model_version_id}'.\"\n",
|
||||
" )\n",
|
||||
" if output_dimensionality not in [-1, 768]:\n",
|
||||
" raise ValueError(f\"Model '{base_model_version_id}' does not support the output_dimensionality parameter.\")\n",
|
||||
" raise ValueError(\n",
|
||||
" f\"Model '{base_model_version_id}' does not support the output_dimensionality parameter.\"\n",
|
||||
" )\n",
|
||||
"\n",
|
||||
"running_interactively = not os.getenv(\"IS_TESTING\")\n",
|
||||
"base_model = TextEmbeddingModel.from_pretrained(base_model_version_id)\n",
|
||||
"if \"TUNING_JOB_ID\" in locals() and TUNING_JOB_ID:\n",
|
||||
" filter = f'pipelineJobUserId=\"{TUNING_JOB_ID}\"'\n",
|
||||
@@ -403,7 +414,7 @@
|
||||
" print(\n",
|
||||
" f\"Got an existing tuning job '{tuning_job.name}' (state: {tuning_job.state.name}).\"\n",
|
||||
" )\n",
|
||||
"else:\n",
|
||||
"elif running_interactively:\n",
|
||||
" tuning_result = base_model.tune_model(\n",
|
||||
" task_type=task_type,\n",
|
||||
" corpus_data=corpus_path,\n",
|
||||
@@ -457,7 +468,6 @@
|
||||
"ACCELERATOR_COUNT = 1 # @param {type: \"number\"}\n",
|
||||
"\n",
|
||||
"# CAVEAT: Colab disruptions may cause 'tuning_result' to be undefined.\n",
|
||||
"running_interactively = not os.getenv(\"IS_TESTING\")\n",
|
||||
"if \"tuning_job\" not in locals() and running_interactively:\n",
|
||||
" message = \"[Action Required] Run the preceding code cells to define 'tuning_job'.\"\n",
|
||||
" raise RuntimeError(message)\n",
|
||||
|
||||
Reference in New Issue
Block a user