Jan 14, 2025 · Jan 14, 2025
diff --git a/pgml-cms/docs/open-source/korvus/guides/constructing-pipelines.md b/pgml-cms/docs/open-source/korvus/guides/constructing-pipelines.md
 The `Pipeline` above generates embeddings and tsvectors for the `abstract` and splits and generates embeddings and tsvectors for the `text`.

 We can now perform search over both the `text` and `abstract` key of our documents. See the [guide for vector search](vector-search) for more information on how to do this.

 ## Self-Hosting Specific Parameters

 **This section is only relevant for self hosted instances of PostgresML**. These parameters are never required for instances hosted by PostgresML.

 ### Trust Remote Code

 Some HuggingFace models require the argument `trust_remote_code=true`. To enable this, pass it as a parameter in the pipeline construction:

 ```python
 pipeline = Pipeline(
    "v0",
    {
        "text": {
            "semantic_search": {
                "model": "Alibaba-NLP/gte-base-en-v1.5",
                "parameters": {
                    "trust_remote_code": True
                }
            }
        }
    }
 )
 ```

 ### HuggingFace authentication

 Pass your HuggingFace token into the pipeline to access gated repos:

 ```python
 pipeline = Pipeline(
    "v0",
    {
        "text": {
            "semantic_search": {
                "model": "Alibaba-NLP/gte-base-en-v1.5",
                "parameters": {
                    "trust_remote_code": True,
                    "token": "YOUR_TOKEN"
                }
            }
        }
    }
 )
 ```
Original file line number	Diff line number	Diff line change
Expand Up		@@ -162,3 +162,48 @@ pipeline = Pipeline(
		The `Pipeline` above generates embeddings and tsvectors for the `abstract` and splits and generates embeddings and tsvectors for the `text`.

		We can now perform search over both the `text` and `abstract` key of our documents. See the [guide for vector search](vector-search) for more information on how to do this.

		## Self-Hosting Specific Parameters

		This section is only relevant for self hosted instances of PostgresML. These parameters are never required for instances hosted by PostgresML.

		### Trust Remote Code

		Some HuggingFace models require the argument `trust_remote_code=true`. To enable this, pass it as a parameter in the pipeline construction:

		```python
		pipeline = Pipeline(
		"v0",
		{
		"text": {
		"semantic_search": {
		"model": "Alibaba-NLP/gte-base-en-v1.5",
		"parameters": {
		"trust_remote_code": True
		}
		}
		}
		}
		)
		```

		### HuggingFace authentication

		Pass your HuggingFace token into the pipeline to access gated repos:

		```python
		pipeline = Pipeline(
		"v0",
		{
		"text": {
		"semantic_search": {
		"model": "Alibaba-NLP/gte-base-en-v1.5",
		"parameters": {
		"trust_remote_code": True,
		"token": "YOUR_TOKEN"
		}
		}
		}
		}
		)
		```