content-semantic-chunking-semanticsplitternodeparser-s-build-semantic-nodes--993e33e4

THEN Ensure you are using a version of llama_index where node.text is populated (fix merged after 0.10.11). If using an older version, manually populate node.text from the original splits. Note that node.embedding is not set at parse time and will be populated later during index creation.

from llama_index.core.node_parser import SemanticSplitterNodeParser
from llama_index.core import Document

embed_model = ...
chunker = SemanticSplitterNodeParser(embed_model=embed_model)
nodes = chunker.build_semantic_nodes_from_documents([Document(text=text)])
# Ensure nodes have text; if not, update library or populate manually:
for node in nodes:
    if not node.text:
        node.text = node.get_content()