docs/how_to/semantic-chunker/

[giscus[bot]]

docs/how_to/semantic-chunker/

Taken from Greg Kamradt's wonderful notebook:

https://python.langchain.com/docs/how_to/semantic-chunker/

[engmargo]

Heflin-Torbert Judicial Building, commonly called the Alabama Judicial Building, is a state government building in Montgomery, Alabama. It houses several state judicial agencies, most notably the Supreme Court of Alabama, Alabama Court of Civil Appeals, and Alabama Court of Criminal Appeals. It is the first state court building in the United States to house all three courts under one roof. Additionally, it houses the State Law Library.\nThe neoclassical-style structure was completed in 1994 at a cost of approximately $35 million.\nIn 2001, Roy Moore, who was Chief Justice at the time, placed a Ten Commandments monument on public display in the rotunda of the building.

[sanjeed5]

1. What @engmargo

[engmargo]

OMG, I'm so sorry for my stupid mistake. I just noticed I posted a comment somehow, which actually is a piece of the article. Never mind! ( how can I delete this comment by the way)

[viniciussousaazevedo]

Hello! Is this still experimental? I wanted to use in my project but this 'experimental' thing made me a bit scared to use it, ngl.

[sanjeed5]

you can use it I think, it works fine

[C97-BADREDDINE]

i hope you add experimental library in javascript

[xhuyvn]

I got an error: TypeError: SemanticChunker.create_documents() got an unexpected keyword argument 'breakpoint_threshold_type'

[code]
docs = text_splitter.create_documents([data],breakpoint_threshold_type="standard_deviation")

I use local ollama embeding for testing.

[OleksandrKarhol]

Thanks!

It would be awesome if you added max_chunk_size and chunk_overlap arguments