Add a ListRerank document compressor
#13311
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
dosubot
bot
added
the
🤖:enhancement
denver1117
commented
Nov 24, 2023
| type="array[dict]", | ||
| ) | ||
| ] | ||
| output_parser = StructuredOutputParser.from_response_schemas(response_schemas) |
There was a problem hiding this comment.
I experimented with a Pydantic parser that defines the full nested structure explicitly and saw notably more output parsing errors. Expressing the array[dict] type as an implicit nested type within a single ResponseSchema type argument was much more successful.
dosubot
bot
added
the
size:L
hwchase17
approved these changes
Nov 29, 2023
dosubot
bot
added
the
lgtm
dosubot
bot
removed
the
lgtm
Thanks @hwchase17. I fixed the lint issues and added to the documentation, it LGTM in the Vercel preview: 
This is ready to go from my perspective. |
hwchase17
approved these changes
Dec 5, 2023
dosubot
bot
added
the
lgtm
…7/langchain into feature/list-rerank-compressor
|
I fixed the order of arguments (need arg first, not kwarg) and fixed the bad type in the Callable arg as noted by the failed check. |
|
needs a re-review and clean up, but i generally like it |
ccurme
approved these changes
Jul 18, 2024
olgamurraft
pushed a commit
to olgamurraft/langchain
that referenced
this pull request
Aug 16, 2024
- **Description:** This PR adds a new document compressor called `ListRerank`. It's derived from `BaseDocumentCompressor`. It's a near exact implementation of introduced by this paper: [Zero-Shot Listwise Document Reranking with a Large Language Model](https://arxiv.org/pdf/2305.02156.pdf) which it finds to outperform pointwise reranking, which is somewhat implemented in LangChain as [LLMChainFilter](https://github.com/langchain-ai/langchain/blob/master/libs/langchain/langchain/retrievers/document_compressors/chain_filter.py). - **Issue:** None - **Dependencies:** None - **Tag maintainer:** @hwchase17 @izzymsft - **Twitter handle:** @HarrisEMitchell Notes: 1. I didn't add anything to `docs`. I wasn't exactly sure which patterns to follow as [cohere reranker is under Retrievers](https://python.langchain.com/docs/integrations/retrievers/cohere-reranker) with other external document retrieval integrations, but other contextual compression is [here](https://python.langchain.com/docs/modules/data_connection/retrievers/contextual_compression/). Happy to contribute to either with some direction. 2. I followed syntax, docstrings, implementation patterns, etc. as well as I could looking at nearby modules. One thing I didn't do was put the default prompt in a separate `.py` file like [Chain Filter](https://github.com/langchain-ai/langchain/blob/master/libs/langchain/langchain/retrievers/document_compressors/chain_filter_prompt.py) and [Chain Extract](https://github.com/langchain-ai/langchain/blob/master/libs/langchain/langchain/retrievers/document_compressors/chain_extract_prompt.py). Happy to follow that pattern if it would be preferred. --------- Co-authored-by: Harrison Chase <[email protected]> Co-authored-by: Bagatur <[email protected]> Co-authored-by: Chester Curme <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
ListRerank. It's derived fromBaseDocumentCompressor. It's a near exact implementation of introduced by this paper: Zero-Shot Listwise Document Reranking with a Large Language Model which it finds to outperform pointwise reranking, which is somewhat implemented in LangChain as LLMChainFilter.Notes:
docs. I wasn't exactly sure which patterns to follow as cohere reranker is under Retrievers with other external document retrieval integrations, but other contextual compression is here. Happy to contribute to either with some direction..pyfile like Chain Filter and Chain Extract. Happy to follow that pattern if it would be preferred.