langchain/fix-cohere-reranker-rerank-method #19486

jjovalle99 · 2024-03-24T19:12:40Z

Description

Fixed the following error with rerank method from CohereRerank:

---> [79](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:79) results = self.client.rerank(
     [80](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:80)     query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc
     [81](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:81) )
     [82](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:82) result_dicts = []
     [83](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:83) for res in results.results:

TypeError: BaseCohere.rerank() takes 1 positional argument but 4 positional arguments (and 2 keyword-only arguments) were given

This was easily fixed going from this:

   def rerank(
        self,
        documents: Sequence[Union[str, Document, dict]],
        query: str,
        *,
        model: Optional[str] = None,
        top_n: Optional[int] = -1,
        max_chunks_per_doc: Optional[int] = None,
    ) -> List[Dict[str, Any]]:
         ...
        if len(documents) == 0:  # to avoid empty api call
            return []
        docs = [
            doc.page_content if isinstance(doc, Document) else doc for doc in documents
        ]
        model = model or self.model
        top_n = top_n if (top_n is None or top_n > 0) else self.top_n
        results = self.client.rerank(
            query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc
        )
        result_dicts = []
        for res in results:
            result_dicts.append(
                {"index": res.index, "relevance_score": res.relevance_score}
            )
        return result_dicts

to this:

    def rerank(
        self,
        documents: Sequence[Union[str, Document, dict]],
        query: str,
        *,
        model: Optional[str] = None,
        top_n: Optional[int] = -1,
        max_chunks_per_doc: Optional[int] = None,
    ) -> List[Dict[str, Any]]:
         ...
        if len(documents) == 0:  # to avoid empty api call
            return []
        docs = [
            doc.page_content if isinstance(doc, Document) else doc for doc in documents
        ]
        model = model or self.model
        top_n = top_n if (top_n is None or top_n > 0) else self.top_n
        results = self.client.rerank(
            query=query, documents=docs, model=model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc <-------------
        )
        result_dicts = []
        for res in results.results:  <-------------
            result_dicts.append(
                {"index": res.index, "relevance_score": res.relevance_score}
            )
        return result_dicts

Unit & Integration tests

I added a unit test to check the behaviour of rerank. Also fixed the original integration test which was failing.

Format & Linting

Everything worked properly with make lint_diff, make format_diff and make format. However I noticed an error coming from other part of the library when doing make lint:

(langchain-py3.9) ➜  langchain git:(master) make format
[ "." = "" ] || poetry run ruff format .
1636 files left unchanged
[ "." = "" ] || poetry run ruff --select I --fix .
(langchain-py3.9) ➜  langchain git:(master) make lint
./scripts/check_pydantic.sh .
./scripts/lint_imports.sh
poetry run ruff .
[ "." = "" ] || poetry run ruff format . --diff
1636 files already formatted
[ "." = "" ] || poetry run ruff --select I .
[ "." = "" ] || mkdir -p .mypy_cache && poetry run mypy . --cache-dir .mypy_cache
langchain/agents/openai_assistant/base.py:252: error: Argument "file_ids" to "create" of "Assistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]"  [arg-type]
langchain/agents/openai_assistant/base.py:374: error: Argument "file_ids" to "create" of "AsyncAssistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]"  [arg-type]
Found 2 errors in 1 file (checked 1634 source files)
make: *** [Makefile:65: lint] Error 1

vercel · 2024-03-24T19:12:46Z

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)	Visit Preview		Mar 28, 2024 6:27am

mspronesti

This issue seems to be due to the changes made to Cohere API since v5:

cohere.Client.rerank now expects named parameters
The return value is now a RerankResponse class, which doesn't implement __iter__.

I recommend updating cohere's version in the requirements though (from v4 to v5).

jjovalle99 · 2024-03-25T02:22:12Z

@mspronesti What do you recommend to be the next steps for this PR? The changes I made handle the changes in the Cohere API.

ddl-john-alexander · 2024-03-25T15:34:42Z

This is a blocking issue for using Rerank. And it was working previously: #19461 Can we get this merged or another solution?

mspronesti · 2024-03-25T15:58:36Z

@jjovalle99 Only to update Cohere's version in the .toml file (at the line I pointed you to in my previous comment).

jjovalle99 · 2024-03-25T16:58:45Z

@mspronesti @baskaryan Cohere version updated from @^4 to @^5. Let me know if something else is needed.

baskaryan · 2024-03-25T23:30:24Z

the cohere team is actually working on splitting all cohere integrations into their own partner package, and in the process updating them to have cohere v5 support! see #19049. the classes in langchain-community will be deprecated in favor of the ones in langchain-cohere

ddl-john-alexander · 2024-03-25T23:46:13Z

Ok, thanks for the update, @baskaryan! Will switching to langchain-cohere now fix this issue?

ddl-john-alexander · 2024-03-26T02:14:11Z

@baskaryan the https://github.com/langchain-ai/langchain/blob/master/libs/langchain/pyproject.toml#L39 cohere version here still needs to be updated from 4 to 5... as referenced above by @mspronesti.

billytrend-cohere · 2024-03-27T00:43:49Z

This fix looks good to me, will copy it into our partner repo, thanks @jjovalle99 and folks

jjovalle99 · 2024-03-27T00:58:34Z

This fix looks good to me, will copy it into our partner repo, thanks @jjovalle99 and folks

Awesome! willing to help if something comes up @billytrend-cohere

billytrend-cohere · 2024-03-27T01:07:21Z

Created the pr, we can probably merge this PR as well but recommend you use our partner package now

jjovalle99 · 2024-03-27T01:11:19Z

Created the pr, we can probably merge this PR as well but recommend you use our partner package now

Sounds great!

libs/langchain/langchain/retrievers/document_compressors/cohere_rerank.py

Fix cohere rerank inspired by #19486

libs/langchain/pyproject.toml

jjovalle99 · 2024-03-28T03:38:58Z

@baskaryan Done!

AndreaEr · 2024-03-29T10:31:16Z

how should i change if i'm using CohereRerank() in ipynb

jjovalle99 · 2024-03-29T14:09:43Z

how should i change if i'm using CohereRerank() in ipynb

@AndreaEr I would say that you should now use everything from the Cohere partner library langchain_cohere.

So using langchain_cohere.rerank.CohereRerank() will be the appropriate way. If for any reason you can't change to their partner library, you can still use the deprecated CohereRerank, and it will also work, but it is not recommended.

Fix cohere rerank inspired by langchain-ai#19486

…angchain-ai#19486) #### Description Fixed the following error with `rerank` method from `CohereRerank`: ``` ---> [79](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:79) results = self.client.rerank( [80](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:80) query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc [81](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:81) ) [82](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:82) result_dicts = [] [83](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:83) for res in results.results: TypeError: BaseCohere.rerank() takes 1 positional argument but 4 positional arguments (and 2 keyword-only arguments) were given ``` This was easily fixed going from this: ``` def rerank( self, documents: Sequence[Union[str, Document, dict]], query: str, *, model: Optional[str] = None, top_n: Optional[int] = -1, max_chunks_per_doc: Optional[int] = None, ) -> List[Dict[str, Any]]: ... if len(documents) == 0: # to avoid empty api call return [] docs = [ doc.page_content if isinstance(doc, Document) else doc for doc in documents ] model = model or self.model top_n = top_n if (top_n is None or top_n > 0) else self.top_n results = self.client.rerank( query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc ) result_dicts = [] for res in results: result_dicts.append( {"index": res.index, "relevance_score": res.relevance_score} ) return result_dicts ``` to this: ``` def rerank( self, documents: Sequence[Union[str, Document, dict]], query: str, *, model: Optional[str] = None, top_n: Optional[int] = -1, max_chunks_per_doc: Optional[int] = None, ) -> List[Dict[str, Any]]: ... if len(documents) == 0: # to avoid empty api call return [] docs = [ doc.page_content if isinstance(doc, Document) else doc for doc in documents ] model = model or self.model top_n = top_n if (top_n is None or top_n > 0) else self.top_n results = self.client.rerank( query=query, documents=docs, model=model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc <------------- ) result_dicts = [] for res in results.results: <------------- result_dicts.append( {"index": res.index, "relevance_score": res.relevance_score} ) return result_dicts ``` #### Unit & Integration tests I added a unit test to check the behaviour of `rerank`. Also fixed the original integration test which was failing. #### Format & Linting Everything worked properly with `make lint_diff`, `make format_diff` and `make format`. However I noticed an error coming from other part of the library when doing `make lint`: ``` (langchain-py3.9) ➜ langchain git:(master) make format [ "." = "" ] || poetry run ruff format . 1636 files left unchanged [ "." = "" ] || poetry run ruff --select I --fix . (langchain-py3.9) ➜ langchain git:(master) make lint ./scripts/check_pydantic.sh . ./scripts/lint_imports.sh poetry run ruff . [ "." = "" ] || poetry run ruff format . --diff 1636 files already formatted [ "." = "" ] || poetry run ruff --select I . [ "." = "" ] || mkdir -p .mypy_cache && poetry run mypy . --cache-dir .mypy_cache langchain/agents/openai_assistant/base.py:252: error: Argument "file_ids" to "create" of "Assistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]" [arg-type] langchain/agents/openai_assistant/base.py:374: error: Argument "file_ids" to "create" of "AsyncAssistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]" [arg-type] Found 2 errors in 1 file (checked 1634 source files) make: *** [Makefile:65: lint] Error 1 ``` --------- Co-authored-by: Bagatur <[email protected]> Co-authored-by: Bagatur <[email protected]>

Fix cohere rerank inspired by langchain-ai/langchain#19486

Fix cohere rerank inspired by #19486

…19486) #### Description Fixed the following error with `rerank` method from `CohereRerank`: ``` ---> [79](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:79) results = self.client.rerank( [80](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:80) query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc [81](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:81) ) [82](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:82) result_dicts = [] [83](https://vscode-remote+wsl-002bubuntu.vscode-resource.vscode-cdn.net/home/jjmov99/legal-colombia/~/legal-colombia/.venv/lib/python3.11/site-packages/langchain/retrievers/document_compressors/cohere_rerank.py:83) for res in results.results: TypeError: BaseCohere.rerank() takes 1 positional argument but 4 positional arguments (and 2 keyword-only arguments) were given ``` This was easily fixed going from this: ``` def rerank( self, documents: Sequence[Union[str, Document, dict]], query: str, *, model: Optional[str] = None, top_n: Optional[int] = -1, max_chunks_per_doc: Optional[int] = None, ) -> List[Dict[str, Any]]: ... if len(documents) == 0: # to avoid empty api call return [] docs = [ doc.page_content if isinstance(doc, Document) else doc for doc in documents ] model = model or self.model top_n = top_n if (top_n is None or top_n > 0) else self.top_n results = self.client.rerank( query, docs, model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc ) result_dicts = [] for res in results: result_dicts.append( {"index": res.index, "relevance_score": res.relevance_score} ) return result_dicts ``` to this: ``` def rerank( self, documents: Sequence[Union[str, Document, dict]], query: str, *, model: Optional[str] = None, top_n: Optional[int] = -1, max_chunks_per_doc: Optional[int] = None, ) -> List[Dict[str, Any]]: ... if len(documents) == 0: # to avoid empty api call return [] docs = [ doc.page_content if isinstance(doc, Document) else doc for doc in documents ] model = model or self.model top_n = top_n if (top_n is None or top_n > 0) else self.top_n results = self.client.rerank( query=query, documents=docs, model=model, top_n=top_n, max_chunks_per_doc=max_chunks_per_doc <------------- ) result_dicts = [] for res in results.results: <------------- result_dicts.append( {"index": res.index, "relevance_score": res.relevance_score} ) return result_dicts ``` #### Unit & Integration tests I added a unit test to check the behaviour of `rerank`. Also fixed the original integration test which was failing. #### Format & Linting Everything worked properly with `make lint_diff`, `make format_diff` and `make format`. However I noticed an error coming from other part of the library when doing `make lint`: ``` (langchain-py3.9) ➜ langchain git:(master) make format [ "." = "" ] || poetry run ruff format . 1636 files left unchanged [ "." = "" ] || poetry run ruff --select I --fix . (langchain-py3.9) ➜ langchain git:(master) make lint ./scripts/check_pydantic.sh . ./scripts/lint_imports.sh poetry run ruff . [ "." = "" ] || poetry run ruff format . --diff 1636 files already formatted [ "." = "" ] || poetry run ruff --select I . [ "." = "" ] || mkdir -p .mypy_cache && poetry run mypy . --cache-dir .mypy_cache langchain/agents/openai_assistant/base.py:252: error: Argument "file_ids" to "create" of "Assistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]" [arg-type] langchain/agents/openai_assistant/base.py:374: error: Argument "file_ids" to "create" of "AsyncAssistants" has incompatible type "Optional[Any]"; expected "Union[list[str], NotGiven]" [arg-type] Found 2 errors in 1 file (checked 1634 source files) make: *** [Makefile:65: lint] Error 1 ``` --------- Co-authored-by: Bagatur <[email protected]> Co-authored-by: Bagatur <[email protected]>

ling-chun · 2024-07-05T02:34:59Z

thx, thats help me a lot

jjovalle99 added 3 commits March 24, 2024 13:32

Fixing cohere reranker call

ed86d32

Adding relevant unit test

4624df5

Fixing integration test for Cohere Reranker

42de343

dosubot bot added the size:M This PR changes 30-99 lines, ignoring generated files. label Mar 24, 2024

dosubot bot added Ɑ: retriever Related to retriever module 🔌: cohere Primarily related to Cohere integrations 🤖:bug Related to a bug, vulnerability, unexpected error with an existing feature labels Mar 24, 2024

mspronesti reviewed Mar 24, 2024

View reviewed changes

Merge branch 'master' into fix/cohere-reranker-rerank-method

d0c9a38

ddl-john-alexander mentioned this pull request Mar 25, 2024

community: Add support for cohere SDK v5 (keeps v4 backwards compatibility) #19084

Merged

Updating cohere@^4 to cohere@^5

2edc098

billytrend-cohere mentioned this pull request Mar 27, 2024

cohere: Fix cohere rerank #19624

Merged

Merge branch 'master' into fix/cohere-reranker-rerank-method

4d26d19

baskaryan reviewed Mar 27, 2024

View reviewed changes

libs/langchain/langchain/retrievers/document_compressors/cohere_rerank.py Outdated Show resolved Hide resolved

baskaryan pushed a commit that referenced this pull request Mar 27, 2024

cohere[patch]: Fix cohere rerank (#19624)

85f57ab

Fix cohere rerank inspired by #19486

baskaryan reviewed Mar 28, 2024

View reviewed changes

libs/langchain/pyproject.toml Show resolved Hide resolved

jjovalle99 added 2 commits March 27, 2024 22:37

Updating cohere to version = '>=4,<6'

7755c27

Backward compatibility cohere rerank

fdcc170

fmt

f7b63f5

baskaryan approved these changes Mar 28, 2024

View reviewed changes

dosubot bot added the lgtm PR looks good. Use to confirm that a PR is ready for merging. label Mar 28, 2024

baskaryan enabled auto-merge (squash) March 28, 2024 06:27

baskaryan merged commit 51baa1b into langchain-ai:master Mar 28, 2024
41 checks passed

ddl-john-alexander mentioned this pull request Mar 28, 2024

langchain: update /libs/langchain/pyproject.toml w Cohere latest release #19605

Closed

gkorland pushed a commit to FalkorDB/langchain that referenced this pull request Mar 30, 2024

cohere[patch]: Fix cohere rerank (langchain-ai#19624)

20404ea

Fix cohere rerank inspired by langchain-ai#19486

Je-Cp pushed a commit to Je-Cp/jcp-langchain that referenced this pull request Apr 2, 2024

cohere[patch]: Fix cohere rerank (#19624)

8427316

Fix cohere rerank inspired by langchain-ai/langchain#19486

hinthornw pushed a commit that referenced this pull request Apr 26, 2024

cohere[patch]: Fix cohere rerank (#19624)

9cdc2dc

Fix cohere rerank inspired by #19486

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

langchain/fix-cohere-reranker-rerank-method #19486

langchain/fix-cohere-reranker-rerank-method #19486

jjovalle99 commented Mar 24, 2024

vercel bot commented Mar 24, 2024 •

edited

Loading

mspronesti left a comment •

edited

Loading

jjovalle99 commented Mar 25, 2024

ddl-john-alexander commented Mar 25, 2024

mspronesti commented Mar 25, 2024

jjovalle99 commented Mar 25, 2024

baskaryan commented Mar 25, 2024

ddl-john-alexander commented Mar 25, 2024

ddl-john-alexander commented Mar 26, 2024

billytrend-cohere commented Mar 27, 2024

jjovalle99 commented Mar 27, 2024

billytrend-cohere commented Mar 27, 2024

jjovalle99 commented Mar 27, 2024

jjovalle99 commented Mar 28, 2024

AndreaEr commented Mar 29, 2024

jjovalle99 commented Mar 29, 2024

ling-chun commented Jul 5, 2024

langchain/fix-cohere-reranker-rerank-method #19486

langchain/fix-cohere-reranker-rerank-method #19486

Conversation

jjovalle99 commented Mar 24, 2024

Description

Unit & Integration tests

Format & Linting

vercel bot commented Mar 24, 2024 • edited Loading

mspronesti left a comment • edited Loading

Choose a reason for hiding this comment

jjovalle99 commented Mar 25, 2024

ddl-john-alexander commented Mar 25, 2024

mspronesti commented Mar 25, 2024

jjovalle99 commented Mar 25, 2024

baskaryan commented Mar 25, 2024

ddl-john-alexander commented Mar 25, 2024

ddl-john-alexander commented Mar 26, 2024

billytrend-cohere commented Mar 27, 2024

jjovalle99 commented Mar 27, 2024

billytrend-cohere commented Mar 27, 2024

jjovalle99 commented Mar 27, 2024

jjovalle99 commented Mar 28, 2024

AndreaEr commented Mar 29, 2024

jjovalle99 commented Mar 29, 2024

ling-chun commented Jul 5, 2024

vercel bot commented Mar 24, 2024 •

edited

Loading

mspronesti left a comment •

edited

Loading