This PR introduces a new module, `elasticsearch_embeddings.py`, which
provides a wrapper around Elasticsearch embedding models. The new
ElasticsearchEmbeddings class allows users to generate embeddings for
documents and query texts using a [model deployed in an Elasticsearch
cluster](https://www.elastic.co/guide/en/machine-learning/current/ml-nlp-model-ref.html#ml-nlp-model-ref-text-embedding).
### Main features:
1. The ElasticsearchEmbeddings class initializes with an Elasticsearch
connection object and a model_id, providing an interface to interact
with the Elasticsearch ML client through
[infer_trained_model](https://elasticsearch-py.readthedocs.io/en/v8.7.0/api.html?highlight=trained%20model%20infer#elasticsearch.client.MlClient.infer_trained_model)
.
2. The `embed_documents()` method generates embeddings for a list of
documents, and the `embed_query()` method generates an embedding for a
single query text.
3. The class supports custom input text field names in case the deployed
model expects a different field name than the default `text_field`.
4. The implementation is compatible with any model deployed in
Elasticsearch that generates embeddings as output.
### Benefits:
1. Simplifies the process of generating embeddings using Elasticsearch
models.
2. Provides a clean and intuitive interface to interact with the
Elasticsearch ML client.
3. Allows users to easily integrate Elasticsearch-generated embeddings.
Related issue https://github.com/hwchase17/langchain/issues/3400
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
Some LLM's will produce numbered lists with leading whitespace, i.e. in
response to "What is the sum of 2 and 3?":
```
Plan:
1. Add 2 and 3.
2. Given the above steps taken, please respond to the users original question.
```
This commit updates the PlanningOutputParser regex to ignore leading
whitespace before the step number, enabling it to correctly parse this
format.
# Allowing openAI fine-tuned models
Very simple fix that checks whether a openAI `model_name` is a
fine-tuned model when loading `context_size` and when computing call's
cost in the `openai_callback`.
Fixes#2887
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Fix typo + add wikipedia package installation part in
human_input_llm.ipynb
This PR
1. Fixes typo ("the the human input LLM"),
2. Addes wikipedia package installation part (in accordance with
`WikipediaQueryRun`
[documentation](https://python.langchain.com/en/latest/modules/agents/tools/examples/wikipedia.html))
in `human_input_llm.ipynb`
(`docs/modules/models/llms/examples/human_input_llm.ipynb`)
# Add link to Psychic from document loaders documentation page
In my previous PR I forgot to update `document_loaders.rst` to link to
`psychic.ipynb` to make it discoverable from the main documentation.
# Add AzureCognitiveServicesToolkit to call Azure Cognitive Services
API: achieve some multimodal capabilities
This PR adds a toolkit named AzureCognitiveServicesToolkit which bundles
the following tools:
- AzureCogsImageAnalysisTool: calls Azure Cognitive Services image
analysis API to extract caption, objects, tags, and text from images.
- AzureCogsFormRecognizerTool: calls Azure Cognitive Services form
recognizer API to extract text, tables, and key-value pairs from
documents.
- AzureCogsSpeech2TextTool: calls Azure Cognitive Services speech to
text API to transcribe speech to text.
- AzureCogsText2SpeechTool: calls Azure Cognitive Services text to
speech API to synthesize text to speech.
This toolkit can be used to process image, document, and audio inputs.
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Add a WhyLabs callback handler
* Adds a simple WhyLabsCallbackHandler
* Add required dependencies as optional
* protect against missing modules with imports
* Add docs/ecosystem basic example
based on initial prototype from @andrewelizondo
> this integration gathers privacy preserving telemetry on text with
whylogs and sends stastical profiles to WhyLabs platform to monitoring
these metrics over time. For more information on what WhyLabs is see:
https://whylabs.ai
After you run the notebook (if you have env variables set for the API
Keys, org_id and dataset_id) you get something like this in WhyLabs:
Co-authored-by: Andre Elizondo <andre@whylabs.ai>
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Improve TextSplitter.split_documents, collect page_content and
metadata in one iteration
## Who can review?
Community members can review the PR once tests pass. Tag
maintainers/contributors who might be interested:
@eyurtsev In the case where documents is a generator that can only be
iterated once making this change is a huge help. Otherwise a silent
issue happens where metadata is empty for all documents when documents
is a generator. So we expand the argument from `List[Document]` to
`Union[Iterable[Document], Sequence[Document]]`
---------
Co-authored-by: Steven Tartakovsky <tartakovsky.developer@gmail.com>
Implementation is similar to search_distance and where_filter
# adds 'additional' support to Weaviate queries
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
OpenLM is a zero-dependency OpenAI-compatible LLM provider that can call
different inference endpoints directly via HTTP. It implements the
OpenAI Completion class so that it can be used as a drop-in replacement
for the OpenAI API. This changeset utilizes BaseOpenAI for minimal added
code.
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Add Mastodon toots loader.
Loader works either with public toots, or Mastodon app credentials. Toot
text and user info is loaded.
I've also added integration test for this new loader as it works with
public data, and a notebook with example output run now.
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Assign `current_time` to `datetime.now()` if it `current_time is None`
in `time_weighted_retriever`
Fixes#4825
As implemented, `add_documents` in `TimeWeightedVectorStoreRetriever`
assigns `doc.metadata["last_accessed_at"]` and
`doc.metadata["created_at"]` to `datetime.datetime.now()` if
`current_time` is not in `kwargs`.
```python
def add_documents(self, documents: List[Document], **kwargs: Any) -> List[str]:
"""Add documents to vectorstore."""
current_time = kwargs.get("current_time", datetime.datetime.now())
# Avoid mutating input documents
dup_docs = [deepcopy(d) for d in documents]
for i, doc in enumerate(dup_docs):
if "last_accessed_at" not in doc.metadata:
doc.metadata["last_accessed_at"] = current_time
if "created_at" not in doc.metadata:
doc.metadata["created_at"] = current_time
doc.metadata["buffer_idx"] = len(self.memory_stream) + i
self.memory_stream.extend(dup_docs)
return self.vectorstore.add_documents(dup_docs, **kwargs)
```
However, from the way `add_documents` is being called from
`GenerativeAgentMemory`, `current_time` is set as a `kwarg`, but it is
given a value of `None`:
```python
def add_memory(
self, memory_content: str, now: Optional[datetime] = None
) -> List[str]:
"""Add an observation or memory to the agent's memory."""
importance_score = self._score_memory_importance(memory_content)
self.aggregate_importance += importance_score
document = Document(
page_content=memory_content, metadata={"importance": importance_score}
)
result = self.memory_retriever.add_documents([document], current_time=now)
```
The default of `now` was set in #4658 to be None. The proposed fix is
the following:
```python
def add_documents(self, documents: List[Document], **kwargs: Any) -> List[str]:
"""Add documents to vectorstore."""
current_time = kwargs.get("current_time", datetime.datetime.now())
# `current_time` may exist in kwargs, but may still have the value of None.
if current_time is None:
current_time = datetime.datetime.now()
```
Alternatively, we could just set the default of `now` to be
`datetime.datetime.now()` everywhere instead. Thoughts @hwchase17? If we
still want to keep the default to be `None`, then this PR should fix the
above issue. If we want to set the default to be
`datetime.datetime.now()` instead, I can update this PR with that
alternative fix. EDIT: seems like from #5018 it looks like we would
prefer to keep the default to be `None`, in which case this PR should
fix the error.
# changed ValueError to ImportError
Code cleaning.
Fixed inconsistencies in ImportError handling. Sometimes it raises
ImportError and sometime ValueError.
I've changed all cases to the `raise ImportError`
Also:
- added installation instruction in the error message, where it missed;
- fixed several installation instructions in the error message;
- fixed several error handling in regards to the ImportError
Added link option in _process_response
<!--
In _process_respons "snippet" provided non working links for the case
that "links" had the correct answer. Thus added an elif statement before
snippet
-->
<!-- Remove if not applicable -->
Fixes # (issue)
In _process_response link provided correct answers while the snippet
reply provided non working links
@vowelparrot
## Before submitting
<!-- If you're adding a new integration, include an integration test and
an example notebook showing its use! -->
## Who can review?
Community members can review the PR once tests pass. Tag
maintainers/contributors who might be interested:
<!-- For a quicker response, figure out the right person to tag with @
@hwchase17 - project lead
Tracing / Callbacks
- @agola11
Async
- @agola11
DataLoaders
- @eyurtsev
Models
- @hwchase17
- @agola11
Agents / Tools / Toolkits
- @vowelparrot
VectorStores / Retrievers / Memory
- @dev2049
-->
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
`from langchain.experimental.autonomous_agents.autogpt.agent import
AutoGPT` results in an import error as AutoGPT is not defined in the
__init__.py file
https://python.langchain.com/en/latest/use_cases/autonomous_agents/marathon_times.html
An Alternate, way would be to be directly update the import statement to
be `from langchain.experimental import AutoGPT`
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# fix a bug in the add_texts method of Weaviate vector store that creats
wrong embeddings
The following is the original code in the `add_texts` method of the
Weaviate vector store, from line 131 to 153, which contains a bug. The
code here includes some extra explanations in the form of comments and
some omissions.
```python
for i, doc in enumerate(texts):
# some code omitted
if self._embedding is not None:
# variable texts is a list of string and doc here is just a string.
# list(doc) actually breaks up the string into characters.
# so, embeddings[0] is just the embedding of the first character
embeddings = self._embedding.embed_documents(list(doc))
batch.add_data_object(
data_object=data_properties,
class_name=self._index_name,
uuid=_id,
vector=embeddings[0],
)
```
To fix this bug, I pulled the embedding operation out of the for loop
and embed all texts at once.
Co-authored-by: Shawn91 <zyx199199@qq.com>
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# PowerBI major refinement in working of tool and tweaks in the rest
I've gained some experience with more complex sets and the earlier
implementation had too many tries by the agent to create DAX, so
refactored the code to run the LLM to create dax based on a question and
then immediately run the same against the dataset, with retries and a
prompt that includes the error for the retry. This works much better!
Also did some other refactoring of the inner workings, making things
clearer, more concise and faster.
# Row-wise cosine similarity between two equal-width matrices and return
the max top_k score and index, the score all greater than
threshold_score.
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
Enhance the code to support SSL authentication for Elasticsearch when
using the VectorStore module, as previous versions did not provide this
capability.
@dev2049
---------
Co-authored-by: caidong <zhucaidong1992@gmail.com>
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Improve pinecone hybrid search retriever adding metadata support
I simply remove the hardwiring of metadata to the existing
implementation allowing one to pass `metadatas` attribute to the
constructors and in `get_relevant_documents`. I also add one missing pip
install to the accompanying notebook (I am not adding dependencies, they
were pre-existing).
First contribution, just hoping to help, feel free to critique :)
my twitter username is `@andreliebschner`
While looking at hybrid search I noticed #3043 and #1743. I think the
former can be closed as following the example right now (even prior to
my improvements) works just fine, the latter I think can be also closed
safely, maybe pointing out the relevant classes and example. Should I
reply those issues mentioning someone?
@dev2049, @hwchase17
---------
Co-authored-by: Andreas Liebschner <a.liebschner@shopfully.com>
This is a highly optimized update to the pull request
https://github.com/hwchase17/langchain/pull/3269
Summary:
1) Added ability to MRKL agent to self solve the ValueError(f"Could not
parse LLM output: `{llm_output}`") error, whenever llm (especially
gpt-3.5-turbo) does not follow the format of MRKL Agent, while returning
"Action:" & "Action Input:".
2) The way I am solving this error is by responding back to the llm with
the messages "Invalid Format: Missing 'Action:' after 'Thought:'" &
"Invalid Format: Missing 'Action Input:' after 'Action:'" whenever
Action: and Action Input: are not present in the llm output
respectively.
For a detailed explanation, look at the previous pull request.
New Updates:
1) Since @hwchase17 , requested in the previous PR to communicate the
self correction (error) message, using the OutputParserException, I have
added new ability to the OutputParserException class to store the
observation & previous llm_output in order to communicate it to the next
Agent's prompt. This is done, without breaking/modifying any of the
functionality OutputParserException previously performs (i.e.
OutputParserException can be used in the same way as before, without
passing any observation & previous llm_output too).
---------
Co-authored-by: Deepak S V <svdeepak99@users.noreply.github.com>
tldr: The docarray [integration
PR](https://github.com/hwchase17/langchain/pull/4483) introduced a
pinned dependency to protobuf. This is a docarray dependency, not a
langchain dependency. Since this is handled by the docarray
dependencies, it is unnecessary here.
Further, as a pinned dependency, this quickly leads to incompatibilities
with application code that consumes the library. Much less with a
heavily used library like protobuf.
Detail: as we see in the [docarray
integration](https://github.com/hwchase17/langchain/pull/4483/files#diff-50c86b7ed8ac2cf95bd48334961bf0530cdc77b5a56f852c5c61b89d735fd711R81-R83),
the transitive dependencies of docarray were also listed as langchain
dependencies. This is unnecessary as the docarray project has an
appropriate
[extras](a01a05542d/pyproject.toml (L70)).
The docarray project also does not require this _pinned_ version of
protobuf, rather [a minimum
version](a01a05542d/pyproject.toml (L41)).
So this pinned version was likely in error.
To fix this, this PR reverts the explicit hnswlib and protobuf
dependencies and adds the hnswlib extras install for docarray (which
installs hnswlib and protobuf, as originally intended). Because version
`0.32.0`
of the docarray hnswlib extras added protobuf, we bump the docarray
dependency from `^0.31.0` to `^0.32.0`.
# revert docarray explicit transitive dependencies and use extras
instead
## Who can review?
@dev2049 -- reviewed the original PR
@eyurtsev -- bumped the pinned protobuf dependency a few days ago
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
Update to pull request https://github.com/hwchase17/langchain/pull/3215
Summary:
1) Improved the sanitization of query (using regex), by removing python
command (since gpt-3.5-turbo sometimes assumes python console as a
terminal, and runs python command first which causes error). Also
sometimes 1 line python codes contain single backticks.
2) Added 7 new test cases.
For more details, view the previous pull request.
---------
Co-authored-by: Deepak S V <svdeepak99@users.noreply.github.com>
Let user inspect the token ids in addition to getting th enumber of tokens
---------
Co-authored-by: Zach Schillaci <40636930+zachschillaci27@users.noreply.github.com>
Extract the methods specific to running an LLM or Chain on a dataset to
separate utility functions.
This simplifies the client a bit and lets us separate concerns of LCP
details from running examples (e.g., for evals)
# docs: `deployments` page moved into `ecosystem/`
The `Deployments` page moved into the `Ecosystem/` group
Small fixes:
- `index` page: fixed order of items in the `Modules` list, in the `Use
Cases` list
- item `References/Installation` was lost in the `index` page (not on
the Navbar!). Restored it.
- added `|` marker in several places.
NOTE: I also thought about moving the `Additional Resources/Gallery`
page into the `Ecosystem` group but decided to leave it unchanged.
Please, advise on this.
## Who can review?
Community members can review the PR once tests pass. Tag
maintainers/contributors who might be interested:
@dev2049
Without the addition of 'in its original language', the condensing
response, more often than not, outputs the rephrased question in
English, even when the conversation is in another language. This
question in English then transfers to the question in the retrieval
prompt and the chatbot is stuck in English.
I'm sometimes surprised that this does not happen more often, but
apparently the GPT models are smart enough to understand that when the
template contains
Question: ....
Answer:
then the answer should be in in the language of the question.
### Submit Multiple Files to the Unstructured API
Enables batching multiple files into a single Unstructured API requests.
Support for requests with multiple files was added to both
`UnstructuredAPIFileLoader` and `UnstructuredAPIFileIOLoader`. Note that
if you submit multiple files in "single" mode, the result will be
concatenated into a single document. We recommend using this feature in
"elements" mode.
### Testing
The following should load both documents, using two of the example docs
from the integration tests folder.
```python
from langchain.document_loaders import UnstructuredAPIFileLoader
file_paths = ["examples/layout-parser-paper.pdf", "examples/whatsapp_chat.txt"]
loader = UnstructuredAPIFileLoader(
file_paths=file_paths,
api_key="FAKE_API_KEY",
strategy="fast",
mode="elements",
)
docs = loader.load()
```
# Corrected Misspelling in agents.rst Documentation
<!--
Thank you for contributing to LangChain! Your PR will appear in our next
release under the title you set. Please make sure it highlights your
valuable contribution.
Replace this with a description of the change, the issue it fixes (if
applicable), and relevant context. List any dependencies required for
this change.
After you're done, someone will review your PR. They may suggest
improvements. If no one reviews your PR within a few days, feel free to
@-mention the same people again, as notifications can get
-->
In the
[documentation](https://python.langchain.com/en/latest/modules/agents.html)
it says "in fact, it is often best to have an Action Agent be in
**change** of the execution for the Plan and Execute agent."
**Suggested Change:** I propose correcting change to charge.
Fix for issue: #5039
# Add documentation for Databricks integration
This is a follow-up of https://github.com/hwchase17/langchain/pull/4702
It documents the details of how to integrate Databricks using langchain.
It also provides examples in a notebook.
## Who can review?
@dev2049 @hwchase17 since you are aware of the context. We will promote
the integration after this doc is ready. Thanks in advance!
# Fixes an annoying typo in docs
<!--
Thank you for contributing to LangChain! Your PR will appear in our next
release under the title you set. Please make sure it highlights your
valuable contribution.
Replace this with a description of the change, the issue it fixes (if
applicable), and relevant context. List any dependencies required for
this change.
After you're done, someone will review your PR. They may suggest
improvements. If no one reviews your PR within a few days, feel free to
@-mention the same people again, as notifications can get lost.
-->
<!-- Remove if not applicable -->
Fixes Annoying typo in docs - "Therefor" -> "Therefore". It's so
annoying to read that I just had to make this PR.
# Streaming only final output of agent (#2483)
As requested in issue #2483, this Callback allows to stream only the
final output of an agent (ie not the intermediate steps).
Fixes#2483
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
# Ensuring that users pass a single prompt when calling a LLM
- This PR adds a check to the `__call__` method of the `BaseLLM` class
to ensure that it is called with a single prompt
- Raises a `ValueError` if users try to call a LLM with a list of prompt
and instructs them to use the `generate` method instead
## Why this could be useful
I stumbled across this by accident. I accidentally called the OpenAI LLM
with a list of prompts instead of a single string and still got a
result:
```
>>> from langchain.llms import OpenAI
>>> llm = OpenAI()
>>> llm(["Tell a joke"]*2)
"\n\nQ: Why don't scientists trust atoms?\nA: Because they make up everything!"
```
It might be better to catch such a scenario preventing unnecessary costs
and irritation for the user.
## Proposed behaviour
```
>>> from langchain.llms import OpenAI
>>> llm = OpenAI()
>>> llm(["Tell a joke"]*2)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/marcus/Projects/langchain/langchain/llms/base.py", line 291, in __call__
raise ValueError(
ValueError: Argument `prompt` is expected to be a single string, not a list. If you want to run the LLM on multiple prompts, use `generate` instead.
```
# Add self query translator for weaviate vectorstore
Adds support for the EQ comparator and the AND/OR operators.
Co-authored-by: Dominic Chan <dchan@cppib.com>
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
