Haven't gotten to all of them, but this:
- Updates some of the tools notebooks to actually instantiate a tool
(many just show a 'utility' rather than a tool. More changes to come in
separate PR)
- Move the `Tool` and decorator definitions to `langchain/tools/base.py`
(but still export from `langchain.agents`)
- Add scene explain to the load_tools() function
- Add unit tests for public apis for the langchain.tools and langchain.agents modules
Move tool validation to each implementation of the Agent.
Another alternative would be to adjust the `_validate_tools()` signature
to accept the output parser (and format instructions) and add logic
there. Something like
`parser.outputs_structured_actions(format_instructions)`
But don't think that's needed right now.
- Add langchain.llms.GooglePalm for text completion,
- Add langchain.chat_models.ChatGooglePalm for chat completion,
- Add langchain.embeddings.GooglePalmEmbeddings for sentence embeddings,
- Add example field to HumanMessage and AIMessage so that users can feed
in examples into the PaLM Chat API,
- Add system and unit tests.
Note async completion for the Text API is not yet supported and will be
included in a future PR.
Happy for feedback on any aspect of this PR, especially our choice of
adding an example field to Human and AI Message objects to enable
passing example messages to the API.
This pull request adds unit tests for various output parsers
(BooleanOutputParser, CommaSeparatedListOutputParser, and
StructuredOutputParser) to ensure their correct functionality and to
increase code reliability and maintainability. The tests cover both
valid and invalid input cases.
Changes:
Added unit tests for BooleanOutputParser.
Added unit tests for CommaSeparatedListOutputParser.
Added unit tests for StructuredOutputParser.
Testing:
All new unit tests have been executed, and they pass successfully.
The overall test suite has been run, and all tests pass.
Notes:
These tests cover both successful parsing scenarios and error handling
for invalid inputs.
If any new output parsers are added in the future, corresponding unit
tests should also be created to maintain coverage.
Enum to string conversion handled differently between python 3.9 and
3.11, currently breaking in 3.11 (see #3788). Thanks @peter-brady for
catching this!
In the current solution, AgentType and AGENT_TO_CLASS are placed in two
separate files and both manually maintained. This might cause
inconsistency when we update either of them.
— latest —
based on the discussion with hwchase17, we don’t know how to further use
the newly introduced AgentTypeConfig type, so it doesn’t make sense yet
to add it. Instead, it’s better to move the dictionary to another file
to keep the loading.py file clear. The consistency is a good point.
Instead of asserting the consistency during linting, we added a unittest
for consistency check. I think it works as auto unittest is triggered
every time with clear failure notice. (well, force push is possible, but
we all know what we are doing, so let’s show trust. :>)
~~This PR includes~~
- ~~Introduced AgentTypeConfig as the source of truth of all AgentType
related meta data.~~
- ~~Each AgentTypeConfig is a annotated class type which can be used for
annotation in other places.~~
- ~~Each AgentTypeConfig can be easily extended when we have more meta
data needs.~~
- ~~Strong assertion to ensure AgentType and AGENT_TO_CLASS are always
consistent.~~
- ~~Made AGENT_TO_CLASS automatically generated.~~
~~Test Plan:~~
- ~~since this change is focusing on annotation, lint is the major test
focus.~~
- ~~lint, format and test passed on local.~~
This **partially** addresses
https://github.com/hwchase17/langchain/issues/1524, but it's also useful
for some of our use cases.
This `DocstoreFn` allows to lookup a document given a function that
accepts the `search` string without the need to implement a custom
`Docstore`.
This could be useful when:
* you don't want to implement a `Docstore` just to provide a custom
`search`
* it's expensive to construct an `InMemoryDocstore`/dict
* you retrieve documents from remote sources
* you just want to reuse existing objects
Add other File Utilities, include
- List Directory
- Search for file
- Move
- Copy
- Remove file
Bundle as toolkit
Add a notebook that connects to the Chat Agent, which somewhat supports
multi-arg input tools
Update original read/write files to return the original dir paths and
better handle unsupported file paths.
Add unit tests
I think the logic of
https://github.com/hwchase17/langchain/pull/3684#pullrequestreview-1405358565
is too confusing.
I prefer this alternative because:
- All `Tool()` implementations by default will be treated the same as
before. No breaking changes.
- Less reliance on pydantic magic
- The decorator (which only is typed as returning a callable) can infer
schema and generate a structured tool
- Either way, the recommended way to create a custom tool is through
inheriting from the base tool
Tradeoffs here:
- No lint-time checking for compatibility
- Differs from JS package
- The signature inference, etc. in the base tool isn't simple
- The `args_schema` is optional
Pros:
- Forwards compatibility retained
- Doesn't break backwards compatibility
- User doesn't have to think about which class to subclass (single base
tool or dynamic `Tool` interface regardless of input)
- No need to change the load_tools, etc. interfaces
Co-authored-by: Hasan Patel <mangafield@gmail.com>
This catches the warning raised when using duckdb, asserts that it's as expected.
The goal is to resolve all existing warnings to make unit-testing much stricter.
This PR introduces a Blob data type and a Blob loader interface.
This is the first of a sequence of PRs that follows this proposal:
https://github.com/hwchase17/langchain/pull/2833
The primary goals of these abstraction are:
* Decouple content loading from content parsing code.
* Help duplicated content loading code from document loaders.
* Make lazy loading a default for langchain.
This commit adds a new unit test for the _merge_splits function in the
text splitter. The new test verifies that the function merges text into
chunks of the correct size and overlap, using a specified separator. The
test passes on the current implementation of the function.
Fix for: [Changed regex to cover new line before action
serious.](https://github.com/hwchase17/langchain/issues/3365)
---
This PR fixes the issue where `ValueError: Could not parse LLM output:`
was thrown on seems to be valid input.
Changed regex to cover new lines before action serious (after the
keywords "Action:" and "Action Input:").
regex101: https://regex101.com/r/CXl1kB/1
---------
Co-authored-by: msarskus <msarskus@cisco.com>
- Proactively raise error if a tool subclasses BaseTool, defines its
own schema, but fails to add the type-hints
- fix the auto-inferred schema of the decorator to strip the
unneeded virtual kwargs from the schema dict
Helps avoid silent instances of #3297
- Permit the specification of a `root_dir` to the read/write file tools
to specify a working directory
- Add validation for attempts to read/write outside the directory (e.g.,
through `../../` or symlinks or `/abs/path`'s that don't lie in the
correct path)
- Add some tests for all
One question is whether we should make a default root directory for
these? tradeoffs either way
`langchain.prompts.PromptTemplate` and
`langchain.prompts.FewShotPromptTemplate` do not validate
`input_variables` when initialized as `jinja2` template.
```python
# Using langchain v0.0.144
template = """"\
Your variable: {{ foo }}
{% if bar %}
You just set bar boolean variable to true
{% endif %}
"""
# Missing variable, should raise ValueError
prompt_template = PromptTemplate(template=template,
input_variables=["bar"],
template_format="jinja2",
validate_template=True)
# Extra variable, should raise ValueError
prompt_template = PromptTemplate(template=template,
input_variables=["bar", "foo", "extra", "thing"],
template_format="jinja2",
validate_template=True)
```
- Remove dynamic model creation in the `args()` property. _Only infer
for the decorator (and add an argument to NOT infer if someone wishes to
only pass as a string)_
- Update the validation example to make it less likely to be
misinterpreted as a "safe" way to run a repl
There is one example of "Multi-argument tools" in the custom_tools.ipynb
from yesterday, but we could add more. The output parsing for the base
MRKL agent hasn't been adapted to handle structured args at this point
in time
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
`langchain.prompts.PromptTemplate` is unable to infer `input_variables`
from jinja2 template.
```python
# Using langchain v0.0.141
template_string = """\
Hello world
Your variable: {{ var }}
{# This will not get rendered #}
{% if verbose %}
Congrats! You just turned on verbose mode and got extra messages!
{% endif %}
"""
template = PromptTemplate.from_template(template_string, template_format="jinja2")
print(template.input_variables) # Output ['# This will not get rendered #', '% endif %', '% if verbose %']
```
---------
Co-authored-by: engkheng <ongengkheng929@example.com>
Add a time-weighted memory retriever and a notebook that approximates a
Generative Agent from https://arxiv.org/pdf/2304.03442.pdf
The "daily plan" components are removed for now since they are less
useful without a virtual world, but the memory is an interesting
component to build off.
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Use numexpr evaluate instead of the python REPL to avoid malicious code
injection.
Tested against the (limited) math dataset and got the same score as
before.
For more permissive tools (like the REPL tool itself), other approaches
ought to be provided (some combination of Sanitizer + Restricted python
+ unprivileged-docker + ...), but for a calculator tool, only
mathematical expressions should be permitted.
See https://github.com/hwchase17/langchain/issues/814
When the code ran by the PythonAstREPLTool contains multiple statements
it will fallback to exec() instead of using eval(). With this change, it
will also return the output of the code in the same way the
PythonREPLTool will.
Currently, the output type of a number of OutputParser's `parse` methods
is `Any` when it can in fact be inferred.
This PR makes BaseOutputParser use a generic type and fixes the output
types of the following parsers:
- `PydanticOutputParser`
- `OutputFixingParser`
- `RetryOutputParser`
- `RetryWithErrorOutputParser`
The output of the `StructuredOutputParser` is corrected from `BaseModel`
to `Any` since there are no type guarantees provided by the parser.
Fixes issue #2715
`combine_docs` does not go through the standard chain call path which
means that chain callbacks won't be triggered, meaning QA chains won't
be traced properly, this fixes that.
Also fix several errors in the chat_vector_db notebook
Right now, eval chains require an answer for every question. It's
cumbersome to collect this ground truth so getting around this issue
with 2 things:
* Adding a context param in `ContextQAEvalChain` and simply evaluating
if the question is answered accurately from context
* Adding chain of though explanation prompting to improve the accuracy
of this w/o GT.
This also gets to feature parity with openai/evals which has the same
contextual eval w/o GT.
TODO in follow-up:
* Better prompt inheritance. No need for seperate prompt for CoT
reasoning. How can we merge them together
---------
Co-authored-by: Vashisht Madhavan <vashishtmadhavan@Vashs-MacBook-Pro.local>
This still doesn't handle the following
- non-JSON media types
- anyOf, allOf, oneOf's
And doesn't emit the typescript definitions for referred types yet, but
that can be saved for a separate PR.
Also, we could have better support for Swagger 2.0 specs and OpenAPI
3.0.3 (can use the same lib for the latter) recommend offline conversion
for now.
`AgentExecutor` already has support for limiting the number of
iterations. But the amount of time taken for each iteration can vary
quite a bit, so it is difficult to place limits on the execution time.
This PR adds a new field `max_execution_time` to the `AgentExecutor`
model. When called asynchronously, the agent loop is wrapped in an
`asyncio.timeout()` context which triggers the early stopping response
if the time limit is reached. When called synchronously, the agent loop
checks for both the max_iteration limit and the time limit after each
iteration.
When used asynchronously `max_execution_time` gives really tight control
over the max time for an execution chain. When used synchronously, the
chain can unfortunately exceed max_execution_time, but it still gives
more control than trying to estimate the number of max_iterations needed
to cap the execution time.
---------
Co-authored-by: Zachary Jones <zjones@zetaglobal.com>
This pull request adds an enum class for the various types of agents
used in the project, located in the `agent_types.py` file. Currently,
the project is using hardcoded strings for the initialization of these
agents, which can lead to errors and make the code harder to maintain.
With the introduction of the new enums, the code will be more readable
and less error-prone.
The new enum members include:
- ZERO_SHOT_REACT_DESCRIPTION
- REACT_DOCSTORE
- SELF_ASK_WITH_SEARCH
- CONVERSATIONAL_REACT_DESCRIPTION
- CHAT_ZERO_SHOT_REACT_DESCRIPTION
- CHAT_CONVERSATIONAL_REACT_DESCRIPTION
In this PR, I have also replaced the hardcoded strings with the
appropriate enum members throughout the codebase, ensuring a smooth
transition to the new approach.
Fix issue#1645: Parse either whitespace or newline after 'Action Input:'
in llm_output in mrkl agent.
Unittests added accordingly.
Co-authored-by: ₿ingnan.ΞTH <brillliantz@outlook.com>
I was getting the same issue reported in #1339 by
[MacYang555](https://github.com/MacYang555) when running the test suite
on my Mac. I implemented the fix they suggested to use a regex match in
the output assertion for the scenario under test.
Resolves#1339
Seeing a lot of issues in Discord in which the LLM is not using the
correct LIMIT clause for different SQL dialects. ie, it's using `LIMIT`
for mssql instead of `TOP`, or instead of `ROWNUM` for Oracle, etc.
I think this could be due to us specifying the LIMIT statement in the
example rows portion of `table_info`. So the LLM is seeing the `LIMIT`
statement used in the prompt.
Since we can't specify each dialect's method here, I think it's fine to
just replace the `SELECT... LIMIT 3;` statement with `3 rows from
table_name table:`, and wrap everything in a block comment directly
following the `CREATE` statement. The Rajkumar et al paper wrapped the
example rows and `SELECT` statement in a block comment as well anyway.
Thoughts @fpingham?
This class enables us to send a dictionary containing an output key and
the expected format, which in turn allows us to retrieve the result of
the matching formats and extract specific information from it.
To exclude irrelevant information from our return dictionary, we can
prompt the LLM to use a specific command that notifies us when it
doesn't know the answer. We refer to this variable as the
"no_update_value".
Regarding the updated regular expression pattern
(r"{}:\s?([^.'\n']*).?"), it enables us to retrieve a format as 'Output
Key':'value'.
We have improved the regex by adding an optional space between ':' and
'value' with "s?", and by excluding points and line jumps from the
matches using "[^.'\n']*".
Provide shared memory capability for the Agent.
Inspired by #1293 .
## Problem
If both Agent and Tools (i.e., LLMChain) use the same memory, both of
them will save the context. It can be annoying in some cases.
## Solution
Create a memory wrapper that ignores the save and clear, thereby
preventing updates from Agent or Tools.
This PR adds
* `ZeroShotAgent.as_sql_agent`, which returns an agent for interacting
with a sql database. This builds off of `SQLDatabaseChain`. The main
advantages are 1) answering general questions about the db, 2) access to
a tool for double checking queries, and 3) recovering from errors
* `ZeroShotAgent.as_json_agent` which returns an agent for interacting
with json blobs.
* Several examples in notebooks
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Follow-up of @hinthornw's PR:
- Migrate the Tool abstraction to a separate file (`BaseTool`).
- `Tool` implementation of `BaseTool` takes in function and coroutine to
more easily maintain backwards compatibility
- Add a Toolkit abstraction that can own the generation of tools around
a shared concept or state
---------
Co-authored-by: William FH <13333726+hinthornw@users.noreply.github.com>
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Co-authored-by: Francisco Ingham <fpingham@gmail.com>
Co-authored-by: Dhruv Anand <105786647+dhruv-anand-aintech@users.noreply.github.com>
Co-authored-by: cragwolfe <cragcw@gmail.com>
Co-authored-by: Anton Troynikov <atroyn@users.noreply.github.com>
Co-authored-by: Oliver Klingefjord <oliver@klingefjord.com>
Co-authored-by: William Fu-Hinthorn <whinthorn@Williams-MBP-3.attlocal.net>
Co-authored-by: Bruno Bornsztein <bruno.bornsztein@gmail.com>
This approach has several advantages:
* it improves the readability of the code
* removes incompatibilities between SQL dialects
* fixes a bug with `datetime` values in rows and `ast.literal_eval`
Huge thanks and credits to @jzluo for finding the weaknesses in the
current approach and for the thoughtful discussion on the best way to
implement this.
---------
Co-authored-by: Francisco Ingham <>
Co-authored-by: Jon Luo <20971593+jzluo@users.noreply.github.com>
Currently the chain is getting the column names and types on the one
side and the example rows on the other. It is easier for the llm to read
the table information if the column name and examples are shown together
so that it can easily understand to which columns do the examples refer
to. For an instantiation of this, please refer to the changes in the
`sqlite.ipynb` notebook.
Also changed `eval` for `ast.literal_eval` when interpreting the results
from the sample row query since it is a better practice.
---------
Co-authored-by: Francisco Ingham <>
---------
Co-authored-by: Francisco Ingham <fpingham@gmail.com>
Supporting asyncio in langchain primitives allows for users to run them
concurrently and creates more seamless integration with
asyncio-supported frameworks (FastAPI, etc.)
Summary of changes:
**LLM**
* Add `agenerate` and `_agenerate`
* Implement in OpenAI by leveraging `client.Completions.acreate`
**Chain**
* Add `arun`, `acall`, `_acall`
* Implement them in `LLMChain` and `LLMMathChain` for now
**Agent**
* Refactor and leverage async chain and llm methods
* Add ability for `Tools` to contain async coroutine
* Implement async SerpaPI `arun`
Create demo notebook.
Open questions:
* Should all the async stuff go in separate classes? I've seen both
patterns (keeping the same class and having async and sync methods vs.
having class separation)
