Commit Graph

50 Commits

Author SHA1 Message Date
Deepak S V
5cd12102be
Improving Resilience of MRKL Agent (#5014)
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>
2023-05-22 11:08:08 -07:00
Harrison Chase
88a3a56c1a
Add Spark SQL support (#4602) (#4956)
# Add Spark SQL support 
* Add Spark SQL support. It can connect to Spark via building a
local/remote SparkSession.
* Include a notebook example

I tried some complicated queries (window function, table joins), and the
tool works well.
Compared to the [Spark Dataframe

agent](https://python.langchain.com/en/latest/modules/agents/toolkits/examples/spark.html),
this tool is able to generate queries across multiple tables.

---------

# Your PR Title (What it does)

<!--
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 # (issue)

## 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: Gengliang Wang <gengliang@apache.org>
Co-authored-by: Mike W <62768671+skcoirz@users.noreply.github.com>
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Co-authored-by: UmerHA <40663591+UmerHA@users.noreply.github.com>
Co-authored-by: 张城铭 <z@hyperf.io>
Co-authored-by: assert <zhangchengming@kkguan.com>
Co-authored-by: blob42 <spike@w530>
Co-authored-by: Yuekai Zhang <zhangyuekai@foxmail.com>
Co-authored-by: Richard He <he.yucheng@outlook.com>
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
Co-authored-by: Leonid Ganeline <leo.gan.57@gmail.com>
Co-authored-by: Alexey Nominas <60900649+Chae4ek@users.noreply.github.com>
Co-authored-by: elBarkey <elbarkey@gmail.com>
Co-authored-by: Davis Chase <130488702+dev2049@users.noreply.github.com>
Co-authored-by: Jeffrey D <1289344+verygoodsoftwarenotvirus@users.noreply.github.com>
Co-authored-by: so2liu <yangliu35@outlook.com>
Co-authored-by: Viswanadh Rayavarapu <44315599+vishwa-rn@users.noreply.github.com>
Co-authored-by: Chakib Ben Ziane <contact@blob42.xyz>
Co-authored-by: Daniel Chalef <131175+danielchalef@users.noreply.github.com>
Co-authored-by: Daniel Chalef <daniel.chalef@private.org>
Co-authored-by: Jari Bakken <jari.bakken@gmail.com>
Co-authored-by: escafati <scafatieugenio@gmail.com>
2023-05-18 20:53:08 -07:00
Harrison Chase
fbfa49f2c1
agent serialization (#4642) 2023-05-13 21:47:10 -07:00
Zander Chase
d969f43ed8
Load HuggingFace Tool (#4475)
# Add option to `load_huggingface_tool`

Expose a method to load a huggingface Tool from the HF hub

---------

Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
2023-05-11 00:07:36 -07:00
Davis Chase
04475bea7d
Mv plan and execute to experimental (#4459) 2023-05-10 08:31:53 -07:00
Harrison Chase
6b8d144ccc
Harrison/plan and solve (#4422) 2023-05-09 21:07:56 -07:00
Zander Chase
35c9e6ab40
Pass Callbacks through load_tools (#4298)
- Update the load_tools method to properly accept `callbacks` arguments.
- Add a deprecation warning when `callback_manager` is passed
- Add two unit tests to check the deprecation warning is raised and to
confirm the callback is passed through.

Closes issue #4096
2023-05-08 08:44:26 -07:00
Zander Chase
2f087d63af
Fix Python RePL Tool (#4137)
Filter out kwargs from inferred schema when determining if a tool is
single input.

Add a couple unit tests.

Move tool unit tests to the tools dir
2023-05-04 20:31:16 -07:00
Mike Wang
67db495fcf
[agent] Add Spark Agent (#4020)
- added support for spark through pyspark library.
- added jupyter notebook as example.
2023-05-03 22:45:23 -07:00
Zander Chase
afa9d1292b
Re-Permit Partials in Tool (#4058)
Resolved issue #4053

Now that StructuredTool is a separate class, this constraint is no
longer needed.

Added/updated a unit test
2023-05-03 13:16:41 -07:00
Zander Chase
c582f2e9e3
Add Structure Chat Agent (#3912)
Create a new chat agent that is compatible with the Multi-input tools
2023-05-01 20:34:50 -07:00
Zander Chase
9b9b231e10
Update some Tools Docs (#3913)
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
2023-05-01 19:07:26 -07:00
Zander Chase
84ea17b786
Move Tool Validation (#3923)
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.
2023-05-01 18:44:24 -07:00
Ankush Gola
d3ec00b566
Callbacks Refactor [base] (#3256)
Co-authored-by: Nuno Campos <nuno@boringbits.io>
Co-authored-by: Davis Chase <130488702+dev2049@users.noreply.github.com>
Co-authored-by: Zander Chase <130414180+vowelparrot@users.noreply.github.com>
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
2023-04-30 11:14:09 -07:00
Mike Wang
512c24fc9c
[annotation improvement] Make AgentType->Class Conversion More Scalable (#3749)
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.~~
2023-04-28 21:17:28 -07:00
Zander Chase
da7b51455c
Dynamic tool -> single purpose (#3697)
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
2023-04-28 09:38:41 -07:00
Zander Chase
4654c58f72
Add validation on agent instantiation for multi-input tools (#3681)
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>
2023-04-27 15:36:11 -07:00
Mindaugas Sharskus
a4d85f7fd5
[Fix #3365]: Changed regex to cover new line before action serious (#3367)
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>
2023-04-24 22:05:31 -07:00
Zander Chase
49122a96e7
Structured Tool Bugfixes (#3324)
- 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
2023-04-24 09:58:29 -07:00
Zander Chase
90ef705ced
Update Tool Input (#3103)
- 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>
2023-04-18 18:18:33 -07:00
Harrison Chase
aad0a498ac
Harrison/output error (#3094)
Co-authored-by: yummydum <sumita@nowcast.co.jp>
2023-04-18 08:59:56 -07:00
Harrison Chase
db968284f8
tools refactor (#2961)
Co-authored-by: vowelparrot <130414180+vowelparrot@users.noreply.github.com>
2023-04-17 21:35:29 -07:00
Harrison Chase
e12e00df12
use output parsers in agents (#2987) 2023-04-16 13:15:21 -07:00
Harrison Chase
705596b46a
Harrison/fix create sql agent (#2870)
Co-authored-by: Timothé Pearce <timothe.pearce@gmail.com>
2023-04-13 22:07:58 -07:00
Zach Jones
13d1df2140
Feature: AgentExecutor execution time limit (#2399)
`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>
2023-04-06 12:54:32 -07:00
leo-gan
fd69cc7e42
Removed duplicate BaseModel dependencies (#2471)
Removed duplicate BaseModel dependencies in class inheritances.
Also, sorted imports by `isort`.
2023-04-06 12:45:16 -07:00
Harrison Chase
c7b083ab56
bump version to 131 (#2391) 2023-04-04 07:21:50 -07:00
Shrined
10dab053b4
Add Enum for agent types (#2321)
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.
2023-04-03 21:56:20 -07:00
Harrison Chase
5c907d9998
Harrison/base agent without docs (#2166) 2023-03-29 22:11:25 -07:00
Harrison Chase
f5a4bf0ce4
remove prep (#2136)
agents should be stateless or async stuff may not work
2023-03-29 14:38:21 -07:00
Harrison Chase
9e74df2404
Fix issue#1645: Parse llm_output even there's newline (#2092) (#2099)
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>
2023-03-28 08:14:09 -07:00
b7f392fdd6
[agent_executor] convenience func: lookup tool by name (#2001)
A quick convenience function to lookup a tool by name

Co-authored-by: blob42 <spike@w530>
2023-03-27 23:10:34 -07:00
Harrison Chase
aed9f9febe
Harrison/return intermediate (#1633)
Co-authored-by: Mario Kostelac <mario@intercom.io>
2023-03-13 07:54:29 -07:00
Harrison Chase
b7708bbec6
rfc: callback changes (#1165)
conceptually, no reason a tool should know what an "agent action" is

unless any objections, can change in all callback handlers
2023-02-20 22:54:15 -08:00
Ankush Gola
7b5e160d28
Make Tools own model, add ToolKit Concept (#1095)
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>
2023-02-18 13:40:43 -08:00
Jason Liu
54f9e4287f
Pass kwargs from initialize_agent into agent classmethod (#799)
# Problem
I noticed that in order to change the prefix of the prompt in the
`zero-shot-react-description` agent
we had to dig around to subset strings deep into the agent's attributes.
It requires the user to inspect a long chain of attributes and classes.

`initialize_agent -> AgentExecutor -> Agent -> LLMChain -> Prompt from
Agent.create_prompt`

``` python
agent = initialize_agent(
    tools=tools,
    llm=fake_llm,
    agent="zero-shot-react-description"
)
prompt_str = agent.agent.llm_chain.prompt.template
new_prompt_str = change_prefix(prompt_str)
agent.agent.llm_chain.prompt.template = new_prompt_str
```

# Implemented Solution

`initialize_agent` accepts `**kwargs` but passes it to `AgentExecutor`
but not `ZeroShotAgent`, by simply giving the kwargs to the agent class
methods we can support changing the prefix and suffix for one agent
while allowing future agents to take advantage of `initialize_agent`.


```
agent = initialize_agent(
    tools=tools,
    llm=fake_llm,
    agent="zero-shot-react-description",
    agent_kwargs={"prefix": prefix, "suffix": suffix}
)
```

To be fair, this was before finding docs around custom agents here:
https://langchain.readthedocs.io/en/latest/modules/agents/examples/custom_agent.html?highlight=custom%20#custom-llmchain
but i find that my use case just needed to change the prefix a little.


# Changes

* Pass kwargs to Agent class method
* Added a test to check suffix and prefix

---------

Co-authored-by: Jason Liu <jason@jxnl.coA>
2023-01-30 14:54:09 -08:00
Harrison Chase
1ad7973cc6
Harrison/tool decorator (#790)
Co-authored-by: Jason Liu <jxnl@users.noreply.github.com>
Co-authored-by: Jason Liu <jason@jxnl.coA>
2023-01-28 18:26:24 -08:00
Amos Ng
6ad360bdef
Suggestions for better debugging (#765)
Please feel free to disregard any changes you disagree with
2023-01-28 08:05:20 -08:00
Harrison Chase
a2eeaf3d43
strip whitespace (#680) 2023-01-21 16:03:48 -08:00
Harrison Chase
1ac3319e45
simplify parsing of the final answer (#621) 2023-01-15 16:39:27 -08:00
Harrison Chase
4974f49bb7
add return_direct flag to tool (#537)
adds a return_direct flag to tools, which just returns the tool output
as the final output
2023-01-06 06:40:32 -08:00
Harrison Chase
9e04c34e20
Add BaseCallbackHandler and CallbackManager (#478)
Co-authored-by: Ankush Gola <9536492+agola11@users.noreply.github.com>
2023-01-04 07:54:25 -08:00
Harrison Chase
d0f194de73
add logic for agent stopping (#420) 2022-12-29 08:21:11 -05:00
Harrison Chase
ee3b8e89b3
better parsing of agent output (#418) 2022-12-25 09:53:36 -05:00
Harrison Chase
cf98f219f9
Harrison/tools exp (#372) 2022-12-18 21:51:23 -05:00
Harrison Chase
78b31e5966
Harrison/cache (#343) 2022-12-15 07:53:32 -08:00
Harrison Chase
9bb7195085
Harrison/llm saving (#331)
Co-authored-by: Akash Samant <70665700+asamant21@users.noreply.github.com>
2022-12-13 06:46:01 -08:00
John McDonnell
68666d6a22
Gracefully degrade when model asks for nonexistent tool (#268)
Not yet tested, but very simple change, assumption is that we're cool
with just producing a generic output when tool is not found
2022-12-06 21:52:48 -08:00
Harrison Chase
6eab5254e5
add docs for custom agents (#196) 2022-11-26 06:03:08 -08:00
Harrison Chase
d3a7429f61
(WIP) agents (#171) 2022-11-22 06:16:26 -08:00