Recent Releases of swarms
swarms - 2.5.0
Swarms Version 2.5.0 Changelog
Commits on Dec 1, 2023
Fix: executefuturesdict
Commit ID: 452bf7d Fix: executefuturesdict
Commit ID: 1c280b1 Cleanup
Commit ID: 64ecedb AbstractLLM
Commit ID: a0a0128 Update basemultimodalmodel.py
Commit ID: 061db36 Feature: Readme
Commit ID: b726f04 Feature: Docs for AbstractLLM, Docs for BaseMultiModalModel
Commit ID: a7a6d54 Feature: AutoScaler Prototype, Tests
Commit ID: 3ae305e Feature: Agent prompt cleanup
Commit ID: 84a8449 Commits on Nov 30, 2023
Cleanup
Commit ID: 0555a2b AbstractLLM
Commit ID: d8fb6f8 Update stable_diffusion.py
Commit ID: e09001f Create idea2img.py
Commit ID: 8213769 Create idea2img.py
Commit ID: b382d6b Create idea2img.py
Commit ID: 9eabc91 Cleanup
Commit ID: a8449d2 Feature: AbstractLLM
Commit ID: a37b314 Code Quality
Commit ID: ec51149 Readme
Commit ID: 4e47528 Feature: SequentialWorkflow
Commit ID: 6e42d54 Feature: Scripts for pytest, SequentialWorkflow
Commit ID: a47f3a7 Commits on Nov 29, 2023
Cleanup: Multi agent debate
Commit ID: 2b6b050 Cleanup: Tests cleanup
Commit ID: 4275efa Cleanup
Commit ID: b1bf3ef Cleanup
Commit ID: d08a2cb Cleanup: Non Existent tests
Commit ID: 45c6b11 Feature: SequentialWorkflow
Commit ID: a679d01 Create logistics.py
Commit ID: 5de20b3 Upload files
Commit ID: 0d8f657 Create logistics.py
Commit ID: b027c22 Create logistics.py
Commit ID: dacc472 Update pip
Commit ID: 0ff1aa0 Code Quality
Commit ID: dc55006 Merge pull request: Browse the repository
Commit ID: 87e8b90
- Python
Published by kyegomez about 2 years ago
swarms - 2.4.2
2.4.2 Release
75f3cf1: [FEAT][Flow.run() img = None for conditional img inputs, BaseMultiMod... a92a6a5: FEAT: [BEAUTIFY in GPT4Vision][Disable logging in init of swarms] f895497: Jarvis demo, base multimmodalmodel, whisperx -> whisperxmodel 51c82cf: Commits on Nov 24, 2023 gpt4vision features and api 9390efb: cleanup b9fea7b: gpt4v docs 66d9f70: GPT4Vision + multimodal dfa4197: Merge pull request #190 from elder-plinius/master 539bfea: Add files via upload 97a79ac: Create assembly.py a4348a8: clean up outputs of multi modal autonomous agents 9e6c427: tests for gpt4visionapi 0802091: gpt4vision api 9c3a292: dockerfile 399099e: Merge pull request #181 from evelynmitchell/master 55a1dfa: nougat fix ee1ac00: clean up d88a31a: positivemed demo; b554d17: Merge pull request #184 from elder-plinius/master 179ca14: Merge branch 'kyegomez:master' into master 673b8ac: prompt name fixes for positive med 29b6c9b: Update autotempexample.py ad3e8e7: Update and rename bloggenexample.py to bloggenexample.py 66b9ba0: Rename bloggen to bloggen.py cd82fa6: Update and rename bloggen.py to blog_gen 14570d1: positive med ed071e0: renamed tests to allow for pytest autodiscovery a229685: black formatting d1144ec: Commits on Nov 23, 2023 requirements.txt clean up 36a641b: requirements.txt 9f20592: code quality fixes: line length = 80 49c7b97: torch verisoning d97de1c: multi agent docs + playground code quality
- Python
Published by kyegomez about 2 years ago
swarms - 2.3.1
Changelog for version 2.3.1:
Commit on Nov 14, 2023: kyegomez committed b2dd2ef - tests for yi, stable diffusion, timm models, etc kyegomez committed dfea671 - no api key kyegomez committed 720a728 - mistral caller, openai version 2.8, llama function caller, tests for ... kyegomez committed 699c943 - mistral caller, openai version 2.8, llama function caller, tests for ... kyegomez committed 5f56023 Commits on Nov 13, 2023: kyegomez committed 7db6930 - clean kyegomez committed 0d0e5e5 - Merge pull request #129 from kyegomez/dependabot/githubactions/codac... kyegomez committed 694100e - Merge pull request #131 from kyegomez/dependabot/githubactions/actio... kyegomez committed 39ba632 - Merge pull request #130 from kyegomez/dependabot/githubactions/sigst... kyegomez committed 2b05329 - Merge pull request #132 from kyegomez/dependabot/githubactions/actio... kyegomez committed b2b9665 - Bump actions/checkout from 3 to 4 dependabot[bot] committed 1203d84 - Bump actions/setup-python from 3 to 4 dependabot[bot] committed 1cd0779 - Bump sigstore/cosign-installer from 3.1.1 to 3.2.0 dependabot[bot] committed 1046e8a - Bump codacy/codacy-analysis-cli-action from 1.1.0 to 4.3.0
- Python
Published by kyegomez over 2 years ago
swarms - 2.1.9
Changelog for version 2.1.9 - Commits on Nov 11, 2023
CLEANUP: swarms.agents, removed unused files [07bcd22] 50+ tests for cohere [bdc7337] Tests clean up, import file paths, and env PSG connection strings [4197920] Environment setup [0c6daf7] Tests for groupchats, anthropic [9402dab] New version [76fd9a0] Tests, workflow fixes, torch version [2e6efb4] Release workflow [e8e024f] Created codacy.yml [d2b5fc9] Created codeql.yml [6dcaab6] Created docker-publish.yml [1d01402] Created python-app.yml [aae6c03] Created docker-image.yml [fdcfa0c] Tools for flow and general cleanup [0e335b1] Commits on Nov 10, 2023
2.1.7, no langchain experimental [a610ff7] dalle3 production grade ready [41e5f17] No schemas [d26531f] Commits on Nov 9, 2023
Dockerfile [4596ddc] Dockerfile running [991979d] Dynamic max loops, GPT4 clean up [371da79]
- Python
Published by kyegomez over 2 years ago
swarms - 2.1.7
Changelog:
[2e6efb4] Release workflow [e8e024f] Create codacy.yml [d2b5fc9] Create codeql.yml [6dcaab6] Create docker-publish.yml [1d01402] Create python-app.yml [aae6c03] Create docker-image.yml [fdcfa0c] Tools for flow and general cleanup [0e335b1] Commits on Nov 10, 2023 - 2.1.7, no langchain experimental [a610ff7] Dalle3 production grade ready [41e5f17] No schemas [d26531f] Commits on Nov 9, 2023 - Dockerfile [4596ddc] Dockerfile running [991979d] Dynamic max loops, + gpt4 clean up [371da79] Account swarm + layout document fix [1dc1e8f] Commits on Nov 8, 2023 - Clean up [fe19f21] Yapf code quality [2e7905d] Code quality + new version + fuyu fixes [e0c712b] Update zephyr.py [73b549b] Fuyu + zephyr fixes [9aa4084] Anthropic + kosmo2 + fastvit [6b4c2d4] Anthropic working [f74a4da] Accountant swarm + autotemp agent [11b02a5] Commits on Nov 7, 2023 - Security [3d186fd] Create CODEOFCONDUCT.md [f16515a] Create SECURITY.md [570ed6a] 2.0.5 [7a5e82b] Clean up unused code [bb496f4] Version [7bc6b9c] Docs for misral and groupchat [4f68c54] Merge pull request #100 from vyomakesh09/master
- Python
Published by kyegomez over 2 years ago
swarms - 2.0.5
Changelog for Release 2.0.5:
Nov 7, 2023:
- Cleaned up unused code. [commit: 7a5e82b]
- Removed useless code, no more worker, etc. [commit: bb496f4]
- Added version. [commit: 7bc6b9c]
- Added docs for Misral and Groupchat. [commit: 55a4e9a]
- Merged pull request #100 from vyomakesh09/master. [commit: 4f68c54]
- Fixed typo in bfloat16. [commit: b61fcf3]
Nov 6, 2023:
- Removed open interpreter, cleaned up docs, added add messages to flow. [commit: 16176e8]
- Updated swarms. [commit: 62a4135]
- Cleaned up code. [commit: 8dc9064]
- Updated anthropic docs. [commit: 4c6bbad]
- Fixed saved state error in dashboard. [commit: 97aa8bc]
- Fixed playground and flow docs. [commit: 336bffe]
- Python
Published by kyegomez over 2 years ago
swarms - 2.0.2
Changelog:
Commits on Nov 6, 2023
swarms (62a4135) clean up (8dc9064) anthropic docs (4c6bbad) saved state in dashboard error (97aa8bc) playground + flow docs fix (336bffe) Dependencies clean up (a70a2b0) dalle3 (4fb38c1) docs for gpt4v (c7d128d) docs (b7aa21f) GPT4Vision + Dalle3 -> modules + tests + documentation (fd8919d) guides (fe48ec1) m (a97f759) docs clean up -> corporate folder (881ec11) multi modal auto agent + removed workflow.py (c520cda) swarms docs corporate + sequential workflow (6010c9d) sequential workflow (71da697) workflow states (383412b) Commits on Nov 5, 2023
sequential workflow docs (c94512d) sequential workflow tests, prototype with documentation (310230a) auto saved + fixed run method of flow (ba28f40) Commits on Nov 4, 2023
anthropic tests (d4bd4fa) fuyu fix (1fb1932) docs for DistilWhisperModel (6e6fe8d) tests for distilled whisperx (7e1d486) distilled whisperx (75ebbe0) flow -> example.py (2f31a63) groupchat (154f50c) Commits on Nov 3, 2023
flow example, save and load state (7d888c6) flow (f53236a) expanded permissions to allow welcome action run (4392b29) added labeler.yml (f190d84) no stream (1b0cb87) flow (1162271) Merge pull request #61 from kyegomez/dependabot/github_actions/action... (17ff736)
- Python
Published by kyegomez over 2 years ago
swarms - 1.1.41
Changelog
[Unreleased]
Added
- Added
CodeInterpreterToolclass to adhere to the providedToolabstract base class. This new tool has flexible and reliable error handling and supports both synchronous and asynchronous operations. - Enhanced
GooglePalmclass by making optional parameters for user inputs. This improves customization and flexibility. - Comprehensive unit tests for the
OmniWorkerAgentclass have been added. - Pytest fixtures and parameterization in unit tests to improve code reuse and readability.
- Improved code readability in the
MultiModalVisualAgentToolclass and added comprehensive unit tests. - Enhanced the
WorkerUltraNodeandWorkerUltraNodeInitializerclasses' test coverage by adding more test cases and edge cases. - A GitHub Actions workflow file has been added to automatically run pytest tests upon each
pushevent. This CI pipeline ensures the reliability of the codebase by preventing merging of breaking changes into the main branch.
Fixed
- Improved error handling for the
GooglePalmclass by incorporating exponential backoff retry logic to handle network and API-related issues. - Fixed missing prompt arguments by transforming message objects to the dictionary format in
GooglePalm. - Unexpected Argument Error in
WorkerNodehas been fixed. TheWorkerNodeclass was not expecting aworker_nameargument which was passed to it during the initialization of theHierarchicalSwarmclass. This bug was fixed by either removing theworker_nameargument from theWorkerNodeinitialization in theHierarchicalSwarmclass or by modifying the__init__method of theWorkerNodeclass to accept theworker_nameargument. - Fixed incorrect parameter type checking in
OmniWorkerAgentclass, methods now correctly validate types for required arguments. The methodchatalso now properly checks forapi_key,api_endpoint, andapi_type. - Updated
MultiModalVisualAgentToolclass tests, the_runmethod was not being properly tested, as it was not being invoked with correct arguments. - Fixed
WorkerUltraNodeclass to handle invalid initialization arguments properly. Added appropriate type checking and error handling for better robustness. - Improved error handling in
WorkerUltraNodeInitializerclass. Now, all methods have try-catch blocks to handle unexpected errors gracefully and provide useful error messages.
[1.1.2] - 2023-07-30
Added
- Initial creation of the
GooglePalmclass, a wrapper around the Google's PaLM Chat API. Toolabstract base class for creating tool-like objects with a consistent interface.CodeInterpreterSessionexample usage in both synchronous and asynchronous manners.
Improvements
- Better error messages and exception handling have been added to inform the user about issues more precisely. For example, included checks and corresponding error messages for invalid inputs for API keys, invalid objectives, failed vectorstore initialization, and so on.
- The documentation and descriptions for classes and methods have been improved, providing a clearer understanding of what each class and method does, and what arguments they accept.
- Improved the initialization of resources like
llm_classandtoolsin theWorkerNodeclass, making the code more robust and easy to understand.
- Python
Published by kyegomez over 2 years ago
swarms - 1.1.4
Changelog
[Unreleased]
Added
- Added
CodeInterpreterToolclass to adhere to the providedToolabstract base class. This new tool has flexible and reliable error handling and supports both synchronous and asynchronous operations. - Enhanced
GooglePalmclass by making optional parameters for user inputs. This improves customization and flexibility. - Comprehensive unit tests for the
OmniWorkerAgentclass have been added. - Pytest fixtures and parameterization in unit tests to improve code reuse and readability.
- Improved code readability in the
MultiModalVisualAgentToolclass and added comprehensive unit tests. - Enhanced the
WorkerUltraNodeandWorkerUltraNodeInitializerclasses' test coverage by adding more test cases and edge cases. - A GitHub Actions workflow file has been added to automatically run pytest tests upon each
pushevent. This CI pipeline ensures the reliability of the codebase by preventing merging of breaking changes into the main branch.
Fixed
- Improved error handling for the
GooglePalmclass by incorporating exponential backoff retry logic to handle network and API-related issues. - Fixed missing prompt arguments by transforming message objects to the dictionary format in
GooglePalm. - Unexpected Argument Error in
WorkerNodehas been fixed. TheWorkerNodeclass was not expecting aworker_nameargument which was passed to it during the initialization of theHierarchicalSwarmclass. This bug was fixed by either removing theworker_nameargument from theWorkerNodeinitialization in theHierarchicalSwarmclass or by modifying the__init__method of theWorkerNodeclass to accept theworker_nameargument. - Fixed incorrect parameter type checking in
OmniWorkerAgentclass, methods now correctly validate types for required arguments. The methodchatalso now properly checks forapi_key,api_endpoint, andapi_type. - Updated
MultiModalVisualAgentToolclass tests, the_runmethod was not being properly tested, as it was not being invoked with correct arguments. - Fixed
WorkerUltraNodeclass to handle invalid initialization arguments properly. Added appropriate type checking and error handling for better robustness. - Improved error handling in
WorkerUltraNodeInitializerclass. Now, all methods have try-catch blocks to handle unexpected errors gracefully and provide useful error messages.
[1.1.2] - 2023-07-30
Added
- Initial creation of the
GooglePalmclass, a wrapper around the Google's PaLM Chat API. Toolabstract base class for creating tool-like objects with a consistent interface.CodeInterpreterSessionexample usage in both synchronous and asynchronous manners.
Improvements
- Better error messages and exception handling have been added to inform the user about issues more precisely. For example, included checks and corresponding error messages for invalid inputs for API keys, invalid objectives, failed vectorstore initialization, and so on.
- The documentation and descriptions for classes and methods have been improved, providing a clearer understanding of what each class and method does, and what arguments they accept.
- Improved the initialization of resources like
llm_classandtoolsin theWorkerNodeclass, making the code more robust and easy to understand.
- Python
Published by kyegomez over 2 years ago
swarms - 0.8.4
Changelog
Bugs
- The flash attention module was missing in the original codebase. This resulted in a module not found error during the execution.
- Incorrect implementation of flash attention integration with the main attention module. The
forwardmethod in theAttendclass wasn't correctly handling flash attention. - The
flash_attnfunction within theAttendclass had incorrect assumptions about the dimensions of thekandvtensors. This led to dimension mismatch errors during the tensor operations. - The original
flash_attnmethod was not handling the scale correctly whenqk_normwas set toTrue.
Improvements
- Integrated the flash attention module into the main codebase and ensured the dimensions and operations are correct.
- Modified the
forwardmethod in theAttendclass to handle flash attention correctly. It checks whether flash attention is enabled and accordingly calls the correct attention method. - Adjusted the
flash_attnmethod to account for possible missing dimensions inq,k, andvtensors, and to correct for possible dimension mismatches. - Included a check to determine if the tensor is on a CUDA device and if so, to leverage the appropriate CUDA configuration for efficient attention.
- Correctly handled the scale in the
flash_attnmethod whenqk_normwasTrue. - Added assertions and informative error messages for incompatible options such as 'talking heads' and 'flash attention'.
- Ensured compatibility with PyTorch version 2.0 and above for using flash attention.
- Python
Published by kyegomez over 2 years ago
swarms - 0.8.1
Bug Fixes
Issue: Lack of error handling and logging. Root Cause: The initial code did not have any error handling or logging mechanisms in place. This could lead to silent failures or uninformative error messages when something goes wrong. Resolution: Added extensive error handling and logging throughout the code. Each method now checks for required parameters and raises a ValueError with a descriptive message if a required parameter is missing. Additionally, try/except blocks have been added around the main logic of each method to catch any exceptions that might occur, log an error message, and re-raise the exception.
Issue: Lack of input validation. Root Cause: The initial code did not validate the inputs to the methods. This could lead to unexpected behavior or errors if invalid inputs are provided. Resolution: Added checks at the beginning of each method to validate the inputs. If an input is invalid (e.g., None when it should not be), an error message is logged and a ValueError or TypeError is raised.
Features Implemented
Feature: Modularization of parameters. Description: Made all meaningful parameters in the createagent and createworker_node methods optional with default values. This allows the user to customize the agent and worker node as needed without having to provide all parameters.
Feature: Documentation. Description: Added docstrings to each method to explain what they do. This will help users understand how to use the methods and what to expect from them.
Feature: Improved logging. Description: Configured the logging module to log messages with the timestamp, log level, and message. This will make the logs more informative and easier to read.
- Python
Published by kyegomez over 2 years ago
swarms - 0.8.0
Changelog
Created
LLMclass which serves as a wrapper for OpenAI's ChatGPT and Hugging Face's language models.Added optional parameters
openai_api_key,hf_repo_id,hf_api_token, andmodel_kwargsto theLLMclass.Introduced logic to switch between OpenAI's ChatGPT and Hugging Face's language models based on the provided optional parameters.
Improved error handling in the
LLMclass. The class now raises aValueErrorif neitheropenai_api_keynorhf_repo_idare provided. It also checks for the presence ofHUGGINGFACEHUB_API_TOKENin the environment variables.Improved modularity in the
LLMclass. Added the ability for the user to provide their own model parameters through themodel_kwargsparameter.Updated
LLMclass to be shapeless, allowing users to provide their inputs in any shape they wish.Updated the error handling system to provide detailed error messages and log exceptions.
Created a comprehensive suite of unit tests for the
LLMclass to ensure robustness and reliability of the code.Tests cover initializations, proper usage of underlying models, error handling for missing parameters, and environment variable checks.
Please note that this changelog reflects the changes made during our interactions and your instructions. The versioning is arbitrary and for the purpose of this demonstration only. In a real-world scenario, versioning would likely be more granular and follow semantic versioning principles.
- Python
Published by kyegomez over 2 years ago
swarms - 0.6.4
Bug:
TypeError: BossNode.__init__() got multiple values for argument 'max_iterations'- Root Cause: This error was caused because the
BossNodeclass was incorrectly trying to assign a value to themax_iterationsparameter in theBabyAGIinitialization. - Resolution: Commented out the
max_iterationsparameter in theBabyAGIinitialization inside theBossNodeclass.
- Root Cause: This error was caused because the
Bug: The same
TypeErrorpersisted.- Root Cause: The
max_iterationsparameter was being passed to theBossNodeinitialization ininitialize_boss_nodefunction ofSwarmsclass but it was not handled in theBossNodeclass. - Resolution: Reintroduced the
max_iterationsparameter in theBossNodeclass, this time to be used as an attribute of theBossNodeitself and not in theBabyAGIinitialization.
- Root Cause: The
Bug:
AttributeError: 'BossNode' object has no attribute 'create_task'- Root Cause: This error was caused because
run_swarmsmethod inSwarmsclass was trying to call acreate_taskmethod on aBossNodeobject, which was not defined in theBossNodeclass. - Resolution: Added a
create_taskmethod inBossNodeclass that returns a dictionary containing the objective.
- Root Cause: This error was caused because
- Python
Published by kyegomez over 2 years ago