Releases: lion-agi/lionagi
v0.5.5
Modified
- updated branch_mixin for truly model agnostic experience
- now all non-openai models can invoke actions and run operations like openai models
Deprecated
- deprecated the ability to invoke action in
branch.communicate
function
Note:
it is recommended to use branch.operate
and go structured output when you need actions. You need to set actions
to True
What's Changed
- Update actions by @ohdearquant in #501
- Update action system by @ohdearquant in #502
Full Changelog: v0.5.4...v0.5.5
v0.5.4
Added:
strategies for operations,
for brainstorm and plan, now can choose among 4 different strategies to explore/exeucte
- concurrent
- sequential
- concurrent_sequential_chunk
- sequential_concurrent_chunk
What's Changed
- Add strategies by @ohdearquant in #500
Full Changelog: v0.5.3...v0.5.4
v0.5.3
Fixed:
- brainstorm function
Updated:
- various prompts,
What's Changed
- Updating prompts by @ohdearquant in #499
Full Changelog: v0.5.2...v0.5.3
v0.5.2
LION v0.5.2
🚀 Major Enhancement
- New ID System: Complete overhaul from string-based IDs to typed UUID implementation
- Significantly faster ID validation (type check vs string format check)
- More robust ID handling across operations
- Better type safety with dedicated IDType wrapper
⚠️ Breaking Changes
- Previous LION ID format is not compatible with this version
- Old serialized data will need to be migrated to new UUID format
🛠 Technical Improvements
- Updated core components to use new ID system
- Enhanced timestamp validation
- More efficient collection operations with typed IDs
📦 Installation
pip install lionagi
What's Changed
- Update ID system by @ohdearquant in #481
Full Changelog: v0.5.1...v0.5.2
v0.5.1
LION v0.5.1
🚀 What's New
- Added support for model-specific role handling via
allowed_roles
property - Improved system message handling for providers without system role support
- Enhanced structured output with
response_format
parameter - Added conversation clearing with
clear_messages()
🛠 Improvements
- Added
allowed_roles
to all model services (Anthropic, Groq, OpenAI, etc.) - Fixed API key handling and environment variable names
- Enhanced error messages and parameter validation
- Improved documentation with new examples and workflows
📦 Installation
uv pip install lionagi
What's Changed
- Update service by @ohdearquant in #480
Full Changelog: v0.5.0...v0.5.1
v0.5.0
quite a change, tutorials will be arriving daily.
What's Changed
- major updates, completely not backward compatible - announcing making of lion2 by @ohdearquant in #467
- V5 patch1 - code clean up by @ohdearquant in #469
- V5 patch2 debug lib by @ohdearquant in #470
- Updating service by @ohdearquant in #471
- feat: implement strategy executors for concurrent and sequential proc… by @ohdearquant in #472
- Adding strategy - patch 2 by @ohdearquant in #473
- Debugging protocols by @ohdearquant in #474
- Check protocols debug libs by @ohdearquant in #475
- Update messages by @ohdearquant in #476
- Updates by @ohdearquant in #477
- Updates by @ohdearquant in #478
- updated by @ohdearquant in #479
Full Changelog: v0.4.0...v0.5.0
v0.4.0
merged commits
What's Changed
- v0.2.3 Reorgaization by @ohdearquant in #403
- Updating libs by @ohdearquant in #404
- Updating libs by @ohdearquant in #405
- added token calculator by @ohdearquant in #406
- 0.3.0 dev by @ohdearquant in #407
- Updating libs by @ohdearquant in #409
- settled on v0.3 structure by @ohdearquant in #411
- organizing folders by @ohdearquant in #412
- Update GitHub workflows for CI/CD by @ohdearquant in #414
- updated converter by @ohdearquant in #420
- Dev imodel by @ohdearquant in #438
- Service by @ohdearquant in #439
- Switching to lionfuncs by @ohdearquant in #452
- Changing base by @ohdearquant in #458
- Dev ops4 by @ohdearquant in #457
- Dev ops by @ohdearquant in #455
- Bump to 4 by @ohdearquant in #453
- dev operations by @ohdearquant in #451
- v0.4.0 nightly by @ohdearquant in #441
- Dev expert by @ohdearquant in #437
- Service Test Version by @ohdearquant in #421
- V0.3 dev by @ohdearquant in #460
- Transition to core by @ohdearquant in #459
Full Changelog: v0.3.8...v0.4.0
v0.3.8
Release Note (v0.3.8)
🎉 New Features
NoteBook: https://github.com/lion-agi/lionagi/blob/main/notebooks/test_agi_ops.ipynb
Intelligent Operations Suite
-
brainstorm
: Generate and evaluate multiple solution approachesideas = await brainstorm( num_steps=3, instruction="Solve user authentication", reason=True )
-
rank
: Multi-agent ranking system with consensus buildingrankings = await rank( choices=candidates, num_scorers=10, )
-
score
: Flexible scoring system with customizable parametersresult = await score( instruction="Evaluate code quality", score_range=(1, 100), precision=2 )
-
select
: Intelligent selection with fuzzy matchingselected = await select( choices=options, max_num_selections=2, return_enum=True )
🔨 Improvements
Core Enhancements
- Enhanced operation modularity and customization
- Improved error handling and debugging
- Better type safety and validation
- More reliable dependency management
Architecture
- Cleaner module organization
- Better separation of concerns
- More efficient resource usage
- Enhanced testing infrastructure
📝 Breaking Changes
- Tutorial data now requires separate download
- Updated core dependency requirements
- Modified some API signatures for better type safety
What's Changed
- Dev ops3 by @ohdearquant in #456
Full Changelog: v0.3.7...v0.3.8
v0.3.7
Modified
changed id to use core id system,
BREAKING CHANGE
the format for timestamp have changed to be standard iso string, replacing previous custom version
What's Changed
- Switching to lionfuncs by @ohdearquant in #449
- Updating using lionfuncs by @ohdearquant in #454
Full Changelog: v0.3.6...v0.3.7
v0.3.6
Added:
various models for direct
, such as ReasonModel
, ActionModel
...
new implementation of select
, now can also return enum value, will replace existing select function in direct
interface in future versions. probably v0.4.0
What's Changed
- added new select by @ohdearquant in #450
Full Changelog: v0.3.5...v0.3.6