Releases: Sagacify/saga-llm-evaluation
Releases · Sagacify/saga-llm-evaluation
v0.10.0
Feature
- pyproject: Remove bleurt from dependencies, add it to readme (
49ffac0
)
v0.9.0
Feature
- cd: Add token pypi to poetry in cd (
0879230
)
v0.8.0
Feature
- Make publishing public to PyPI (
64fbb80
)
v0.7.2
Fix
- cd: Add password for private repo (
c509d8a
)
- cd: Add username cred for private repo (
6e02408
)
- pyproject: Re-add private repo to poetry (
7360b4a
)
v0.7.1
Fix
- cd: Update credentials for private repo (
d2832a3
)
v0.7.0
Feature
- project: Llama model can be now load from local, minor fix, improve doc (
6064ffc
)
- project: Redesigned scorer, llm_metrics support lists as input, scorer test skipped (
99932ea
)
- llm_metric: For each llm_metric, evaluation model can be passed as input (
d01ca44
)
Fix
- llm_metrics: Unify default value for models (
c467e28
)
Documentation
- language_metrics: Adjust doc for q_squared compute function (
8fa7e97
)
v0.6.0
Feature
- add_aspects: Add aspects to geval and gptscore small changes in arguments (
d76676d
)
- scorer: Add scorer class support (
c457d31
)
v0.5.0
Feature
- llm_metrics: Refactor gptscore and add support for geval and selfcheck (
980acdb
)
- gptscore: Revamp to make multiple predictions at once (
ce999f8
)
- gptscore: Add gptscore support (
09d2cba
)
Fix
v0.4.0
Feature
- language_metrics: Add q squared (
329b7bd
)
Fix
- versions: Update versions of tensorflow back (
a25acd7
)
v0.3.0
Feature
- metadata: Add support for computing simple metadata (
1f12800
)
Fix
- ci: Downgraded elemeta to 1.0.7 (
c8c872c
)
- ci: Attempt for fixing ci (
74b2078
)