Skip to content

Releases: Ontotext-AD/graphrag-eval

6.2.0

29 May 12:37
d759a1f

Choose a tag to compare

What's Changed

WKG-33:

  • Fix custom evaluation metrics
  • Upgrade litellm to 1.85.1
  • Make answer correctness evaluator get_correctness_dict async for consistency
  • Make custom evaluator evaluate async for consistency
  • Rename llm.py to llm_factory.py
  • Improve tests:
    • Fix instructions on how to run the tests
    • Rename all system tests to start with "test_*" for consistency
    • Read the configurations for all system tests in the test methods, not in the file globally

by @nelly-hateva in #81

Full Changelog: 6.1.0...6.2.0

6.1.0

29 Apr 12:43

Choose a tag to compare

What's Changed

  • Statnett-282: Allow matches of IRI discovery with SPARQL Query by @nelly-hateva in #74

Full Changelog: 6.0.0...6.1.0

6.0.0

28 Apr 06:40
98f5fd6

Choose a tag to compare

What's Changed

  • Statnett-282: Calculate steps score based on all reference steps, not just the ones in the last reference group & Evaluation of time series and IRI discovery steps by @nelly-hateva in #38
  • TTYG-160 Implement custom metric by @pgan002 in #46
  • TTYG-165: Upgrade 3rd-party libraries by @nelly-hateva in #47
  • TTYG-160 Fix custom evaluations reserved keys by @pgan002 in #52
  • TTYG-166 Configurable LLM by @pgan002 in #50
  • Statnett-358: Fix evaluation in case an actual sparql query is a DESCRIBE query, and the output is a string by @nelly-hateva in #57
  • Statnett-282: Bump vulnerable dependencies and release version 6.0.0 by @nelly-hateva in #73

Full Changelog: 5.3.1...6.0.0

5.3.1

28 Nov 10:44
4241c89

Choose a tag to compare

What's Changed

Full Changelog: 5.3.0...5.3.1

5.3.0

28 Nov 10:05

Choose a tag to compare

What's Changed

  • Statnett-278: Take into account the error samples, when calculating averages by @nelly-hateva in #33

Full Changelog: 5.2.0...5.3.0

5.2.0

21 Oct 14:16

Choose a tag to compare

Statnett-240: Compare SPARQL results with duplicated binding values

5.1.2

06 Oct 11:59

Choose a tag to compare

Bug fixes

Statnett-142: Cast micro and macro aggregation statistics to dict, so that the yaml serialization works as expected

5.1.1

06 Oct 06:58

Choose a tag to compare

Bug fixes

Statnett-142: Fix bug in the calculation of the aggregated results in case of an output from DESCRIBE or CONSTRUCT query containing the string "results"

5.1.0

30 Sep 09:56

Choose a tag to compare

Bug fixes

TTYG-126: Fix ragas errors caused by incompatibility of some libraries
TTYG-126: Rename openai extra and dependency group to ragas

5.0.2

26 Sep 16:21
75cdbbe

Choose a tag to compare

Bug fixes

TTYG-130: Package the prompts as part of the package