Module: Ask::Eval
- Defined in:
- lib/ask/eval.rb,
lib/ask/eval/dsl.rb,
lib/ask/eval/judge.rb,
lib/ask/eval/runner.rb,
lib/ask/eval/version.rb,
lib/ask/eval/recorder.rb,
lib/ask/eval/test_case.rb,
lib/ask/eval/assertions.rb,
lib/ask/eval/judges/bias.rb,
lib/ask/eval/cost_tracker.rb,
lib/ask/eval/session_eval.rb,
lib/ask/eval/configuration.rb,
lib/ask/eval/judges/faithful.rb,
lib/ask/eval/judges/toxicity.rb,
lib/ask/eval/reporters/junit.rb,
lib/ask/eval/assertions/judge.rb,
lib/ask/eval/reporters/github.rb,
lib/ask/eval/reporters/console.rb,
lib/ask/eval/judges/correctness.rb,
lib/ask/eval/judges/hallucination.rb,
lib/ask/eval/assertions/deterministic.rb
Defined Under Namespace
Modules: Assertions, DSL, Judges, Reporters Classes: AssertionError, Configuration, CostTracker, Error, Judge, Recorder, Runner, SessionEval, TestCase
Constant Summary collapse
- VERSION =
"0.2.0"
Class Method Summary collapse
-
.configuration ⇒ Configuration
The global configuration.
-
.configure {|Configuration| ... } ⇒ Object
Configure ask-eval globally.
-
.cost_report ⇒ Hash
Cost report from all evaluations.
-
.evaluate(output, assertions, context: nil, input: nil, expected: nil) ⇒ Array<Hash>
Run a batch evaluation.
Class Method Details
.configuration ⇒ Configuration
Returns the global configuration.
57 58 59 |
# File 'lib/ask/eval.rb', line 57 def configuration @configuration ||= Configuration.new end |
.configure {|Configuration| ... } ⇒ Object
Configure ask-eval globally.
52 53 54 |
# File 'lib/ask/eval.rb', line 52 def configure yield configuration end |
.cost_report ⇒ Hash
Returns cost report from all evaluations.
62 63 64 |
# File 'lib/ask/eval.rb', line 62 def cost_report configuration.cost_report end |
.evaluate(output, assertions, context: nil, input: nil, expected: nil) ⇒ Array<Hash>
Run a batch evaluation.
40 41 42 43 44 45 46 47 48 |
# File 'lib/ask/eval.rb', line 40 def evaluate(output, assertions, context: nil, input: nil, expected: nil) test_case = TestCase.new( actual_output: output, input: input, context: context, expected_output: expected ) Assertions.evaluate_all(test_case, assertions) end |