Closed devmotion closed 2 years ago
Merging #94 (019bb59) into master (4994a79) will decrease coverage by
0.25%
. The diff coverage is95.00%
.
@@ Coverage Diff @@
## master #94 +/- ##
==========================================
- Coverage 97.74% 97.48% -0.26%
==========================================
Files 7 7
Lines 222 239 +17
==========================================
+ Hits 217 233 +16
- Misses 5 6 +1
Impacted Files | Coverage Δ | |
---|---|---|
src/AbstractMCMC.jl | 100.00% <ø> (ø) |
|
src/sample.jl | 97.66% <95.00%> (-0.40%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 4994a79...019bb59. Read the comment docs.
The MCMCChains errors are unrelated and known.
https://github.com/TuringLang/AbstractMCMC.jl/pull/99 should be merged first since it is non-breaking.
Implements the suggestion in https://github.com/TuringLang/AbstractMCMC.jl/issues/92#issuecomment-1016745828.
With this PR, samplers that support
init_params
for single chain sampling (such as e.g. AdvancedMH, EllipticalSliceSampling or Turing) get support forinit_params
with ensemble methods for free. E.g., with this PR and the latest release of EllipticalSliceSampling (no changes needed):Samplers that do not support
init_params
are not affected by this PR.