Fitness package reorg - #31
Merged
Merged
Conversation
Share assay validation and Polars workflows across scorers. Preserve repeated measurements and reject invalid inputs and undefined assay correlations. Remove unused helpers and update the existing workflow tests for the simpler interfaces.
Share verified weights across the fitness, 2D and 3D benchmarks. Pin all 15 Hub checkpoints and the separate NTv3 model and tokenizer code. Use locked, atomic downloads and verify the Evo2 40B shards. Isolate Triton caches by environment and retain the native Evo loader behavior.
Score GenSLM with mean next-codon log likelihood, excluding padding. Use float32 Evo1 and Evo1.5 inference to preserve rankings across GPUs. Update the leaderboard from the regenerated predictions and add a GenSLM regression test for causal targets, score direction and padding.
Rebuild the full leaderboard and compare 128 variants from each of the 31 ncRNA assays. Require score Spearman >= 0.95 and matching rows, missing values and finite scores. Add native GenSLM and EVmutation checks, numerical regression fixtures, CPU CI and full reproduction coverage with an 80% minimum.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
data now in
data/fitnessand fitness code now inrnagym/fitnessmodule. reproducible pixi envs + shared pixi run -e predict tasks. cut down complexity and verbose docs. model scores exactly reproduce previously generated results within reasonable error (allowing for eg, accumulation of some fp8 rounding discrepancies in Evo models)also added GenSLM fixes, EVmutation scores, and reproducibility tests