Repository navigation

All pull requests

Find the PET adaptive cutoff by bisection
#1331sirmarcel opened Oct 27, 2026Contributor
4
Add mean-variance estimation (MVE) heads to FlashMD + LLPR
#1329wiseodd opened Oct 27, 20262 of 5 tasks
38
Add functions to couple and uncouple spherical targets.
#1321pfebrer opened Oct 22, 2026Contributor
9
Use vesin.NeighborList directly for neighbor lists
#1319LPinede opened Oct 21, 2026
1
Set an explicit 10s timeout for all tests
#1318Luthaf opened Oct 21, 2026Member3 of 5 tasks
4
Implement per-block heads and species-conditioned readouts in PET
#1308jwa7 opened Oct 16, 2026Member3 of 4 tasks
14
Keep fine-tunes working after the non_conservative_force rename
#1293EricBoittier opened Oct 6, 2026Contributor2 of 5 tasks
Create the directories in mtt train -o, and fix copying outputs
#1283ppegolo opened Sep 30, 2026Contributor2 of 5 tasks
[integration, do not merge] PET input pipeline: instrumentation, persistent workers, tensor transport, pinned batches
#1275EricBoittier opened Sep 18, 2026Contributor2 of 5 tasks
Time each collate transform separately
#1274EricBoittier opened Sep 18, 2026Contributor1 of 5 tasks
Pin batches so the transfer to the GPU can overlap compute
#1273EricBoittier opened Sep 18, 2026Contributor5 tasks
Name the collate transforms after what they do
#1272EricBoittier opened Sep 18, 2026Contributor5 tasks
Keep dataloader workers alive between epochs
#1271EricBoittier opened Sep 18, 2026Contributor1 of 5 tasks
Split collation from serialization in CollateFn
#1270EricBoittier opened Sep 18, 2026Contributor1 of 5 tasks
Let unpack_batch() accept an already-collated Batch
#1269EricBoittier opened Sep 18, 2026Contributor1 of 5 tasks
Add per-stage timing of the PET training step, and a benchmark
#1268EricBoittier opened Sep 18, 2026Contributor2 of 5 tasks
read periodicity from pbc.npy in memmapdataset
#1260sofiia-chorna opened Sep 9, 2026Collaborator2 of 5 tasks
9
Add possibility to build docs without running examples
#1224pfebrer opened Jul 27, 2026Contributor
6
Add hooks for indirect learning.
#1209pfebrer opened Jul 18, 2026Contributor
23
add mtt show
#1204sofiia-chorna opened Jul 15, 2026Collaborator
13
memmap extension to disk dataset
#1171jwa7 opened Jun 25, 2026Member5 tasks
fix: handle missing gradients in remove_additive for mixed datasets
#1140ChaitanyaParate opened May 10, 2026Contributor
4
Move all files generated by examples in a data subfolder
#1078Luthaf opened Mar 18, 2026Member2 of 5 tasks
Add the possibility to retrieve descriptions from datasets
#1052pfebrer opened Feb 25, 2026Contributor
4
feat(torch.compile): PET FX compilation
#1029HaoZeke opened Feb 2, 2026Member4 of 5 tasks
26

Navigation Menu

Repository navigation

All pull requests

Find the PET adaptive cutoff by bisection

Add mean-variance estimation (MVE) heads to FlashMD + LLPR

Add functions to couple and uncouple spherical targets.

Use vesin.NeighborList directly for neighbor lists

Set an explicit 10s timeout for all tests

Implement per-block heads and species-conditioned readouts in PET

Keep fine-tunes working after the non_conservative_force rename

Create the directories in mtt train -o, and fix copying outputs

[integration, do not merge] PET input pipeline: instrumentation, persistent workers, tensor transport, pinned batches

Time each collate transform separately

Pin batches so the transfer to the GPU can overlap compute

Name the collate transforms after what they do

Keep dataloader workers alive between epochs

Split collation from serialization in CollateFn

Let unpack_batch() accept an already-collated Batch

Add per-stage timing of the PET training step, and a benchmark

read periodicity from pbc.npy in memmapdataset

Add possibility to build docs without running examples

Add hooks for indirect learning.

add mtt show

memmap extension to disk dataset

fix: handle missing gradients in remove_additive for mixed datasets

Move all files generated by examples in a `data` subfolder

Add the possibility to retrieve descriptions from datasets

feat(torch.compile): PET FX compilation

Pull requests