issues
search
stanford-crfm
/
levanter
Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax
https://levanter.readthedocs.io/en/latest/
Apache License 2.0
495
stars
80
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Simplifiy tokenization pipeline, make it work with large numbers of shards again, (re)add configuration metadata to cache
#752
dlwh
opened
1 day ago
0
Bump tensorstore from 0.1.63 to 0.1.66
#751
dependabot[bot]
opened
2 days ago
0
Bump ray[default] from 2.34.0 to 2.37.0
#750
dependabot[bot]
opened
2 days ago
0
Minor fix for docker TPU workflow
#749
ahmeda14960
opened
5 days ago
0
don't require local_cpu_mesh for dataloading functions. we should be doing it automatically...
#748
dlwh
opened
5 days ago
0
Tweaks to Ray TPU stuff
#747
dlwh
closed
5 days ago
0
Adding supervised data config
#746
TheQuantumFractal
opened
6 days ago
0
don't require output_exemplar for user code
#745
dlwh
opened
1 week ago
0
Make new tokenization ~67% faster
#744
dlwh
closed
1 week ago
1
bump levanter version
#743
dlwh
closed
1 week ago
0
see if it's this file in particular
#742
dlwh
closed
1 week ago
1
Automatically set up wandb workspace for people
#741
dlwh
opened
1 week ago
1
fix llama 3 rotary embeddings
#740
dlwh
closed
1 week ago
0
Bump equinox from 0.11.6 to 0.11.7
#739
dependabot[bot]
closed
1 week ago
1
fix sequence parallel attention in splash attention
#738
dlwh
closed
1 week ago
0
Support for running in a Ray cluster
#737
dlwh
closed
1 week ago
2
New Tokenization Pipeline Fails Silently
#736
ahmeda14960
opened
1 week ago
1
Tensorboard logging breaks if value is of the type "string"
#735
mhmaqbool
opened
1 week ago
2
Display summary statistics on stdout
#734
mhmaqbool
opened
1 week ago
1
Levanter unit test issue
#733
DwarKapex
opened
1 week ago
3
Update datasets requirement from ~=2.18 to >=2.18,<4.0
#732
dependabot[bot]
closed
2 weeks ago
0
Bump tensorstore from 0.1.64 to 0.1.65
#731
dependabot[bot]
closed
2 weeks ago
0
Bump equinox from 0.11.3 to 0.11.6
#730
dependabot[bot]
closed
2 weeks ago
0
add bits-per-byte calculation to levanter
#729
dlwh
closed
2 weeks ago
0
get rid of eraconfig b/c draccus can't handle it
#728
dlwh
closed
2 weeks ago
0
Crash on GPU in Roberta branch
#727
dlwh
opened
2 weeks ago
0
Reset optimizer state is not working
#726
ahmeda14960
opened
2 weeks ago
0
Fix eqx
#725
dlwh
closed
2 weeks ago
0
fix extra context docker build bug
#724
blahBlahhhJ
closed
3 weeks ago
0
Update gcsfs requirement from <2024.7,>=2024.2 to >=2024.2,<2024.10
#723
dependabot[bot]
closed
3 weeks ago
0
Update fsspec[http] requirement from <2024.7,>=2024.2 to >=2024.2,<2024.10
#722
dependabot[bot]
closed
3 weeks ago
0
Bump equinox from 0.11.4 to 0.11.5
#721
dependabot[bot]
closed
3 weeks ago
0
ModuleNotFoundError: No module named 'async_lru'
#720
mhmaqbool
closed
3 weeks ago
3
attempt at launcing small fast
#719
dlwh
closed
2 weeks ago
0
unpin ray
#718
dlwh
closed
3 weeks ago
0
add sequence packing for evals
#717
dlwh
opened
3 weeks ago
0
WIP Completely rework dataset/cache system: instant resume, perfect shuffle, stable mixtures and more
#716
dlwh
closed
3 weeks ago
3
use hf config from checkpoint by default
#715
dlwh
closed
4 weeks ago
5
Bump ray[default] from 2.34.0 to 2.35.0
#714
dependabot[bot]
closed
4 weeks ago
0
fix device kind for mfu v5e
#713
dlwh
closed
1 month ago
0
will this work
#712
dlwh
closed
1 month ago
0
Do whatever work to get autocheckpoint working
#711
dlwh
opened
1 month ago
0
suppress stderr in describe_tpu since it usually logs a dumb error
#710
dlwh
closed
1 month ago
0
add haps configuration (cycle lr schedule)
#709
blahBlahhhJ
closed
1 month ago
2
Fix tpu vm autoshutdown
#708
dlwh
closed
1 month ago
0
Fix base again
#707
dlwh
closed
1 month ago
0
Llama mixture
#706
abhinavg4
closed
1 month ago
2
grr
#705
dlwh
closed
1 month ago
0
fix incremental build on CI
#704
dlwh
closed
1 month ago
0
publish full tpu image
#703
dlwh
closed
1 month ago
0
Next