issues
search
stanford-crfm
/
haliax
Named Tensors for Legible Deep Learning in JAX
Apache License 2.0
140
stars
9
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Use Generics
#97
thomasahle
opened
2 days ago
1
Another attempt at shard
#96
dlwh
closed
1 week ago
0
Run levanter tests
#95
dlwh
closed
3 weeks ago
0
No default in_resources
#94
dlwh
closed
3 weeks ago
1
Use sqrt(fan_in) as default in and truncated_normal
#93
dlwh
closed
3 weeks ago
0
scan/fold should not consider the `init` args for scanning over
#92
rjpower
closed
4 weeks ago
5
Expose the filter_checkpoint helper: it's useful for custom training loops.
#91
rjpower
closed
1 month ago
0
Move StateDict to Haliax, clean it up a lot.
#90
dlwh
opened
1 month ago
0
revert previous PR
#89
blahBlahhhJ
closed
1 month ago
0
Fix pspec method for levanter
#88
blahBlahhhJ
closed
1 month ago
0
Add replica axes
#87
blahBlahhhJ
closed
1 month ago
1
fix hax.where in the presence of scalar NamedArrays
#86
dlwh
closed
2 months ago
0
add in-place mutation ops for NamedArrays, support lists/1-d JAX Arrays for indexing
#85
dlwh
closed
2 months ago
0
allow einsum to accept aliases similar to rearrange
#84
dlwh
closed
2 months ago
0
switch to pdm:
#83
dlwh
closed
2 months ago
0
fix revision number calculation for auto-publishing
#82
dlwh
closed
2 months ago
0
Poetry + Dev Builds
#81
dlwh
closed
3 months ago
0
Fix multi device shard
#80
dlwh
closed
3 months ago
0
Simplify sharding
#79
dlwh
closed
3 months ago
0
Make Conv work when input type is not the same as kernel type
#78
dlwh
opened
3 months ago
0
Add FP8 support to Haliax
#77
dlwh
closed
3 months ago
0
No Scan Layers
#76
dlwh
closed
3 months ago
0
jax.random.KeyArray deprecated
#75
jennifgcrl
closed
2 months ago
2
workaround current limitation in jax_metals
#74
dlwh
closed
4 months ago
0
explicitly shard scalars in named_jit
#73
dlwh
closed
4 months ago
0
Support for partitioning/sharded data with Pallas kernels?
#72
G-Levine
opened
4 months ago
10
tile and repeat
#71
dlwh
closed
4 months ago
0
add a non-scan-layers version of Stacked
#70
dlwh
closed
3 months ago
0
Add single argument mode for hax.where
#69
blahBlahhhJ
closed
4 months ago
0
Don't check shapes in tree_unflatten. Fixes #66
#68
dlwh
closed
4 months ago
0
Make a helper function to squash/unsquash all axes (except some) into a single batch axis
#67
dlwh
closed
4 months ago
1
named arrays and `eqxi.while_loop(mode=checkpointed)` do not get a long
#66
dlwh
closed
4 months ago
0
Mixed Precision and Resource Envs
#65
dlwh
opened
5 months ago
0
Lp norm function?
#64
rohan-mehta-1024
opened
5 months ago
1
Einsum
#63
dlwh
closed
5 months ago
0
auto_shard inside zeros etc
#62
dlwh
opened
5 months ago
0
top_k returns indices in addition to values
#61
blahBlahhhJ
closed
5 months ago
1
Add checkpointed_scan
#60
dlwh
opened
5 months ago
0
add htu.tree_map etc
#59
dlwh
closed
5 months ago
0
better exception when calling .item inside jit
#58
dlwh
opened
5 months ago
0
better stack trace for concretization inside scan
#57
dlwh
opened
5 months ago
0
Dev
#55
dlwh
closed
4 months ago
0
Add tree_map etc variants that set is_leaf=is_named_array by default
#54
dlwh
closed
5 months ago
1
make Linear support overlapping input/axis names?
#53
dlwh
opened
6 months ago
0
Support Rearrange Without Tailing Dimensions
#52
karan-dalal
closed
5 months ago
2
Add Pooling operations
#51
dlwh
closed
6 months ago
0
Adaptive Pooling
#50
dlwh
opened
8 months ago
0
refactor dot_product_attention to be more general but also check argu…
#49
dlwh
closed
8 months ago
0
Add a simple MLP module
#48
dlwh
closed
8 months ago
0
Update README.md
#47
jennifgcrl
closed
8 months ago
0
Next