issues
search
jax-ml
/
ml_dtypes
A stand-alone implementation of several NumPy dtype extensions used in machine learning.
Apache License 2.0
146
stars
23
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
pre-commit: update & freeze hooks
#179
jakevdp
opened
1 day ago
0
Internal exhasutive test refactoring
#178
copybara-service[bot]
opened
1 day ago
0
Add FP8 support to the exhaustive tests
#177
copybara-service[bot]
opened
1 day ago
0
Add implicit F8 conversions
#176
copybara-service[bot]
opened
2 days ago
0
CI: update actions & pin via hashes
#175
jakevdp
closed
2 days ago
0
CI: update OSX workflows to macos-12
#174
jakevdp
closed
2 days ago
0
CI: build against the most recent NumPy 2.X release
#173
jakevdp
closed
2 days ago
0
BUG: fix float divmod with zero denominator
#172
jakevdp
closed
5 days ago
1
Add float8_e3m4
#171
apivovarov
closed
1 day ago
2
np.floor_divide(?, ml_dtypes.???(0.0)) return NaN but np.float16 returns Inf.
#170
apivovarov
closed
5 days ago
3
Add .clang-format
#169
apivovarov
opened
1 week ago
2
Added CPython free-threading support and basic CI env
#168
vfdev-5
opened
1 week ago
3
Finite precision error calculations always 0 under JIT with bfloat16
#167
colehaus
closed
2 weeks ago
2
Add `float8_e8m0_fnu` (E8M0) OCP MX scale format
#166
balancap
opened
2 weeks ago
3
Updating v0.2.0: Adding new float8 type (fp8e4_m3)
#165
tapan0007
closed
1 week ago
2
Support unpacked FP4E2M1
#164
justinchuby
opened
3 weeks ago
1
WIP: Add MX `E8M0` datatype to `ml_dtypes`
#163
balancap
closed
3 weeks ago
0
f16 sNaN is casted to f8e5m2 qNaN in Half_To_Float8E5m2 test
#162
apivovarov
opened
3 weeks ago
2
Add float8_e4m3
#161
apivovarov
closed
1 week ago
7
README: fix discussion of 2-bit integer layout
#160
jakevdp
closed
1 month ago
0
Update to Numpy 2.0?
#159
sbrl
closed
1 month ago
2
Update build requirement to numpy 2.0 release
#158
jakevdp
closed
2 months ago
0
CI: add test against oldest supported numpy
#157
jakevdp
closed
2 months ago
0
Update minimum NumPy version to v1.26.4.
#156
hawkinsp
closed
1 week ago
2
Package Requirements Looks Inconsistent
#155
daskol
closed
2 months ago
5
[ml_dtypes] Add int2 and uint2 types.
#154
copybara-service[bot]
closed
3 months ago
0
[ml_dtypes] Remove unused int4 forwarding header.
#153
copybara-service[bot]
closed
3 months ago
0
Support HiF8
#152
dengguojie
closed
2 months ago
0
Refactor FloatPyCast for improved performance using lookup table
#151
vxst
opened
3 months ago
2
TBox
#150
vxst
opened
3 months ago
1
put a description ?
#149
stonebig
opened
4 months ago
1
Deserialization of float8_e5m2
#148
wonjeon
opened
4 months ago
2
Support for non-saturating mode for fp8
#147
wonjeon
opened
4 months ago
0
Fix casts from f8e5m2fnuz to half.
#146
copybara-service[bot]
closed
4 months ago
0
numpy casting error: Cannot cast array data from dtype(bfloat16) to dtype(bfloat16) according to the rule 'unsafe'
#145
idanori
closed
4 months ago
5
Compilation Error in libtensorflow Pointing to float8.h in ml_dtype
#144
HongMJ1315
opened
4 months ago
6
Bump version to 0.4.0.
#143
hawkinsp
closed
4 months ago
0
Basic math ops
#142
vatsalsalla
closed
5 months ago
1
Erroneous issue
#141
awf
closed
5 months ago
0
Generalize int4 to other sizes
#140
copybara-service[bot]
closed
5 months ago
0
CI: add ruff pre-commit
#139
jakevdp
closed
5 months ago
0
Bump version to 0.4.0b1
#138
jakevdp
closed
5 months ago
1
Fix invalid type registrations under NumPy 2.0.
#137
hawkinsp
closed
5 months ago
0
Adjust ml_dtypes to support NumPy 2.0 dtype registration APIs
#136
copybara-service[bot]
closed
6 months ago
0
Adept to NumPy change
#135
seberg
closed
6 months ago
1
bfloat16s (at least) throw error when given a format specifiier
#134
colehaus
opened
7 months ago
0
Bump version to v0.3.2
#133
jakevdp
closed
7 months ago
0
Simplify dtype registration logic
#132
jakevdp
closed
8 months ago
1
CI: add several more pre-commit hooks
#131
jakevdp
closed
8 months ago
0
Remove pybind11 dependency
#130
jakevdp
closed
8 months ago
0
Next