Jakevin
/

kev-4b-ternary-mlx

mixed-precision

post-training-quantization

Model card Files Files and versions

kev-4b-ternary-mlx / runtime /scripts

40.6 kB

Ctrl+K

Ctrl+K

1 contributor

History: 2 commits

Jakevin's picture

Kev-4B v2.0: D3 sensitivity-allocated mixed ternary/3-bit layout (v1.0 kept at tag v1.0)

28fd117 verified 19 days ago

build_packed_model.py

7.37 kB
Kev-4B v2.0: D3 sensitivity-allocated mixed ternary/3-bit layout (v1.0 kept at tag v1.0) 19 days ago
mlx_long_states.py

29.7 kB
Publish standalone Kev-4B ternary MLX v0.1.0 experimental 20 days ago
profile_packed_model.py

3.54 kB
Publish standalone Kev-4B ternary MLX v0.1.0 experimental 20 days ago