Libraries for applying sparsification recipes to neural networks with a few lines of code, enabling faster and smaller models
2.01k
stars
140
forks
source link
[GPTQ UX] Add scheme arg with QuantizationScheme support #2286
Closed
rahul-tuli closed 1 month ago
This PR adds support for a
scheme
arg in GPTQ, this arg can be set to a singleQuantizationScheme
objectrecipe:
test script:
test command:
Output: