Datasets:

jaytonde05
/

blackwell-quantization-results

Dataset card Files Files and versions

blackwell-quantization-results / model_quantization

Commit History

32B: use full max-seq-length 2048 (whole samples) on 2x RTX PRO 6000

70bc323
verified

jaytonde05 commited on Jul 3

32B: use full max-seq-length 2048 (whole samples) on 2x RTX PRO 6000

0ae75fd
verified

jaytonde05 commited on Jul 3

Correct docs: 8B models used default seqlen 2048, not 512

cfead8c
verified

jaytonde05 commited on Jul 3

Add model quantization scripts and command reference

e695de7
verified

jaytonde05 commited on Jul 3