Why so few 8 bit capable models?
1
#13 opened 10 months ago
by
ibivibiv
Can Run "gptq_model-4bit--1g" but not "gptq-4bit-32g-actorder_True"
#12 opened 10 months ago
by
0-hero
comparison with bitsandbytes nf4, hope to increase GPTQ accuracy
12
#11 opened 10 months ago
by
AIReach
Mininum VRAM?
7
#9 opened 10 months ago
by
hierholzer
GGML version possible/coming?
2
#8 opened 10 months ago
by
Thireus
vram requirements
1
#5 opened 10 months ago
by
joujiboi