huihui_ai/
deepseek-v3-abliterated:671b-Q2_K

1,940 5 months ago

A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.

671b
... /
params
d318c0731575 · 160B
{
"num_gpu": 1,
"stop": [
"<｜begin▁of▁sentence｜>",
"<｜end▁of▁sentence｜>",
"<｜User｜>",
"<｜Assistant｜>"
]
}