8000 Use Q4_K for attn_v for Q2_K_S when n_gqa >= 4 by ikawrakow · Pull Request #4996 · ggml-org/llama.cpp · GitHub < 8000 meta name="robots" content="noindex, nofollow" />
[go: up one dir, main page]

Skip to content

Use Q4_K for attn_v for Q2_K_S when n_gqa >= 4#4996

Merged
ggerganov merged 1 commit intomasterfrom
ik/better_q2_k_s
Jan 17, 2024
Merged

Use Q4_K for attn_v for Q2_K_S when n_gqa >= 4#4996
ggerganov merged 1 commit intomasterfrom
ik/better_q2_k_s

Commits

Commits on Jan 17, 2024

0