ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 17:20:01 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
30381fc1fc6a302f9de0487b1e719f4efcc06a00
ik_llama.cpp/github-data/pull_requests/69 - Allow bf16 kv-cache.md
Thomas eaa2510a28 Add GitHub data: filename sanitization (#640)
2025-07-23 13:31:53 +02:00

544 B
Raw Blame History

🔀 #69 - Allow bf16 kv-cache

Author ikawrakow
State ❌ Closed
Created 2024-09-29
Updated 2024-09-29

Description

On the CPU I get the exact same PPL with and without FA using bf16 for kv-cache. But on CUDA the bf16 kv-cache result is about the same as the fp16 kv-cache CPU result, so I'm missing some conversion somewhere. Either way, we can now run on all platforms supported here with bf16 kv-cache.