prabodbc commited on
Commit
0f746b7
·
verified ·
1 Parent(s): e3fc87e

Re-convert with fixed rotary q/k permutation (baseRT-internal#114); NIAH-validated at 2k/8k. Q4 now uses the default-q4 quality profile (embeddings/head at higher precision).

Browse files
Files changed (1) hide show
  1. Llama-3.2-3B-Instruct-Q8.base +1 -1
Llama-3.2-3B-Instruct-Q8.base CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fe94e68856346a8016f83d142b5fac16555e180a6409f357b749e7f708ec7dea
3
  size 3320125440
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b26f0dd86648cec9d1f7f30dbf986c59077395dffe8607a28640a2e8d251177a
3
  size 3320125440