fivetech commited on
Commit
cb41912
·
verified ·
1 Parent(s): 445e9ac

fix: qkv_proj per-rank scale-grid repair (ranks 1..3 of all 48 layers)\n\nPre-fix converter dequantized fused qkv with a flat scale grid, corrupting\nranks 1..NR-1 (rank 0 intact). Repaired from XiaomiMiMo/MiMo-V2.5 fp8 source\nwith the fixed per-rank dequant + identical int8 requant. Verified:\nverify_mimo_qkv 5/5 layers PASS all ranks; ppl prose 9.97 / code 2.07\n(code matches the 12.40/2.07 reference axis); greedy 'The capital of France\nis' -> 'Paris.' See peng findings.md §46.

Browse files
Files changed (1) hide show
  1. out-00001.safetensors +1 -1
out-00001.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bec6ba3b01a6b25163fe7300f1e4489c7723cb39101749ce8a3e88ad2331932e
3
  size 12635280000
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5c62b9dd039d0023449ff75bdff33e60b88c35dbabe5b21c02c1f03d975be22
3
  size 12635280000