Aia haung

Aia12

2

AI & ML interests

None yet

Recent Activity

new activity about 2 months ago

XiaomiMiMo/MiMo-V2.5:Released FP8 weights cannot be loaded with transformers: AttributeError: FP8Expert has no attribute `0` in replace_with_fp8_linear

new activity about 2 months ago

XiaomiMiMo/MiMo-V2.5:modeling_mimo_v2.py splits fused qkv_proj contiguously, but the released weights store it as 4 concatenated TP shards → garbage output (ppl 1.2e6 → 6.67 after fix)

View all activity

Organizations

None yet

models 0

None public yet

datasets 0

None public yet