Hugging Face's logo

embedl
/
gemma-3-270m-it-FlashHead

Safetensors
gemma3_text
Model card Files
xet
 Community
WilhelmT commited on
Commit
85dde8d
·
verified ·
1 Parent(s): 05f2184

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -60,7 +60,7 @@ python3 -m embedl.models.vllm.demo --model embedl/gemma-3-270m-it-FlashHead
60
  | BF16 baseline | 397 | 1.0× |
61
  | **FlashHead (Embedl)** | **526** | **1.32×** |
62
  | W4A16 baseline | 420 | 1.06× |
63
- | **FlashHead W4A16 (Embedl)** | **568** | **1.74×** |
64
 
65
  FlashHead improves end-to-end speed by **1.35×** over state-of-the-art, while maintaining full accuracy parity.
66
 
 
60
  | BF16 baseline | 397 | 1.0× |
61
  | **FlashHead (Embedl)** | **526** | **1.32×** |
62
  | W4A16 baseline | 420 | 1.06× |
63
+ | **FlashHead W4A16 (Embedl)** | **568** | **1.43×** |
64
 
65
  FlashHead improves end-to-end speed by **1.35×** over state-of-the-art, while maintaining full accuracy parity.
66