mlboydaisuke commited on
Commit
a14dee5
·
verified ·
1 Parent(s): d67c80e

card: add M4 Max 224 tok/s

Browse files
Files changed (1) hide show
  1. README.md +2 -1
README.md CHANGED
@@ -57,7 +57,8 @@ Simpler than the [Qwen3-VL port](https://huggingface.co/mlboydaisuke/Qwen3-VL-2B
57
 
58
  ## Measured (iPhone 17 Pro, iOS 27 beta, release)
59
 
60
- - **decode ~51.5 tok/s** (text core 53.4), engine cold-spec ~3–5 s, ~1.5 GB resident (jetsam-safe).
 
61
  - **Numerics**: fp32-torch parity bit-exact (vision cos 1.000000, full overlay logits cos 1.00004);
62
  Core AI engine ≡ python ≡ HF (text 24/24; image path reproduces the HF description modulo one int8
63
  near-tie token, then reconverges).
 
57
 
58
  ## Measured (iPhone 17 Pro, iOS 27 beta, release)
59
 
60
+ - **iPhone 17 Pro decode ~51.5 tok/s** (text core 53.4) · **M4 Max ~224 tok/s** (text core, `llm-benchmark`),
61
+ engine cold-spec ~3–5 s, ~1.5 GB resident (jetsam-safe).
62
  - **Numerics**: fp32-torch parity bit-exact (vision cos 1.000000, full overlay logits cos 1.00004);
63
  Core AI engine ≡ python ≡ HF (text 24/24; image path reproduces the HF description modulo one int8
64
  near-tie token, then reconverges).