Views
No views yet
| Metric | Standard | Flash |
|---|---|---|
| Avg decode tok/s | 99 | 152 |
| Median decode tok/s | 90 | 157 |
| Avg peak GPU memory | 2.64 GB | 2.64 GB |
| Load time | ~0.4 s | ~0.7 s |
| Prompt type | Median tok/s | Example |
|---|---|---|
| Describe (EN, 1 sentence) | 198.5 | "A man in a white lab coat with the name 'De WARON' on it stands behind a desk..." |
| What text appears? | 96.9 | "De Wooning" |
| Count objects/people | 61.4 | "1" |
| Main subject | 170.1 | "The main subject is a man wearing a white lab coat with the name 'De Woon' and a logo on it." |
| Describe (ZH) | 221.2 | "这是一...話是柯尼特, 黃色胸部..." |
mlx-vlm adds a visionpsy_nano handler, load via the MLX port at KaedeTai/mlx-video @ visionpsy-mlx-port:1from huggingface_hub import snapshot_download
2from mlx_video.models.visionpsy_nano import load_visionpsy_nano
3from mlx_video.models.visionpsy_nano.processor import load_processor
4from PIL import Image
5
6path = snapshot_download("KaedeTai/VisionPsy-Nano-460M-Flash-MLX")
7
8model, cfg = load_visionpsy_nano(path)
9proc = load_processor(path, cfg=cfg)
10
11img = Image.open("photo.jpg").convert("RGB")
12batch = proc("Describe this image in one sentence.", image=img)
13
14tokens = list(model.generate(
15 batch["input_ids"],
16 pixel_values=batch["pixel_values"],
17 image_token_id=batch["image_token_id"],
18 max_new_tokens=64,
19 eos_token_id=proc.tokenizer.eos_token_id,
20))
21print(proc.decode(tokens, skip_special_tokens=True))