---
title: SAM2.1-hiera-small-fp16
canonical_url: "https://www.modelscope.cn/models/mlx-community/SAM2.1-hiera-small-fp16"
md_url: "https://www.modelscope.cn/models/mlx-community/SAM2.1-hiera-small-fp16.md"
repository: mlx-community/SAM2.1-hiera-small-fp16
last_updated: 2026-09-23
license: apache-2.0
pipeline_tag: image-segmentation
tasks:
  - image-segmentation
base_model:
  - facebook/sam2.1-hiera-small
base_model_relation: finetune
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 19
stars: 0
tags:
  - mlx
  - segmentation
  - promptable-segmentation
  - sam2
  - sam2.1
---

# SAM2.1-hiera-small-fp16

> SAM2.1-hiera-small-fp16 - mlx-community 在 ModelScope 开源的模型。mlx-community/SAM2.1-hiera-small-fp16

mlx-community/SAM2.1-hiera-small-fp16 是 ModelScope 魔搭社区上的image-segmentation模型，采用 apache-2.0 许可，基于 facebook/sam2.1-hiera-small 构建。

- **Repository**: mlx-community/SAM2.1-hiera-small-fp16
- **License**: apache-2.0
- **Tasks**: image-segmentation
- **Base model**: facebook/sam2.1-hiera-small
- **Tags**: mlx, segmentation, promptable-segmentation, sam2, sam2.1
- **Downloads**: 19
- **Stars**: 0
- **Last updated**: 2026-09-23

Source: https://www.modelscope.cn/models/mlx-community/SAM2.1-hiera-small-fp16

---

# mlx-community/SAM2.1-hiera-small-fp16

[SAM 2.1](https://github.com/facebookresearch/sam2) Hiera-Small, **image mode** (point / box → mask),
converted to Apple **MLX** (fp16) from Meta's official `sam2.1_hiera_small.pt` for the
[`mlx-edgetam-swift`](https://github.com/xocialize/mlx-edgetam-swift) Swift package (`SAM21Package`, an MLXEngine
`promptSegment` ModelPackage).

- **Contents:** the image path only — Hiera trunk + FPN neck + SAM prompt encoder + mask decoder + `no_mem_embed`
  (359 tensors, 38.5 M params). The video memory stack is not included.
- **Keys:** upstream's names, unchanged. Convolutions are NHWC `(O,kH,kW,I)`. The trunk's `pos_embed` and
  `pos_embed_window` are `(1,H,W,C)`. Converter: `oracle/convert_sam21.py`.
- **Parity:** the Swift port matches PyTorch `SAM2ImagePredictor` **through `set_image`** on the CPU fp32 stream.
  On a 1024² flat-shaded image, a 1024² captioned copy and a 1200×800 synthetic page:
  - image_embed relative error ≤ 5.3e-6;
  - every click and box mask matches at IoU 1.0000, with predicted-IoU Δ ≤ 1e-5.
  With these fp16 weights and fp16 activations on the GPU, mask IoU vs fp32 PyTorch is ≥ 0.9996 on
  object-sized prompts.
- **Footprint (M5 Max):** 1024² encode ~0.1 s warm; each further prompt on the same image 11–18 ms; process
  `phys_footprint` 2.4–2.8 GB.

## Why use it next to EdgeTAM
[EdgeTAM](https://huggingface.co/mlx-community/EdgeTAM-fp16) (13.9 M params) is the fast default and handles
video. On single clicks over flat-shaded or synthetic images, SAM 2.1-S is much better:
- a click on a cartoon fox's belly selects the whole fox (IoU 0.968 vs 0.008 for EdgeTAM);
- a click on one small square of a grey page selects only that square, where EdgeTAM also adds a corner
  square every time.

## Use
```swift
// .package(url: "https://github.com/xocialize/mlx-edgetam-swift", from: "0.6.0")
import EdgeTAM
let p = try EdgeTAMPredictor.fromPretrained(weightsPath, dtype: .float16)   // Hiera detected from the weights
p.setImage(sourceCGImage)
let r = p.predict(point: (471, 635))            // r.mask, r.soft (anti-aliased), r.score
let boxed = p.predict(points: [], labels: [], box: [40, 245, 745, 910])
```
Or through MLXEngine: `SAM21Package` (`MLXEdgeTAM`), with `mode: softMatte` for an anti-aliased matte.

Weights: Apache-2.0 (facebookresearch/sam2). Port code: MIT.