Loading
Du hast einen ComfyUI-Workflow? Erstelle direkt im Browser eine verständliche Workflow-Dokumentation.Workflow verständlich machen.Jetzt ausprobieren ↗
GGUF-format weights of Roboflow RF-DETR Base (detection variant) for use with rfdetr.cpp, a C++/ggml implementation that matches the upstream PyTorch model on CPU.
Modellquelle
Quellenbeschreibung
GGUF-format weights of Roboflow RF-DETR Base (detection variant) for use with rfdetr.cpp, a C++/ggml implementation that matches the upstream PyTorch model on CPU.
This repo contains all four standard quantizations of this variant. F16 is the recommended default — same accuracy as F32, 1.85× smaller, and typically the fastest on modern CPUs thanks to ggml's F32×F16 matmul fast path.
Quellen
1 QuelleVerifiziert 27. Sept.
Modellartefakte
4 ArtefakteQuellenauszüge
2 Auszüge| File | Quant | Size (MB) | Recall @ IoU 0.5 | Recall @ IoU 0.95 | Mean |Δscore| | Latency (median ms, T=8) |
|---|---|---|---|---|---|---|
rfdetr-base-f32.gguf | F32 | 119.2 | 1.0000 | 0.9890 | 0.0080 | 144.0 |
rfdetr-base-f16.gguf ← recommended | F16 | 64.2 | 1.0000 | 0.9890 | 0.0082 | 136.1 |
rfdetr-base-q8_0.gguf | Q8_0 | 38.5 | 1.0000 | 0.9890 | 0.0091 | 145.7 |
rfdetr-base-q4_K.gguf | Q4_K | 31.5 |
All accuracy numbers are computed against the upstream PyTorch reference (rfdetr 1.7.0) on 7 COCO val2017 images at threshold 0.5. Latency is measured with rfdetr-cli bench (8 iters + 3 warmup) at T=8 threads on a single AMD Ryzen 9 9950X3D image (coco_kitchen.jpg, 640x427).
ne[0] % 256 != 0 (the decoder's 128-dim MLP halves, 60 tensors) silently fall back to Q8_0 per ggml's quantizer logic — net compression is still ~3.8× over F32. Use only when the size budget is tight; expect a measurable [email protected] drop relative to F16/Q8_0 (see file table above).# 1. Clone + build rfdetr.cpp
git clone https://github.com/mudler/rf-detr.cpp
cd rt-detr.cpp
cmake -B build -DRFDETR_BUILD_CLI=ON && cmake --build build -j
# 2. Download a quant (F16 recommended)
hf download mudler/rfdetr-cpp-base rfdetr-base-f16.gguf --local-dir models/
# 3. Run detection
build/bin/rfdetr-cli detect \
--model models/rfdetr-base-f16.gguf \
--input my_image.jpg \
--threshold 0.5 --threads 8 \
--output detections.json
All accuracy metrics are computed against the upstream PyTorch reference (rfdetr 1.7.0) on 7 COCO val2017 images at threshold 0.5. Each detection match uses greedy Hungarian-style assignment by IoU (≥ 0.5 lenient, ≥ 0.95 strict) with class equality required.
See BENCHMARK.md and benchmarks/results/accuracy_sweep.json for the full sweep across all 32 (variant × quant) cells.
Apache-2.0 — matches the upstream rfdetr license.
rfdetr-base-q4_K.gguf
gguf · 31,5 MB · SHA-256 5d9118d27cd3…9a7d · Hugging Face
--- license: apache-2.0 library_name: rfdetr.cpp tags: - object-detection - rfdetr - gguf - ggml - cpp-inference pipeline_tag: object-detection base_model: roboflow/rfdetr --- # RF-DETR Base — GGUF for rfdetr.cpp GGUF-format weights of [Roboflow RF-DETR Base](https://github.com/roboflow/rf-detr) (detection variant) for use with [rfdetr.cpp](https://github.com/mudler/rf-detr.cpp), a C++/ggml implementation that matches the upstream PyTorch model on CPU. This repo contains all four standard quantizations of this variant. **F16 is the recommended default** — same accuracy as F32, 1.85× smaller, and typically the fastest on modern CPUs thanks to ggml's F32×F16 matmul fast path. ## Available files | File | Quant | Size (MB) | Recall @ IoU 0.5 | Recall @ IoU 0.95 | Mean \|Δscore\| | Latency (median ms, T=8) | |---|---|---:|---:|---:|---:|---:| | `rfdetr-base-f32.gguf` | F32 | 119.2 | 1.0000 | 0.9890 | 0.0080 | 144.0 | | `rfdetr-base-f16.gguf` ← **recommended** | F16 | 64.2 | 1.0000 | 0.9890 | 0.0082 | 136.1 | | `rfdetr-base-q8_0.gguf` | Q8_0 | 38.5 | 1.0000 | 0.9890 | 0.0091 | 145.7 | | `rfdetr-base-q4_K.gguf` | Q4_K | 31.5 | 0.9531 | 0.8794 | 0.0196 | 166.3 | All accuracy numbers are computed against the upstream PyTorch reference (`rfdetr 1.7.0`) on 7 COCO val2017 images at threshold 0.5. Latency is measured with `rfdetr-cli bench` (8 iters + 3 warmup) at T=8 threads on a single AMD Ryzen 9 9950X3D image (`coco_kitchen.jpg`, 640x427). ## Architecture - Backbone: DINOv2-small - Input resolution: 560×560 - Patch size: 14 - Decoder layers: 3 - Object queries: 300 - Task: object detection (boxes only) ## Quantization notes - **F32** — full-precision reference, ~120 MB. Bit-exact PyTorch parity. - **F16** — matmul-multiplicand weights only; LayerNorms, conv kernels, embeddings, biases, and layer-scale gammas stay F32. Lossless on this model and consistently the fastest variant on CPU. - **Q8_0** — best size/accuracy tradeoff under F16; ~3× smaller than F32 with effectively identical detections. - **Q4_K** — smallest practical quant. Rows with `ne[0] % 256 != 0` (the decoder's 128-dim MLP halves, 60 tensors) silently fall back to Q8_0 per ggml's quantizer logic — net compression is still ~3.8× over F32. Use only when the size budget is tight; expect a measurable [email protected] drop relative to F16/Q8_0 (see file table above). ## Usage ```bash # 1. Clo...
Source context: 503 downloads · 0 likes · Pipeline object-detection · Library rfdetr.cpp · Repo mudler/rfdetr-cpp-base
| 0.9531 |
| 0.8794 |
| 0.0196 |
| 166.3 |