---
family: d-fine
license: apache-2.0
model_id: d-fine-s__dfine_grapha_fp16
source_url: https://huggingface.co/litert-community/D-FINE-S-LiteRT
task: object-detection
---

# d-fine-s__dfine_grapha_fp16

| | |
|---|---|
| **Task** | object-detection |
| **Family** | d-fine |
| **Source** | https://huggingface.co/litert-community/D-FINE-S-LiteRT |
| **License** | apache-2.0 |

## Conversion

| | |
|---|---|
| **Tool** | litert-torch 0.10.0 (editable dev checkout 115a136 + local patches) |
| **Command** | `python scripts/build_dfine_split.py && python scripts/build_dfine_fix3.py && python scripts/pack_assets.py` |
| **Quantization** | fp16 |

## Artifacts

| File | SHA-256 | Size (MB) |
|---|---|---|
| `c7232e5b17f66c54c88a72e6db7430b694c3cd02460114b590bd401cbf53546e.tflite` | `9f9bbaa3a063bbc9612f1758e7428fab4427c9d229b02d611023819947891594` | 12.373 |

## Performance

No benchmark data yet.

## Browser (LiteRT.js)

Model-level sweep results — measured behavior of this exact `.tflite` in the browser, only meaningful together with the environment that produced it. Full records (failure class, error evidence, sweep config): `data/sweep/2.5.3/2026-08-11/d-fine-s__dfine_grapha_fp16.json`.

| Backend | Status | Full delegation | Output match | Max rel diff | Latency p50 (ms) | Environment | Date | Provenance |
|---|---|---|---|---|---|---|---|---|
| wasm_xnnpack | pass | - | - | - | 1461.235 | mac-studio-m4-max · chromium 151.0.7922.34 headless · macOS · @litertjs/core 2.5.3 | 2026-08-11 | measured |
| webgpu_mldrift | pass | yes | pass | 22.71427481515067 | 47.127 | mac-studio-m4-max · chromium 151.0.7922.34 headless · macOS · @litertjs/core 2.5.3 | 2026-08-11 | measured |

## Device runs (NPU / LiteRT-LM)

Model-level on-device results — measured behavior of this exact artifact on the recorded device, only meaningful together with that environment. Throughput is conditional on the measured prompt length (Prompt (tokens); '-' = the source stated none): rows differing only there are different measurements, not re-runs. Full records (failure classes, log evidence): `data/device_runs/2.2.0.dev20260804/2026-08-31/d-fine-s__dfine_grapha_fp16__raspberry-pi-5.json`, `data/device_runs/2.2.0/2026-08-25/d-fine-s__dfine_grapha_fp16__galaxy-s26.json`.

| Device | Accelerator | Status | Full delegation | Output match | Prompt (tokens) | Latency p50 (ms) | Prefill tok/s | Decode tok/s | TTFT (ms) | Peak mem (MB) | Environment | Date | Provenance |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| galaxy-s26 | gpu_mldrift | pass | - | - | - | 19.73 | - | - | - | - | Galaxy S26 (SM-S942Q) · Qualcomm SM8850 · litert 2.2.0 · Android 16 | 2026-08-25 | measured |
| galaxy-s26 | npu_qnn | pass | - | - | - | 23.812 | - | - | - | - | Galaxy S26 (SM-S942Q) · Qualcomm SM8850 · litert 2.2.0 · QAIRT (Hexagon, JIT on-device) · Android 16 | 2026-08-25 | measured |
| raspberry-pi-5 | cpu_xnnpack | pass | - | - | - | 358.316 | - | - | - | 176.84 | Raspberry Pi 5 Model B Rev 1.1 · litert 2.2.0.dev20260804 · Linux-6.18.34+rpt-rpi-2712-aarch64-with-glibc2.41 | 2026-08-31 | measured |

## Pitfalls

- Graph A deliberately emits memory_raw multiplied by 2 (a clean-output-buffer workaround for a Mali 3D-token fan-out bug); the host must halve it before feeding Graph B (HF card 'How it runs' / zoo README L324-328).
- The per-token tail (enc_output + enc_bbox_head over the top-300 tokens, weights in host_params.bin) runs on the host between the two graphs (HF card).

---

_Rendered by `edge-card` from `card.json` (the source of truth); edit the inputs, not this file._