File size: 3,461 Bytes
a383b20 5f2cfa9 a383b20 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 | # SAM3 TensorRT Engine
<!-- fury-evidence-20260930:start -->
## Results and visual examples
Hardware-specific SAM3 TensorRT artifacts and runtime metadata. Prompted masks from another SAM3 runtime are not proof for these exact engines. Validate the engine revision, GPU architecture, TensorRT version and metadata together before use.
Available build records: [artifacts/sam3.fixed-prompts@v1/h100-sm90/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/h100-sm90/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json), [artifacts/sam3.fixed-prompts@v1/rtx4090-sm89/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/rtx4090-sm89/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json), [artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/1dd326d35db80e1c5db8cba126155fd8e3cd4416053316ff3434b3bb5dede5dc/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/1dd326d35db80e1c5db8cba126155fd8e3cd4416053316ff3434b3bb5dede5dc/runtime-metadata.json), [artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/2a0f12b7c89d9a671f5e06f29a136937fb2b7ea0522b2bfa975f73d4942dd4a0/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/2a0f12b7c89d9a671f5e06f29a136937fb2b7ea0522b2bfa975f73d4942dd4a0/runtime-metadata.json), [artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/3c7d54ccd0644c963e9526b47a5b39eb9c7a045a1066ee0a41e2d19f08018b25/runtime-metadata.json), [artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/9fc8590ceb69ac518e7c98eb36d5e28a4572b1a08e0af0e71d42f76c4c6fa788/runtime-metadata.json](artifacts/sam3.fixed-prompts@v1/rtx5090-sm120/9fc8590ceb69ac518e7c98eb36d5e28a4572b1a08e0af0e71d42f76c4c6fa788/runtime-metadata.json), [metadata.json](metadata.json).
No checkpoint-matched photo gallery was located in the existing evidence reviewed for this update. The files and runtime records below are the available evidence; no visual quality claim is inferred from their presence.
### Evidence provenance
This documentation update reuses saved results; it does not rerun inference or change weights. The starting repository revision is [`7787e2951766`](https://huggingface.co/carpedm20/sam3-tensorrt/tree/7787e29517660658923d3384685b57bd038e010a). Captions distinguish model predictions, training diagnostics, and aggregate metrics. Qualitative examples are not a representative accuracy estimate.
[Gallery sources and SHA-256 checksums](examples/20260930/provenance.json).
<!-- fury-evidence-20260930:end -->
Pre-built TensorRT engine for [SAM3](facebook/sam3) segmentation.
## Specifications
| Setting | Value |
|---------|-------|
| Resolution | 1008x1008 |
| Batch Size | 1-8 |
| Text Max Length | 32 |
| Precision | fp16 |
| TensorRT | 10.14.1.48 |
| Built on | NVIDIA GeForce RTX 4090 |
| Engine Size | 1597.92 MB |
## Usage
The engine is automatically downloaded when using SAM3 with TensorRT enabled.
```python
from ai.models.sam3 import Sam3SegmentationModel
model = Sam3SegmentationModel()
result = await model.segment_text(image, payload=payload, text_prompts="person")
```
## Rebuild
```bash
make run-ai-local -- python -m scripts.tensorrt.sam3 --force
```
Built at: 2026-01-22 12:47:50 UTC
|