RTMDet (CSPNeXt, tiny / S / M / L / X; nano and M person detectors)
OpenMMLab's RTMDet (tiny, s, m, l, x on COCO; nano and m person detectors): real-time one-stage box detector, a CSPNeXt CNN with NMS. The person detectors supply the boxes for RTMPose. Apache-2.0.
stable Fork of open-mmlab/mmdetection@44ebd17 · backends: portable, kernels
- Paper RTMDet: An Empirical Study of Designing Real-Time Object Detectors arXiv · first released 14 Dec 2022
Paper 2022 Added in 0.2.0 Evaluated with 0.2.0
Use it
pip install fastforks Lean install with only what this model needs: pip install "fastforks-core[torchvision]"
from fastforks import detection
model = detection.load("rtmdet-tiny-coco") # kernels on a supported GPU, else portable PyTorch
result = model(image)
result.boxes # also: scores, labels
result.backend # what ran: mode, device, dtype, kernels fastforks bench rtmdet-tiny-coco Times the model on this machine. load(…, mode="portable") or mode="kernels" picks a backend; usage guide.
Speed and memory
Every backend on one GPU (RTX 5060 Ti 16 GB) and the same input, the authors’ code included. Kernels runs fp16. Portable runs fp16 on CUDA.
Input: synthetic, 800x1344 (the task’s standard size)
| Backend | dtype | Time per call (ms) | vs original fp32 | Peak memory allocated / reserved | Load (s) | First call (s) |
|---|---|---|---|---|---|---|
| Original fp32mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed call | fp32 | 10.7 | – | 73 MiB/96 MiB | 5.2 | 0.43 |
| Original fp16torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed call | fp16 | 13.8 | 0.78× | 48 MiB/64 MiB | 0.2 | 1.24 |
| Portable fp32 | fp32 | not measured | – | – | – | – |
| Portable | fp16 | 5.74 | 1.9× | 36 MiB/56 MiB | 0.2 | 1.37 |
| Kernels | fp16 | 2.46 | 4.3× | 29 MiB/66 MiB | 0.9 | 1.01 |
| Portable + torch.compile | 3.43 | 3.1× | 64 MiB/94 MiB | 14 (build) | – | |
| TensorRT fp16 | fp16 | 0.962 | 11× | 50 MiBenginenot comparable | 19 (build) | – |
Measured on RTX 5060 Ti 16 GB, torch 2.14.1+cu130, 2026-11-10 to 2026-11-16. First call, 3 warm-up, peak reset, median of 3 rounds x median of 5. Ratios divide two
times of this table (same GPU, same input). Original fp16 is the authors’ own precision option, else torch.autocast. torch.compile and TensorRT: build time instead of load time. TensorRT memory (engine, not comparable) is the GPU memory the engine took, read from the driver because TensorRT allocates outside PyTorch, so it can't be compared with the PyTorch peaks of the other rows.
Other devices
| Device | Backend | dtype | ms / call | Peak memory | Output vs reference | Input |
|---|---|---|---|---|---|---|
| Apple M1 Pro 16 GB MPS, torch 2.14.1, 2026-11-12 | Original | fp32 | 35.9 | 1,813 MiB process | – | 800x1344 |
| Portable | fp16 | 27.0 | 1,545 MiB process | matches | 800x1344 | |
| Portable | fp32 | 28.6 | 1,551 MiB process | matches | 800x1344 | |
| Kernels | fp16 | not implemented | – | 800x1344 | ||
CUDA memory is torch’s peak allocated / reserved; Apple has no such counter, so the Mac rows show the process footprint (weights, Python and torch included). “Output vs reference” compares the device’s output with the stored golden output under the card’s tolerance; – = not checked on that device. Kernels “not implemented” = no kernels for this model on that backend yet: it runs portable there.
Licence and weights
Commercial use allowed
- Code
- Apache-2.0 LICENSE
- Weights
- Apache-2.0 source
- Training data
-
- COCO 2017 (train, instance annotations) CC-BY-4.0
- ImageNet-1k (ILSVRC 2012 classification) noncommercial, non-commercial
Our reading of the licences, not legal advice.
Weights
rtmdet_tiny_8xb32-300e_coco_20220902_112414-78e30dcc.pth 57.5 MB· hosted on OpenMMLabfastforks downloads and checks these files on first use (fastforks download rtmdet-tiny-coco).
Accuracy
The test set through the authors’ code and through ours: same checkpoint, same protocol. The distance between the original’s fp32 and fp16 runs is what half precision alone costs; our runs should stay within it.
COCO val2017
every 10th image, box mAP
| Run | dtype | box mAP ↑ | Δ box mAP vs fp32 original | images | ms / image |
|---|---|---|---|---|---|
| Original fp32 | fp32 | 40.51 | reference | 500 | 39.8 |
| Original fp16 | fp16 | 39.95 | −0.56 | 500 | 44.8 |
| Portable fp32 | fp32 | 40.53 | +0.02 | 500 | 20.4 |
| Portable | fp16 | 40.48 | −0.03 | 500 | 21.1 |
| Kernels | fp16 | 40.52 | +0.01 | 500 | 17.6 |
pycocotools COCOeval of the top 100 detections per image. Original = the authors’ code, same checkpoint and protocol. Measured on original fp32, original fp16, portable, kernels: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-10; portable fp32: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 44640884, 2026-11-17.
Closeness to the original
Beyond the test-set scores: the raw output of ours against the original’s fp32 output on the same input, pixel by pixel. CPU, float32: max |diff| <= 0.0037 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.16 and p99 <= 0.64 (absolute error of the output boxes).
| Check | Tolerance | Measured |
|---|---|---|
| CPU fp32, max |diff| | ≤ 0.0037 | not recorded |
| GPU default dtypes, mean |diff| | ≤ 0.16 | 0.30669 |
| GPU default dtypes, p99 |diff| | ≤ 0.64 | 3.78379 |
Recorded for rtmdet-tiny-coco on RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-11, with these kernels: cudnn_haccfvm_convfvm_extrafvm_mbfvm_r3
Details
RTMDet (Lyu et al. 2022, OpenMMLab): real-time one-stage detector, all convolutions: a CSPNeXt backbone (5 x 5 depthwise convs), a CSPNeXt-PAFPN neck and a head that predicts, per cell of the stride 8 / 16 / 32 maps, a score per class and the distances to the four box edges; class-wise NMS. COCO checkpoints (80 classes) from mmdetection and the person-only detectors mmpose ships as the box source of RTMPose (same backbone family as `fastforks.pose`'s RTMPose). The image is fitted into 640 x 640 (nano: 320 x 320); vendored from mmdetection (no mmdet / mmcv dependency). The portable `module()` is the whole network (class logits and decoded boxes per cell, before NMS).
- Input / output
- Input: one RGB image. Output: boxes, class labels, scores and instance masks.
- Input size
- any H x W >= 32, padded to a multiple of 32; tested up to 3000x4000 (491 MiB allocated)
- Options
score_threshold=0.050.0 to 1.0· changes the outputmax_detections=100>= 1· changes the outputcuda_graphs=true
- Outputs
boxes,scores,labels- Checkpoint
deepen=0.167,widen=0.375,neck_blocks=1,exp_on_reg=false- CUDA kernels
cudnn_haccconvolutions with fp16 accumulation (cuDNN)fvm_convfused convolutions for image classifiersfvm_extrafused transformer row operations (rotary split, SwiGLU with LayerNorm)fvm_mbfused MBConv blocks (MobileNet and EfficientNet families)fvm_r3further fused classifier blocks
- Apple (MPS)
- fp32, fp16
Limitations
- The image is resized like OpenCV 4's 8-bit INTER_LINEAR (what mmdetection calls); OpenCV 5 resizes differently by up to two grey levels, so a reference run on OpenCV 5 differs slightly.
- Boxes are not clamped to the image, as upstream: at the right or bottom border they can reach a few pixels into the padding.
- The person detectors' released config names COCO only; the training config stored in the checkpoints (and the file name) also lists Objects365.
- Kernels keep CUDA graphs: their private pool and static outputs stay reserved, so peak reserved is ~65 MiB above portable (peak allocated is not); `cuda_graphs=false` or `release_graphs()` frees it.
Full reference, generated from the repository
OpenMMLab's RTMDet (tiny, s, m, l, x on COCO; nano and m person detectors): real-time one-stage box detector, a CSPNeXt CNN with NMS. The person detectors supply the boxes for RTMPose. Apache-2.0.
Commercial use: yes.
| variant | input | backend | time (ms) | peak memory (MiB, allocated / reserved) |
|---|---|---|---|---|
rtmdet-tiny-coco | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 10.7 13.8 5.74 2.46 3.43 0.962 | 73 / 96 48 / 64 36 / 56 29 / 66 64 / 94 80 / 122 |
rtmdet-s-coco | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 12.1 14.5 6.13 2.75 3.56 1.19 | 86 / 106 73 / 92 49 / 82 38 / 94 76 / 98 93 / 130 |
rtmdet-m-coco | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 14.7 17.8 7.50 4.27 4.78 2.39 | 168 / 200 168 / 198 92 / 112 75 / 182 115 / 130 156 / 212 |
rtmdet-l-coco | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 19.5 20.4 9.59 5.52 10.6 3.59 | 295 / 316 332 / 356 170 / 210 149 / 286 178 / 226 283 / 318 |
rtmdet-x-coco | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 32.2 24.2 13.1 8.30 11.8 6.17 | 480 / 518 585 / 604 254 / 298 229 / 378 271 / 292 471 / 504 |
rtmdet-nano-person | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 12.9 16.0 7.39 1.82 4.23 0.769 | 14 / 30 12 / 30 9 / 28 9 / 52 48 / 64 49 / 58 |
rtmdet-m-person | synthetic (800x1344) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 14.1 16.3 7.18 3.89 4.44 2.38 | 168 / 200 168 / 198 92 / 112 75 / 182 115 / 128 153 / 190 |
Details
- Original code: https://github.com/open-mmlab/mmdetection at
44ebd17b145c - Code licence: Apache-2.0 (permissive); files:
LICENSE - Weights licence: Apache-2.0
- Licence evidence: https://github.com/open-mmlab/mmdetection/blob/44ebd17b145c2372c4b700bfb9cb20dbd28ab64a/configs/rtmdet/README.md
- Training data: COCO 2017 (train, instance annotations), ImageNet-1k (ILSVRC 2012 classification) (COCO 2017 train (300 epochs); the CSPNeXt backbones start from ImageNet-1k pretraining (configs/rtmdet/README.md))
rtmdet-nano-person: COCO 2017 (train, instance annotations), Objects365 (train, box annotations), ImageNet-1k (ILSVRC 2012 classification) (persons of COCO 2017 train and Objects365 train (100 epochs; the training config stored in the checkpoint); the CSPNeXt backbone starts from ImageNet-1k pretraining)rtmdet-m-person: COCO 2017 (train, instance annotations), Objects365 (train, box annotations), ImageNet-1k (ILSVRC 2012 classification) (persons of COCO 2017 train and Objects365 train (100 epochs; the training config stored in the checkpoint); the CSPNeXt backbone starts from ImageNet-1k pretraining)
- Options:
score_threshold= 0.05 (0.0 to 1.0),max_detections= 100 (>= 1),cuda_graphs= True (speed only) - Kernel libraries:
cudnn_hacc,fvm_conv,fvm_extra,fvm_mb,fvm_r3 - Variants:
rtmdet-tiny-coco,rtmdet-s-coco,rtmdet-m-coco,rtmdet-l-coco,rtmdet-x-coco,rtmdet-nano-person,rtmdet-m-person - Outputs: boxes, scores, labels; label_set:
coco-80 - Input size: any H x W >= 32, padded to a multiple of 32
- Tested up to (portable, fp16, RTX 5060 Ti 16 GB):
rtmdet-tiny-coco3000x4000 (0.5 GiB peak allocated, 0.5 reserved);rtmdet-s-coco3000x4000 (0.5 GiB peak allocated, 0.5 reserved);rtmdet-m-coco3000x4000 (0.5 GiB peak allocated, 0.5 reserved);rtmdet-l-coco3000x4000 (0.6 GiB peak allocated, 0.6 reserved);rtmdet-x-coco3000x4000 (0.7 GiB peak allocated, 0.7 reserved);rtmdet-nano-person3000x4000 (0.4 GiB peak allocated, 0.5 reserved);rtmdet-m-person3000x4000 (0.5 GiB peak allocated, 0.5 reserved) rtmdet-tiny-coco800x1344, first call: portable + torch.compile 14 s (cold Inductor cache), TensorRT fp16 19 s (ONNX + engine).rtmdet-s-coco800x1344, first call: portable + torch.compile 14 s (cold Inductor cache), TensorRT fp16 20 s (ONNX + engine).rtmdet-m-coco800x1344, first call: portable + torch.compile 16 s (cold Inductor cache), TensorRT fp16 22 s (ONNX + engine).rtmdet-l-coco800x1344, first call: portable + torch.compile 17 s (cold Inductor cache), TensorRT fp16 25 s (ONNX + engine).rtmdet-x-coco800x1344, first call: portable + torch.compile 19 s (cold Inductor cache), TensorRT fp16 28 s (ONNX + engine).rtmdet-nano-person800x1344, first call: portable + torch.compile 16 s (cold Inductor cache), TensorRT fp16 23 s (ONNX + engine).rtmdet-m-person800x1344, first call: portable + torch.compile 16 s (cold Inductor cache), TensorRT fp16 25 s (ONNX + engine).rtmdet-tiny-cocooriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-tiny-cocooriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-s-cocooriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-s-cocooriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-m-cocooriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-m-cocooriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-l-cocooriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-l-cocooriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-x-cocooriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-x-cocooriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-nano-personoriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-nano-personoriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-m-personoriginal (fp32) 800x1344: mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-m-personoriginal 800x1344: torch.autocast(float16) around the call; mmdet 3.3.0 inference_detector (CPU cv2 resize + pad, network, decode, NMS: torchvision nms in mmcv's batched_nms logic), score_thr 0.05, 100 detections (card defaults), all in the timed callrtmdet-tiny-cocoload / first call: original 0.22 s / 1.242 s, portable 0.31 s / 1.539 s, kernels 2.12 s / 0.261 srtmdet-s-cocoload / first call: original 0.27 s / 0.581 s, portable 0.28 s / 1.582 s, kernels 1.81 s / 0.212 srtmdet-m-cocoload / first call: original 0.46 s / 0.313 s, portable 0.72 s / 1.304 s, kernels 1.66 s / 1.197 srtmdet-l-cocoload / first call: original 0.68 s / 0.323 s, portable 0.99 s / 1.292 s, kernels 2.43 s / 1.167 srtmdet-x-cocoload / first call: original 1.0 s / 0.608 s, portable 1.05 s / 1.256 s, kernels 2.39 s / 1.162 srtmdet-nano-personload / first call: original 0.14 s / 2.695 s, portable 0.19 s / 2.196 s, kernels 4.13 s / 0.602 srtmdet-m-personload / first call: original 0.51 s / 0.025 s, portable 0.31 s / 0.389 s, kernels 1.0 s / 0.246 s- Protocol, first call, 3 warm-up, peak reset, median of 3 rounds x median of 5
Measured
pycocotools COCOeval of the top 100 detections per image. Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp32 | original fp16 | portable | portable fp32 | kernels | note |
|---|---|---|---|---|---|---|---|
rtmdet-tiny-coco | COCO val2017, every 10th image, box mAP | 40.51 | 39.95 | 40.48 | 40.53 | 40.52 | |
rtmdet-s-coco | COCO val2017, every 10th image, box mAP | 45.36 | 44.58 | 45.27 | 45.35 | 45.37 | |
rtmdet-m-coco | COCO val2017, every 10th image, box mAP | 49.11 | 49.14 | 49.11 | 49.15 | 49.14 | |
rtmdet-l-coco | COCO val2017, every 10th image, box mAP | 52.37 | 52.39 | 52.37 | 52.38 | 52.41 | |
rtmdet-x-coco | COCO val2017, every 10th image, box mAP | 52.79 | 52.79 | 52.79 | 52.77 | 52.85 | |
rtmdet-nano-person | COCO val2017, every 10th image, box AP of the person class | 36.86 | 36.79 | 36.91 | 36.86 | 36.94 | |
rtmdet-m-person | COCO val2017, every 10th image, box AP of the person class | 60.20 | 60.21 | 60.20 | 60.21 | 60.21 |
Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-10; portable32: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 44640884, 2026-11-17; kernels: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-10.
Parity
CPU, float32: max |diff| <= 0.0037 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.16 and p99 <= 0.64 (absolute error of the output boxes).
| variant | kernels error vs original, mean / p99 | GPU, torch | kernels used |
|---|---|---|---|
rtmdet-tiny-coco | 0.30669 / 3.78379 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-s-coco | 0.96654 / 9.43566 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-m-coco | 0.09531 / 0.32148 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-l-coco | 0.23124 / 2.21498 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-x-coco | 0.07914 / 0.24827 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-nano-person | 0.10524 / 0.42335 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
rtmdet-m-person | 0.12766 / 0.96977 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, fvm_conv@f1d064e539e1:sm_120, fvm_extra@3208ce525a91:sm_120, fvm_mb@720dd3f29bf7:sm_120, fvm_r3@f54eb890e23c:sm_120 |
Limitations
- The image is resized like OpenCV 4's 8-bit INTER_LINEAR (what mmdetection calls); OpenCV 5 resizes differently by up to two grey levels, so a reference run on OpenCV 5 differs slightly.
- Boxes are not clamped to the image, as upstream: at the right or bottom border they can reach a few pixels into the padding.
- The person detectors' released config names COCO only; the training config stored in the checkpoints (and the file name) also lists Objects365.
- Kernels keep CUDA graphs: their private pool and static outputs stay reserved, so peak reserved is ~65 MiB above portable (peak allocated is not);
cuda_graphs=falseorrelease_graphs()frees it.
Example output
One fixed input through this checkpoint, the same for every model of the task (kernels · float16 · NVIDIA GeForce RTX 5060 Ti).
- person 0.74
Input: Astronaut Eileen Collins, NASA (public domain).