Image-to-Video
TensorRT
ONNX
Safetensors
PyTorch
English
Arabic
audio-to-video
talking-head
portrait-animation
lip-sync
flow-matching
motion-generation
video-generation
wav2vec2
real-time
Instructions to use mohamedemam/float with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- TensorRT
How to use mohamedemam/float with TensorRT:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Download onnx/backends.generated.yaml from mohamedemam/float: direct link, hf CLI and curl.
- Browser
- Download file 4.18 kB
-
https://huggingface.co/mohamedemam/float/resolve/main/onnx/backends.generated.yaml
- Command line
-
hf download hf://mohamedemam/float/onnx/backends.generated.yaml
-
curl -L -o backends.generated.yaml https://huggingface.co/mohamedemam/float/resolve/main/onnx/backends.generated.yaml
4.18 kB
| version: 2 | |
| target: | |
| torch: 2.7.1+cu126 | |
| cuda: true | |
| gpu: NVIDIA RTX A6000 | |
| cc: '8.6' | |
| vram_gb: 47.4 | |
| ort: 1.23.2 | |
| trt_ep_listed: true | |
| trt_libs_loadable: true | |
| trt: true | |
| cuda_ep: true | |
| objective: end_to_end_ms | |
| default_mode: eager | |
| chosen_combo: | |
| emotion: onnx | |
| audio: onnx | |
| fmt: onnx | |
| refenc: compile | |
| e2e: | |
| kind: global_generator | |
| eager_ms: 18.77132902154699 | |
| eager_mad_ms: 0.012680015061050653 | |
| compile_ms: 16.538165014935657 | |
| compile_mad_ms: 0.04346101195551455 | |
| compile_s: 3.469 | |
| speedup: 1.1350309423442417 | |
| significant: true | |
| components: | |
| emotion: | |
| mode: onnx | |
| provider: trt | |
| onnx_path: onnx/emotion_v1.onnx | |
| metrics: | |
| eager_ms: 16.313770989654586 | |
| eager_mad_ms: 0.2377640048507601 | |
| compile_ms: 12.844707001931965 | |
| compile_mad_ms: 0.22261301637627184 | |
| compile_s: 3.432 | |
| onnx_parity: | |
| max_abs: 0.000633 | |
| mean_abs: 0.000164 | |
| cosine: 1.0 | |
| ok: true | |
| onnx_ms: 2.843594003934413 | |
| onnx_mad_ms: 0.004879984771832824 | |
| onnx_provider: trt | |
| onnx_providers: | |
| - TensorrtExecutionProvider | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |
| audio: | |
| mode: onnx | |
| provider: trt | |
| onnx_path: onnx/wav2vec_v1.onnx | |
| metrics: | |
| eager_ms: 7.668967999052256 | |
| eager_mad_ms: 0.03063099575228989 | |
| compile_ms: 4.126863001147285 | |
| compile_mad_ms: 0.03980001201853156 | |
| compile_s: 6.241 | |
| onnx_parity: | |
| max_abs: 0.015054 | |
| mean_abs: 0.000658 | |
| cosine: 0.99999 | |
| ok: true | |
| onnx_ms: 1.743127009831369 | |
| onnx_mad_ms: 0.004930014256387949 | |
| onnx_provider: trt | |
| onnx_providers: | |
| - TensorrtExecutionProvider | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |
| direction: | |
| mode: eager | |
| onnx_path: onnx/direction_v1.onnx | |
| metrics: | |
| eager_ms: 0.037949997931718826 | |
| eager_mad_ms: 0.0008099887054413557 | |
| compile_ms: 0.07996099884621799 | |
| compile_mad_ms: 0.001040025381371379 | |
| compile_s: 0.025 | |
| onnx_parity: | |
| max_abs: 0.000194 | |
| mean_abs: 3.9e-05 | |
| cosine: 1.0 | |
| ok: true | |
| onnx_ms: 0.181223003892228 | |
| onnx_mad_ms: 0.003029999788850546 | |
| onnx_provider: trt | |
| onnx_providers: | |
| - TensorrtExecutionProvider | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |
| reason: no candidate beat eager by 1.10x outside the noise band | |
| fmt: | |
| mode: onnx | |
| provider: trt | |
| onnx_path: onnx/fmt_step_v1.onnx | |
| metrics: | |
| eager_ms: 5.844598985277116 | |
| eager_mad_ms: 0.013921002391725779 | |
| compile_ms: 4.951315990183502 | |
| compile_mad_ms: 0.020679988665506244 | |
| compile_s: 3.774 | |
| onnx_parity: | |
| max_abs: 0.020788 | |
| mean_abs: 0.002348 | |
| cosine: 0.999996 | |
| ok: true | |
| onnx_ms: 2.2519149933941662 | |
| onnx_mad_ms: 0.01677899854257703 | |
| onnx_provider: trt | |
| onnx_providers: | |
| - TensorrtExecutionProvider | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |
| decoder: | |
| mode: eager | |
| onnx_path: onnx/decoder_v1.onnx | |
| metrics: | |
| eager_ms: 25.683184998342767 | |
| eager_mad_ms: 0.08247100049629807 | |
| compile_ms: 24.97261398821138 | |
| compile_mad_ms: 0.13203200069256127 | |
| compile_s: 9.272 | |
| onnx_parity: | |
| max_abs: 0.673292 | |
| mean_abs: 0.003701 | |
| cosine: 0.999929 | |
| ok: true | |
| onnx_ms: 149.67813200200908 | |
| onnx_mad_ms: 2.0826529944315553 | |
| onnx_provider: cuda | |
| onnx_providers: | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |
| reason: no candidate beat eager by 1.10x outside the noise band | |
| refenc: | |
| mode: compile | |
| onnx_path: onnx/refenc_v1.onnx | |
| metrics: | |
| eager_ms: 5.775298981461674 | |
| eager_mad_ms: 0.004929985152557492 | |
| compile_ms: 5.231279996223748 | |
| compile_mad_ms: 0.04106998676434159 | |
| compile_s: 3.105 | |
| onnx_parity: | |
| max_abs: 0.091488 | |
| mean_abs: 0.007204 | |
| cosine: 0.999985 | |
| ok: true | |
| onnx_ms: 11.51109699276276 | |
| onnx_mad_ms: 0.11942200944758952 | |
| onnx_provider: trt | |
| onnx_providers: | |
| - TensorrtExecutionProvider | |
| - CUDAExecutionProvider | |
| - CPUExecutionProvider | |