facebookincubator/AITemplate
AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.
PythonApache-2.0
Issues
- 2
Got cutlass error: Error Internal at: 214
#998 opened by syntrive - 0
FileNotFoundError: [Errno 2] No such file or directory: \\ComfyUI\\custom_nodes\\AITemplate-main\\__init__.py'
#1027 opened by DuckersMcQuack - 0
Called SetConstant on, but can't find in either bound or unbound constant set
#1024 opened by ADongGu - 0
Grouped Transposed Convolutions cutlass errror
#1021 opened by jonpryai - 4
<class 'src.pipeline_stable_diffusion_ait.StableDiffusionAITPipeline'> is incorrectly implemented. Expected {'feature_extractor', 'scheduler', 'tokenizer', 'text_encoder', 'safety_checker', 'unet', 'vae'} to be defined
#975 opened by jiagaoxiang - 7
Confused on the shape of input Tensor
#1009 opened by ningmenghongcha - 2
- 1
model_interface.cu:231: Error: Constant pretrained_model_patch_embed_proj_weight was not set! Set the value with set_constant.
#999 opened by ADongGu - 1
Can i remove vae?
#933 opened by Boom-Hacker - 2
error during inferencing: Error: Constant embeddings_token_embedding_weight was not set! Set the value with set_constant.
#968 opened by mengbingrock - 1
Does Concatenate order matters?
#995 opened by ecilay - 5
windows platform cannot link _binary_constants_bin_end and _binary_constants_bin_start
#990 opened by joye - 0
- 4
multi-gpu at runtime error
#988 opened by ecilay - 1
Does AIT handle if/else in forward function?
#983 opened by jiangwei221 - 6
`Unsupported workload for this conv2d specialization` when using dynamic shape together with permute
#981 opened by jiangwei221 - 2
gcc: internal compiler error
#980 opened by jiangwei221 - 3
- 1
- 2
- 18
Low performance from unnecessary permutations
#936 opened by jonpryai - 4
Stable Diffusion (GLIGEN) Download Error
#955 opened by isouf - 1
- 2
complie controlnet error
#949 opened by dushwe - 1
Compile Diffusers Community Pipelines
#868 opened by djj0s3 - 6
- 9
- 1
- 3
AMD MI210 ResNet test error
#924 opened by crispyberry - 3
- 3
GFX1100 Support
#908 opened by clayscode - 3
sm90/sm90a Hopper architecture Incompatibility
#926 opened by ConsceIeratus - 1
I try to make it work on gfx1030,but
#930 opened by Boom-Hacker - 1
- 3
- 1
fx2ait low performance
#921 opened by Oldpan - 1
Ops unit tests fail on ROCm
#920 opened by duli2012 - 1
making template of sdxl
#912 opened by JAVerma - 1
Add Support for intel arc GPUs 🥹
#906 opened by x-legion - 4
Does AIT support BF16 inference now?
#899 opened by sanbuphy - 2
Could AITemplate support mix precision inference?
#890 opened by sanbuphy - 3
- 2
Slicing tensor along a fixed dimension
#869 opened by CanyonWind - 1
Any plan on supporting Flash Attention 2
#856 opened by CanyonWind - 0
- 0
Error in installing fx2ait setup.py
#884 opened by MahdiMohseni0033 - 4
Option for choosing fp32 gemm backend implementation
#872 opened by zhekunz2 - 1
Dynamic Resolution for Control-Net
#870 opened by MahdiMohseni0033 - 5
How do I make stable diffusion with AIT work with any latent resolution?
#867 opened by comfyanonymous - 2