Instructions to use t-tech/flex-sae with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use t-tech/flex-sae with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("t-tech/flex-sae") - Notebooks
- Google Colab
- Kaggle
| # /// script | |
| # dependencies = [ | |
| # "torch", | |
| # "numpy", | |
| # "kernels", | |
| # ] | |
| # /// | |
| import torch | |
| import numpy as np | |
| from kernels import get_kernel | |
| flex = get_kernel("t-tech/flex-sae") #Fast Kernels | |
| def hierarchical_sae_loss( | |
| indices: torch.Tensor, # [B, K] | |
| weight: torch.Tensor, # [F, D] | |
| vals: torch.Tensor, # [B, K] | |
| bias: torch.Tensor, # [D] | |
| target: torch.Tensor, # [B, D] | |
| ) -> torch.Tensor: | |
| emb = weight[indices].to(torch.float32) # [K, D] | |
| recon_cum = bias.to(torch.float32) + (emb * vals.unsqueeze(-1)).cumsum(dim=1) | |
| diff = recon_cum.to(torch.float32) - target.to(torch.float32).unsqueeze(1) | |
| loss = diff.pow(2).mean() | |
| return loss | |
| B = 2048 | |
| K = 256 | |
| F = 1024 * 128 | |
| D = 1024 | |
| WARMUP = 5 | |
| NUM_ITER = 100 | |
| dtype = torch.float32 | |
| vals = None | |
| decoder = None | |
| bias = None | |
| target = None | |
| indices = None | |
| def init_parameters(): | |
| global vals, decoder, bias, target, indices | |
| vals = torch.randn(B, K, dtype=dtype, device="cuda").abs().requires_grad_() | |
| decoder = torch.randn(F, D, dtype=dtype, device="cuda", requires_grad=True) | |
| bias = torch.randn(D, dtype=dtype, device="cuda", requires_grad=True) | |
| target = torch.randn(B, D, dtype=dtype, device="cuda") | |
| indices = torch.randint(0, F, (B, K), dtype=torch.long, device="cuda") | |
| timing_kernel = [] | |
| timing_vanilla = [] | |
| torch.cuda.reset_peak_memory_stats() | |
| loss_kernel_list = torch.zeros((100,)) | |
| loss_vanilla_list = torch.zeros((100,)) | |
| def zero_grad(): | |
| vals.grad = None | |
| decoder.grad = None | |
| bias.grad = None | |
| torch.cuda.empty_cache() | |
| for i in range(NUM_ITER + WARMUP): | |
| init_parameters() | |
| start_kernel = torch.cuda.Event(enable_timing=True) | |
| end_kernel = torch.cuda.Event(enable_timing=True) | |
| start_vanilla = torch.cuda.Event(enable_timing=True) | |
| end_vanilla = torch.cuda.Event(enable_timing=True) | |
| start_kernel.record() | |
| loss_kernel = flex.triton_hierarchical_sae_loss(indices, decoder, vals, bias, target) | |
| loss_kernel.backward() | |
| end_kernel.record() | |
| zero_grad() | |
| start_vanilla.record() | |
| loss_vanilla = hierarchical_sae_loss(indices, decoder, vals, bias, target) | |
| loss_vanilla.backward() | |
| end_vanilla.record() | |
| if i >= WARMUP: | |
| torch.cuda.synchronize() | |
| timing_kernel.append(start_kernel.elapsed_time(end_kernel)) | |
| timing_vanilla.append(start_vanilla.elapsed_time(end_vanilla)) | |
| loss_kernel_list[i-WARMUP] = loss_kernel.detach() | |
| loss_vanilla_list[i-WARMUP] = loss_vanilla.detach() | |
| zero_grad() | |
| if torch.allclose(loss_kernel, loss_vanilla): | |
| print("β Outputs are close! Everything is good! π") | |
| else: | |
| print("β Outputs mismatch... β οΈπ€") | |
| print(f"π¦ Triton Kernel Time (Ours): {np.mean(timing_kernel):.4f} Β± {np.std(timing_kernel):.4f} ms") | |
| print(f"π₯ Torch Compile Kernel Time: {np.mean(timing_vanilla):.4f} Β± {np.std(timing_vanilla):.4f} ms") | |
| print(f"π Speedup: {np.mean(timing_vanilla) / np.mean(timing_kernel):.2f}x") |