mumu 0.10.0

Lava Mumu is a language for those in the now and that know
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
{
  "id": "fn-gpu-apply_activation",
  "dataComponent": "gpu",
  "heading": {
    "title": "apply_activation",
    "badges": ["GPU"]
  },
  "synopsis": "Applies a specified activation function (e.g., 'relu', 'sigmoid', 'tanh') elementwise to a tensor on the GPU.",
  "codeBlocks": [
    "extend(\"gpu\")\n\nA = gpu:to_tensor([[1.0,-2.0],[3.0,-4.5]])\nB = gpu:apply_activation(\"relu\", A)\n# => [[1.0,0.0],[3.0,0.0]]\nsput(B)\n\nC = gpu:apply_activation(\"sigmoid\", A)\n# => each element => 1 / (1 + exp(-x))\n\n# Use cases:\n# 1) Neural network forward pass for hidden layers.\n# 2) Custom elementwise activation in GPU-based data pipelines."
  ],
  "notes": [
    "gpu:apply_activation expects 2 arguments: (Str activationName, Tensor).",
    "Supported strings might be 'relu', 'sigmoid', 'tanh', or others depending on bridging. If unknown, raises an error.",
    "Returns a new tensor of the same shape, with each element replaced by the activation function result."
  ]
}