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-to_tensor",
  "dataComponent": "gpu",
  "heading": {
    "title": "to_tensor",
    "badges": ["GPU"]
  },
  "synopsis": "Converts an int2D_array or float2D_array into a GPU-ready float32 tensor (rank=2). Logs an error if shape is inconsistent or data is empty.",
  "codeBlocks": [
    "extend(\"gpu\")\\n\\n# 1) Convert 2D int array => rank=2 float32 tensor\\narrI = [[1,2],[3,4],[5,6]]\\nT = gpu:to_tensor(arrI)\\nslog(T)\\n# => TensorValue(...) with shape=[3,2], elem_type=Float32\\n\\n# 2) Convert 2D float array => same concept\\narrF = [[1.1,2.2],[3.3,4.4]]\\nTF = gpu:to_tensor(arrF)\\nslog(TF)\\n# => shape=[2,2], data as float32\\n\\n# If bridging sees an empty array or ragged rows, it raises an error.\\n"
  ],
  "notes": [
    "gpu:to_tensor => exactly 1 argument: a 2D array (either int2D_array or float2D_array). The resulting shape is [rows, cols].",
    "All values become float32. If rows differ in length, bridging fails with an 'inconsistent row lengths' error.",
    "Used as the first step for GPU ops like multiply, add, etc."
  ]
}