search

mljs / libsvm

LIBSVM for the browser and nodejs :fire:
https://mljs.github.io/libsvm/
BSD 3-Clause "New" or "Revised" License
82 stars 14 forks source link

RuntimeError: memory access out of bounds #16

Open chidokun opened 5 years ago

chidokun commented 5 years ago

I have a big problem when loading a model string from file for the first run. libsvm can not deserialize model because of the following RuntimeError. But when I try training a dataset and then load my model from file, it's ok. 

RuntimeError: memory access out of bounds
    at wasm-function[84]:51
    at wasm-function[256]:11
    at SVM.Module._deserialize_model (D:\Documents\Thesis\sources\verizone-api\node_modules\libsvm-js\out\wasm\libsvm.js:79:107637)
    at ccall (D:\Documents\Thesis\sources\verizone-api\node_modules\libsvm-js\out\wasm\libsvm.js:79:4914)
    at D:\Documents\Thesis\sources\verizone-api\node_modules\libsvm-js\out\wasm\libsvm.js:79:5295
    at Function.load (D:\Documents\Thesis\sources\verizone-api\node_modules\libsvm-js\src\loadSVM.js:252:19)
    at Function.<anonymous> (D:\Documents\Thesis\sources\verizone-api\src\ml\MLModel.ts:17:30)
    at Generator.next (<anonymous>)
    at fulfilled (D:\Documents\Thesis\sources\verizone-api\src\ml\MLModel.ts:4:58)

Any solution?

stropitek commented 5 years ago

Hello @chidokun,

What exactly is the difference between the case that works and the one that doesn't?

Can you provide a model file and a code snippet that reproduces the error?

AndreBreh commented 5 years ago

i have the same error. The difference between the case that works and the one that doesn't is the file size. I tested it with a serialized String of size 1,5mb and it works. Then i used a file with size 15mb and it doesnt work

therealglazou commented 4 years ago

I confirm this issue and the diagnostic above. Beyond a certain size, loading a model does not work any more. I'm unable to say if it's because serialization has generated an incorrect output or if it's a mem size issue on load. I would bet on the latter.