Recursion in Fastai library

James_Ying · December 29, 2018, 4:26am

I don’t understand how this Recursion work

stephenjohnson · December 29, 2018, 5:18am

That’s because it isn’t recursion. It’s calling the predict function from model.py

def predict(m, dl):
    preda,_ = predict_with_targs_(m, dl)
    return np.concatenate(preda)

James_Ying · December 29, 2018, 9:50am

How can I find where this function defined?

stephenjohnson · January 1, 2019, 8:26pm

Line 248 of model.py

github.com

fastai/fastai/blob/master/old/fastai/model.py

from .imports import *
from .torch_imports import *
from .core import *
from .layer_optimizer import *
from .swa import *
from .fp16 import *

IS_TORCH_04 = LooseVersion(torch.__version__) >= LooseVersion('0.4')

def cut_model(m, cut):
    return list(m.children())[:cut] if cut else [m]

def predict_to_bcolz(m, gen, arr, workers=4):
    arr.trim(len(arr))
    lock=threading.Lock()
    m.eval()
    for x,*_ in tqdm(gen):
        y = to_np(m(VV(x)).data)
        with lock:
            arr.append(y)

This file has been truncated. show original