srush

import numpy as np
from numpy import ndarray
from chalk import *
from jaxtyping import Float
from chalk.transform import Batched
import numpy.random


def create_t_shaped_data(num_points: int) -> Float[ndarray, "num_points 2"]:
    """Create random 2D data points shaped like a T."""

srush

""" Some language modeling code.

python lm.py  --model LstmLm --devid 1 --lr 0.01 --clip 2 --optim Adam --nlayers 2 --nhid 512 --dropout 0.5 --epochs 50 --bsz 128 --bptt 32 --tieweights
Train: 42.90723478099889, Valid: 75.74921810452948, Test: 72.84913233987702

python lm.py  --model NnLm --devid 3 --lr 0.001 --clip 0 --optim Adam --nlayers 2 --nhid 512 --dropout 0.5 --epochs 20 --bsz 64 --bptt 64
Train: 71.58999479091389, Valid: 158.07431086368382, Test: 146.13046578572258

Tested on torch.__version__ == 0.3.1b0+2b47480

srush

import torch
from torch.autograd import Variable

q = Variable(torch.Tensor(5), requires_grad=True)
leaves = [Variable(torch.zeros(5, 5), requires_grad=True) for _ in range(10)]
intermediates = [l  for i, l in enumerate(leaves)]
loss = sum(q * v * i for i, v in enumerate(intermediates)).sum()

# define a helper for dividing intermediates into groups
def group(l, group_size):

srush

import torch
from torch.autograd import Variable

leaves = [Variable(torch.zeros(5, 5), requires_grad=True) for _ in range(10)]
intermediates = [l + i for i, l in enumerate(leaves)]
loss = sum(v * i for i, v in enumerate(intermediates)).sum()

# define a helper for dividing intermediates into groups
def group(l, group_size):
    """Groups l into chunks of size group_size.

srush

---------------------------------------------------------------------------
Reviewer's Scores
---------------------------------------------------------------------------

                         Appropriateness: 5
                                 Clarity: 5
                             Originality: 3
                 Soundness / Correctness: 4
               Impact of Ideas / Results: 4
                   Meaningful Comparison: 4