danielhavir

ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAINhqM2fhXk0Qgl+6R9kPjJ/48ZrNcXjc/7zTn8oi88nH daniel@Daniels-MacBook-Pro.local

danielhavir

class LoRAConv1DWrapper(nn.Module):
    """SimpleWrapper class that implements LoRA: Low-Rank Adaptation of Large Language Models.
    
    Arxiv link: https://arxiv.org/abs/2106.09685"""
    def __init__(self, conv1dmodule: transformers.pytorch_utils.Conv1D, lora_rank: int):
        super().__init__()

        self.base_module = conv1dmodule
        d1, d2 = self.base_module.weight.size()
        self.A = nn.Parameter(

danielhavir

import argparse
import multiprocessing as mp
from functools import partial

import numpy as np
from scipy.stats import norm
import matplotlib.pyplot as plt

STEPS = 10000

danielhavir

# Full Example: https://gist.github.com/danielhavir/407a6cfd592dfc2ad1e23a1ed3539e07
import os
from typing import Callable, List, Tuple, Generator, Dict

import torch
import torch.utils.data
from PIL.Image import Image as ImageType


def list_items_local(path: str) -> List[str]:

danielhavir

au BufNewFile,BufRead *.py set tabstop=4
au BufNewFile,BufRead *.py set softtabstop=4
au BufNewFile,BufRead *.py set shiftwidth=4
au BufNewFile,BufRead *.py set textwidth=79
au BufNewFile,BufRead *.py set expandtab
au BufNewFile,BufRead *.py set autoindent
au BufNewFile,BufRead *.py set fileformat=unix

highlight BadWhitespace ctermbg=red guibg=red
au BufRead,BufNewFile *.py,*.pyw,*.c,*.h match BadWhitespace /\s\+$/

danielhavir

from visdom import Visdom
import numpy as np
from time import sleep
import logging
from threading import Event
try:
    import psutil
except ImportError:
    logging.error("You must \"pip install psutil\"")
    raise

danielhavir

def style_loss(style_features, generated_features, J):
    loss_style = 0.
    for j in J:
        gram_style = gram_matrix(style_features[j])
        gram_gen = gram_matrix(generated_features[j])
        loss_style += F.mse_loss(gram_gen, gram_style)
    return loss_style

danielhavir

def content_loss(content_features, generated_features, j):
    loss_content = F.mse_loss(generated_features[j], content_features[j])
    return loss_content

danielhavir

def gram_matrix(feature_map):
    C, H, W = feature_map.size()
    reshaped_map = feature_map.view(C, H*W)
    G = reshaped_map.mm(reshaped_map.t()) / (C * H * W)
    return G

danielhavir

package main

import (
	"crypto/rand"
	"flag"
	"fmt"
	"io/ioutil"
	"log"
	"net/http"
	"os/exec"