yoavg

Reinforcement Learning for Language Models

Yoav Goldberg, April 2023.

Why RL?

With the release of the ChatGPT model and followup large language models (LLMs), there was a lot of discussion of the importance of "RLHF training", that is, "reinforcement learning from human feedback". I was puzzled for a while as to why RL (Reinforcement Learning) is better than learning from demonstrations (a.k.a supervised learning) for training language models. Shouldn't learning from demonstrations (or, in language model terminology "instruction fine tuning", learning to immitate human written answers) be sufficient? I came up with a theoretical argument that was somewhat convincing. But I came to realize there is an additional argumment which not only supports the case of RL training, but also requires it, in particular for models like ChatGPT. This additional argument is spelled out in (the first half of) a talk by John Schulman from OpenAI. This post pretty much

FrancescoSaverioZuppichini

version: '3.8'
services:
  caddy:
    container_name: caddy
    image: caddy
    ports:
      #     - '80:80'
      - '443:443'
    restart: unless-stopped
    volumes:

yzh119

from __future__ import print_function
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable

def sample_gumbel(shape, eps=1e-20):
    U = torch.rand(shape).cuda()
    return -Variable(torch.log(-torch.log(U + eps) + eps))

karpathy

"""
A bare bones examples of optimizing a black-box function (f) using
Natural Evolution Strategies (NES), where the parameter distribution is a 
gaussian of fixed standard deviation.
"""

import numpy as np
np.random.seed(0)

# the function we want to optimize

aras-p

// Just before switching jobs:
// Add one of these.
// Preferably into the same commit where you do a large merge.
//
// This started as a tweet with a joke of "C++ pro-tip: #define private public",
// and then it quickly escalated into more and more evil suggestions.
// I've tried to capture interesting suggestions here.
//
// Contributors: @r2d2rigo, @joeldevahl, @msinilo, @_Humus_,
// @YuriyODonnell, @rygorous, @cmuratori, @mike_acton, @grumpygiant,