raphael-sch

import re
import sys
import time

import tqdm
import torch
from datasets import load_dataset, concatenate_datasets
from transformers import AutoTokenizer, LlamaForCausalLM

xenova

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

grey-area

import torch
import math
from torch import Tensor
from typing import Optional


def get_relative_positional_encoding(length1:int, length2:int, d_model:int, device:torch.device):
    xs = torch.arange(length1, device=device).unsqueeze(1)
    ys = torch.arange(length2, device=device).unsqueeze(0)

matpalm

import numpy as np

def slerp(p0, p1, n):
    # https://en.wikipedia.org/wiki/Slerp
    norm = np.linalg.norm(p0) * np.linalg.norm(p1)
    dot = np.sum(p0 * p1 / norm)
    theta_0 = np.arccos(dot)
    sin_theta_0 = np.sin(theta_0)
    interp = []
    for t in np.linspace(0.0, 1.0, n):

ashleve

from pytorch_lightning import LightningDataModule
from torch_geometric.datasets import TUDataset
from torch_geometric.data import DataLoader
from sklearn.model_selection import KFold


class ProteinsKFoldDataModule(LightningDataModule):
    def __init__(
            self,
            data_dir: str = "data/",

jiahao87

"""Script for fine-tuning Pegasus
Example usage:
  # use XSum dataset as example, with first 1000 docs as training data
  from datasets import load_dataset
  dataset = load_dataset("xsum")
  train_texts, train_labels = dataset['train']['document'][:1000], dataset['train']['summary'][:1000]
  
  # use Pegasus Large model as base for fine-tuning
  model_name = 'google/pegasus-large'
  train_dataset, _, _, tokenizer = prepare_data(model_name, train_texts, train_labels)

rbrigden

abdominocardi.ac
autotr.actor
cephalotr.actor
cocontr.actor
coen.actor
cornf.actor
counter.actor
effr.actor
idemf.actor
lithofr.actor