Ben Krause

Erik Nijkamp, Tian Xie, Hiroaki Hayashi et al. · cmu, microsoft-research

Large Language Models (LLMs) have become ubiquitous across various domains, transforming the way we interact with information and conduct research. However, most high-performing LLMs remain confined behind proprietary walls, hindering scientific progress. Most open-source LLMs, on the other hand, are limited in their ability to support longer sequence lengths, which is a key requirement for many tasks that require inference over an input context. To address this, we have trained XGen, a series of 7B parameter models on up to 8K sequence length for up to 1.5T tokens. We have also finetuned the XGen models on public-domain instructional data, creating their instruction-tuned counterparts (XGen-Inst). We open-source our models for both research advancements and commercial applications. Our evaluation on standard benchmarks shows that XGen models achieve comparable or better results when compared with state-of-the-art open-source LLMs. Our targeted evaluation on long sequence modeling tasks shows the benefits of our 8K-sequence models over 2K-sequence open-source LLMs.

Ben Krause, Lucia Chen, Emmanuel Kahembwe

We introduce the AutoGRAMS framework for programming multi-step interactions with language models. AutoGRAMS represents AI agents as a graph, where each node can execute either a language modeling instruction or traditional code. Likewise, transitions in the graph can be governed by either language modeling decisions or traditional branch logic. AutoGRAMS supports using variables as memory and allows nodes to call other AutoGRAMS graphs as functions. We show how AutoGRAMS can be used to design highly sophisticated agents, including self-referential agents that can modify their own graph. AutoGRAMS's graph-centric approach aids interpretability, controllability, and safety during the design, development, and deployment of AI agents. We provide our framework as open source at https://github.com/autograms/autograms .

Alvin Chan, Ali Madani, Ben Krause et al.

Attribute extrapolation in sample generation is challenging for deep neural networks operating beyond the training distribution. We formulate a new task for extrapolation in sequence generation, focusing on natural language and proteins, and propose GENhance, a generative framework that enhances attributes through a learned latent space. Trained on movie reviews and a computed protein stability dataset, GENhance can generate strongly-positive text reviews and highly stable protein sequences without being exposed to similar data during training. We release our benchmark tasks and models to contribute to the study of generative modeling extrapolation and data-driven design in biology and chemistry.

Nazneen Fatema Rajani, Ben Krause, Wengpeng Yin et al.

Interpretability techniques in NLP have mainly focused on understanding individual predictions using attention visualization or gradient-based saliency maps over tokens. We propose using k nearest neighbor (kNN) representations to identify training examples responsible for a model's predictions and obtain a corpus-level understanding of the model's behavior. Apart from interpretability, we show that kNN representations are effective at uncovering learned spurious associations, identifying mislabeled examples, and improving the fine-tuned model's performance. We focus on Natural Language Inference (NLI) as a case study and experiment with multiple datasets. Our method deploys backoff to kNN for BERT and RoBERTa on examples with low model confidence without any update to the model parameters. Our results indicate that the kNN approach makes the finetuned model more robust to adversarial inputs.

Ben Krause, Akhilesh Deepak Gotmare, Bryan McCann et al.

While large-scale language models (LMs) are able to imitate the distribution of natural language well enough to generate realistic text, it is difficult to control which regions of the distribution they generate. This is especially problematic because datasets used for training large LMs usually contain significant toxicity, hate, bias, and negativity. We propose GeDi as an efficient method for using smaller LMs as generative discriminators to guide generation from large LMs to make them safer and more controllable. GeDi guides generation at each step by computing classification probabilities for all possible next tokens via Bayes rule by normalizing over two class-conditional distributions; one conditioned on the desired attribute, or control code, and another conditioned on the undesired attribute, or anti control code. We find that GeDi gives stronger controllability than the state of the art method while also achieving generation speeds more than 30 times faster. Additionally, training GeDi on only four topics allows us to controllably generate new topics zero-shot from just a keyword, unlocking a new capability that previous controllable generation methods do not have. Lastly, we show that GeDi can make GPT-2 (1.5B parameters) significantly less toxic without sacrificing linguistic quality, making it by far the most practical existing method for detoxifying large language models while maintaining a fast generation speed.

Yu Bai, Ben Krause, Huan Wang et al.

We propose \emph{Taylorized training} as an initiative towards better understanding neural network training at finite width. Taylorized training involves training the $k$-th order Taylor expansion of the neural network at initialization, and is a principled extension of linearized training---a recently proposed theory for understanding the success of deep learning. We experiment with Taylorized training on modern neural network architectures, and show that Taylorized training (1) agrees with full neural network training increasingly better as we increase $k$, and (2) can significantly close the performance gap between linearized and full training. Compared with linearized training, higher-order training works in more realistic settings such as standard parameterization and large (initial) learning rate. We complement our experiments with theoretical results showing that the approximation error of $k$-th order Taylorized models decay exponentially over $k$ in wide neural networks.

Ben Krause, Emmanuel Kahembwe, Iain Murray et al.

This research note combines two methods that have recently improved the state of the art in language modeling: Transformers and dynamic evaluation. Transformers use stacked layers of self-attention that allow them to capture long range dependencies in sequential data. Dynamic evaluation fits models to the recent sequence history, allowing them to assign higher probabilities to re-occurring sequential patterns. By applying dynamic evaluation to Transformer-XL models, we improve the state of the art on enwik8 from 0.99 to 0.94 bits/char, text8 from 1.08 to 1.04 bits/char, and WikiText-103 from 18.3 to 16.4 perplexity points.

Joachim Fainberg, Ben Krause, Mihai Dobre et al.

Conversational agents are gaining popularity with the increasing ubiquity of smart devices. However, training agents in a data driven manner is challenging due to a lack of suitable corpora. This paper presents a novel method for gathering topical, unstructured conversational data in an efficient way: self-dialogues through crowd-sourcing. Alongside this paper, we include a corpus of 3.6 million words across 23 topics. We argue the utility of the corpus by comparing self-dialogues with standard two-party conversations as well as data from other corpora.

Ben Krause, Marco Damonte, Mihai Dobre et al.

We present Edina, the University of Edinburgh's social bot for the Amazon Alexa Prize competition. Edina is a conversational agent whose responses utilize data harvested from Amazon Mechanical Turk (AMT) through an innovative new technique we call self-dialogues. These are conversations in which a single AMT Worker plays both participants in a dialogue. Such dialogues are surprisingly natural, efficient to collect and reflective of relevant and/or trending topics. These self-dialogues provide training data for a generative neural network as well as a basis for soft rules used by a matching score component. Each match of a soft rule against a user utterance is associated with a confidence score which we show is strongly indicative of reply quality, allowing this component to self-censor and be effectively integrated with other components. Edina's full architecture features a rule-based system backing off to a matching score, backing off to a generative neural network. Our hybrid data-driven methodology thus addresses both coverage limitations of a strictly rule-based approach and the lack of guarantees of a strictly machine-learning approach.

Ben Krause, Emmanuel Kahembwe, Iain Murray et al.

We present methodology for using dynamic evaluation to improve neural sequence models. Models are adapted to recent history via a gradient descent based mechanism, causing them to assign higher probabilities to re-occurring sequential patterns. Dynamic evaluation outperforms existing adaptation approaches in our comparisons. Dynamic evaluation improves the state-of-the-art word-level perplexities on the Penn Treebank and WikiText-2 datasets to 51.1 and 44.3 respectively, and the state-of-the-art character-level cross-entropies on the text8 and Hutter Prize datasets to 1.19 bits/char and 1.08 bits/char respectively.

Ben Krause, Liang Lu, Iain Murray et al.

We introduce multiplicative LSTM (mLSTM), a recurrent neural network architecture for sequence modelling that combines the long short-term memory (LSTM) and multiplicative recurrent neural network architectures. mLSTM is characterised by its ability to have different recurrent transition functions for each possible input, which we argue makes it more expressive for autoregressive density estimation. We demonstrate empirically that mLSTM outperforms standard LSTM and its deep variants for a range of character level language modelling tasks. In this version of the paper, we regularise mLSTM to achieve 1.27 bits/char on text8 and 1.24 bits/char on Hutter Prize. We also apply a purely byte-level mLSTM on the WikiText-2 dataset to achieve a character level entropy of 1.26 bits/char, corresponding to a word level perplexity of 88.8, which is comparable to word level LSTMs regularised in similar ways on the same task.

Ben Krause

Recurrent Neural Networks (RNNs) have long been recognized for their potential to model complex time series. However, it remains to be determined what optimization techniques and recurrent architectures can be used to best realize this potential. The experiments presented take a deep look into Hessian free optimization, a powerful second order optimization method that has shown promising results, but still does not enjoy widespread use. This algorithm was used to train to a number of RNN architectures including standard RNNs, long short-term memory, multiplicative RNNs, and stacked RNNs on the task of character prediction. The insights from these experiments led to the creation of a new multiplicative LSTM hybrid architecture that outperformed both LSTM and multiplicative RNNs. When tested on a larger scale, multiplicative LSTM achieved character level modelling results competitive with the state of the art for RNNs using very different methodology.