README.md

---
tags:
- summarization
- summary
- booksum
- long-document
- long-form
license: apache-2.0
datasets:
- kmfoda/booksum
metrics:
- rouge
widget:
- text: large earthquakes along a given fault segment do not occur at random intervals
    because it takes time to accumulate the strain energy for the rupture. The rates
    at which tectonic plates move and accumulate strain at their boundaries are approximately
    uniform. Therefore, in first approximation, one may expect that large ruptures
    of the same fault segment will occur at approximately constant time intervals.
    If subsequent main shocks have different amounts of slip across the fault, then
    the recurrence time may vary, and the basic idea of periodic mainshocks must be
    modified. For great plate boundary ruptures the length and slip often vary by
    a factor of 2. Along the southern segment of the San Andreas fault the recurrence
    interval is 145 years with variations of several decades. The smaller the standard
    deviation of the average recurrence interval, the more specific could be the long
    term prediction of a future mainshock.
  example_title: earthquakes
- text: " A typical feed-forward neural field algorithm. Spatiotemporal coordinates\
    \ are fed into a neural network that predicts values in the reconstructed domain.\
    \ Then, this domain is mapped to the sensor domain where sensor measurements are\
    \ available as supervision. Class and Section Problems Addressed Generalization\
    \ (Section 2) Inverse problems, ill-posed problems, editability; symmetries. Hybrid\
    \ Representations (Section 3) Computation & memory efficiency, representation\
    \ capacity, editability: Forward Maps (Section 4) Inverse problems Network Architecture\
    \ (Section 5) Spectral bias, integration & derivatives. Manipulating Neural Fields\
    \ (Section 6) Edit ability, constraints, regularization. Table 2: The five classes\
    \ of techniques in the neural field toolbox each addresses problems that arise\
    \ in learning, inference, and control. (Section 3). We can supervise reconstruction\
    \ via differentiable forward maps that transform Or project our domain (e.g, 3D\
    \ reconstruction via 2D images; Section 4) With appropriate network architecture\
    \ choices, we can overcome neural network spectral biases (blurriness) and efficiently\
    \ compute derivatives and integrals (Section 5). Finally, we can manipulate neural\
    \ fields to add constraints and regularizations, and to achieve editable representations\
    \ (Section 6). Collectively, these classes constitute a 'toolbox' of techniques\
    \ to help solve problems with neural fields There are three components in a conditional\
    \ neural field: (1) An encoder or inference function \u20AC that outputs the conditioning\
    \ latent variable 2 given an observation 0 E(0) =2. 2 is typically a low-dimensional\
    \ vector, and is often referred to aS a latent code Or feature code_ (2) A mapping\
    \ function 4 between Z and neural field parameters O: Y(z) = O; (3) The neural\
    \ field itself $. The encoder \u20AC finds the most probable z given the observations\
    \ O: argmaxz P(2/0). The decoder maximizes the inverse conditional probability\
    \ to find the most probable 0 given Z: arg- max P(Olz). We discuss different encoding\
    \ schemes with different optimality guarantees (Section 2.1.1), both global and\
    \ local conditioning (Section 2.1.2), and different mapping functions Y (Section\
    \ 2.1.3) 2. Generalization Suppose we wish to estimate a plausible 3D surface\
    \ shape given a partial or noisy point cloud. We need a suitable prior over the\
    \ sur- face in its reconstruction domain to generalize to the partial observations.\
    \ A neural network expresses a prior via the function space of its architecture\
    \ and parameters 0, and generalization is influenced by the inductive bias of\
    \ this function space (Section 5)."
  example_title: scientific paper
- text: ' the big variety of data coming from diverse sources is one of the key properties
    of the big data phenomenon. It is, therefore, beneficial to understand how data
    is generated in various environments and scenarios, before looking at what should
    be done with this data and how to design the best possible architecture to accomplish
    this The evolution of IT architectures, described in Chapter 2, means that the
    data is no longer processed by a few big monolith systems, but rather by a group
    of services In parallel to the processing layer, the underlying data storage has
    also changed and became more distributed This, in turn, required a significant
    paradigm shift as the traditional approach to transactions (ACID) could no longer
    be supported. On top of this, cloud computing is becoming a major approach with
    the benefits of reducing costs and providing on-demand scalability but at the
    same time introducing concerns about privacy, data ownership, etc In the meantime
    the Internet continues its exponential growth: Every day both structured and unstructured
    data is published and available for processing: To achieve competitive advantage
    companies have to relate their corporate resources to external services, e.g.
    financial markets, weather forecasts, social media, etc While several of the sites
    provide some sort of API to access the data in a more orderly fashion; countless
    sources require advanced web mining and Natural Language Processing (NLP) processing
    techniques: Advances in science push researchers to construct new instruments
    for observing the universe O conducting experiments to understand even better
    the laws of physics and other domains. Every year humans have at their disposal
    new telescopes, space probes, particle accelerators, etc These instruments generate
    huge streams of data, which need to be stored and analyzed. The constant drive
    for efficiency in the industry motivates the introduction of new automation techniques
    and process optimization: This could not be done without analyzing the precise
    data that describe these processes. As more and more human tasks are automated,
    machines provide rich data sets, which can be analyzed in real-time to drive efficiency
    to new levels. Finally, it is now evident that the growth of the Internet of Things
    is becoming a major source of data. More and more of the devices are equipped
    with significant computational power and can generate a continuous data stream
    from their sensors. In the subsequent sections of this chapter, we will look at
    the domains described above to see what they generate in terms of data sets. We
    will compare the volumes but will also look at what is characteristic and important
    from their respective points of view. 3.1 The Internet is undoubtedly the largest
    database ever created by humans. While several well described; cleaned, and structured
    data sets have been made available through this medium, most of the resources
    are of an ambiguous, unstructured, incomplete or even erroneous nature. Still,
    several examples in the areas such as opinion mining, social media analysis, e-governance,
    etc, clearly show the potential lying in these resources. Those who can successfully
    mine and interpret the Internet data can gain unique insight and competitive advantage
    in their business An important area of data analytics on the edge of corporate
    IT and the Internet is Web Analytics.'
  example_title: data science textbook
- text: "Transformer-based models have shown to be very useful for many NLP tasks.\
    \ However, a major limitation of transformers-based models is its O(n^2)O(n 2)\
    \ time & memory complexity (where nn is sequence length). Hence, it's computationally\
    \ very expensive to apply transformer-based models on long sequences n > 512n>512.\
    \ Several recent papers, e.g. Longformer, Performer, Reformer, Clustered attention\
    \ try to remedy this problem by approximating the full attention matrix. You can\
    \ checkout \U0001F917's recent blog post in case you are unfamiliar with these\
    \ models.\nBigBird (introduced in paper) is one of such recent models to address\
    \ this issue. BigBird relies on block sparse attention instead of normal attention\
    \ (i.e. BERT's attention) and can handle sequences up to a length of 4096 at a\
    \ much lower computational cost compared to BERT. It has achieved SOTA on various\
    \ tasks involving very long sequences such as long documents summarization, question-answering\
    \ with long contexts.\nBigBird RoBERTa-like model is now available in \U0001F917\
    Transformers. The goal of this post is to give the reader an in-depth understanding\
    \ of big bird implementation & ease one's life in using BigBird with \U0001F917\
    Transformers. But, before going into more depth, it is important to remember that\
    \ the BigBird's attention is an approximation of BERT's full attention and therefore\
    \ does not strive to be better than BERT's full attention, but rather to be more\
    \ efficient. It simply allows to apply transformer-based models to much longer\
    \ sequences since BERT's quadratic memory requirement quickly becomes unbearable.\
    \ Simply put, if we would have \u221E compute & \u221E time, BERT's attention\
    \ would be preferred over block sparse attention (which we are going to discuss\
    \ in this post).\nIf you wonder why we need more compute when working with longer\
    \ sequences, this blog post is just right for you!\nSome of the main questions\
    \ one might have when working with standard BERT-like attention include:\nDo all\
    \ tokens really have to attend to all other tokens? Why not compute attention\
    \ only over important tokens? How to decide what tokens are important? How to\
    \ attend to just a few tokens in a very efficient way? In this blog post, we will\
    \ try to answer those questions.\nWhat tokens should be attended to? We will give\
    \ a practical example of how attention works by considering the sentence 'BigBird\
    \ is now available in HuggingFace for extractive question answering'. In BERT-like\
    \ attention, every word would simply attend to all other tokens.\nLet's think\
    \ about a sensible choice of key tokens that a queried token actually only should\
    \ attend to by writing some pseudo-code. Will will assume that the token available\
    \ is queried and build a sensible list of key tokens to attend to.\n>>> # let's\
    \ consider following sentence as an example >>> example = ['BigBird', 'is', 'now',\
    \ 'available', 'in', 'HuggingFace', 'for', 'extractive', 'question', 'answering']\n\
    >>> # further let's assume, we're trying to understand the representation of 'available'\
    \ i.e. >>> query_token = 'available' >>> # We will initialize an empty `set` and\
    \ fill up the tokens of our interest as we proceed in this section. >>> key_tokens\
    \ = [] # => currently 'available' token doesn't have anything to attend Nearby\
    \ tokens should be important because, in a sentence (sequence of words), the current\
    \ word is highly dependent on neighboring past & future tokens. This intuition\
    \ is the idea behind the concept of sliding attention."
  example_title: bigbird blog intro
- text: "To be fair, you have to have a very high IQ to understand Rick and Morty.\
    \ The humour is extremely subtle, and without a solid grasp of theoretical physics\
    \ most of the jokes will go over a typical viewer's head. There's also Rick's\
    \ nihilistic outlook, which is deftly woven into his characterisation- his personal\
    \ philosophy draws heavily from Narodnaya Volya literature, for instance. The\
    \ fans understand this stuff; they have the intellectual capacity to truly appreciate\
    \ the depths of these jokes, to realise that they're not just funny- they say\
    \ something deep about LIFE. As a consequence people who dislike Rick & Morty\
    \ truly ARE idiots- of course they wouldn't appreciate, for instance, the humour\
    \ in Rick's existential catchphrase 'Wubba Lubba Dub Dub,' which itself is a cryptic\
    \ reference to Turgenev's Russian epic Fathers and Sons. I'm smirking right now\
    \ just imagining one of those addlepated simpletons scratching their heads in\
    \ confusion as Dan Harmon's genius wit unfolds itself on their television screens.\
    \ What fools.. how I pity them. \U0001F602\nAnd yes, by the way, i DO have a Rick\
    \ & Morty tattoo. And no, you cannot see it. It's for the ladies' eyes only- and\
    \ even then they have to demonstrate that they're within 5 IQ points of my own\
    \ (preferably lower) beforehand. Nothin personnel kid \U0001F60E"
  example_title: Richard & Mortimer
parameters:
  max_length: 64
  min_length: 8
  no_repeat_ngram_size: 3
  early_stopping: true
  repetition_penalty: 3.5
  length_penalty: 0.3
  encoder_no_repeat_ngram_size: 3
  num_beams: 4
model-index:
- name: pszemraj/long-t5-tglobal-base-16384-book-summary
  results:
  - task:
      type: summarization
      name: Summarization
    dataset:
      name: samsum
      type: samsum
      config: samsum
      split: test
    metrics:
    - name: ROUGE-1
      type: rouge
      value: 33.7197
      verified: true
    - name: ROUGE-2
      type: rouge
      value: 8.5493
      verified: true
    - name: ROUGE-L
      type: rouge
      value: 25.1917
      verified: true
    - name: ROUGE-LSUM
      type: rouge
      value: 29.2739
      verified: true
    - name: loss
      type: loss
      value: .nan
      verified: true
    - name: gen_len
      type: gen_len
      value: 34.464
      verified: true
  - task:
      type: summarization
      name: Summarization
    dataset:
      name: xsum
      type: xsum
      config: default
      split: test
    metrics:
    - name: ROUGE-1
      type: rouge
      value: 20.5398
      verified: true
    - name: ROUGE-2
      type: rouge
      value: 3.4827
      verified: true
    - name: ROUGE-L
      type: rouge
      value: 13.647
      verified: true
    - name: ROUGE-LSUM
      type: rouge
      value: 15.8818
      verified: true
    - name: loss
      type: loss
      value: .nan
      verified: true
    - name: gen_len
      type: gen_len
      value: 81.4964
      verified: true
---

# long-t5-tglobal-base-16384 + BookSum

- summarize long text and get a SparkNotes-esque summary of arbitrary topics!
- generalizes reasonably well to academic & narrative text. 

## Cheeky Proof-of-Concept

A summary of the [infamous navy seals copypasta](https://knowyourmeme.com/memes/navy-seal-copypasta):

> The narrator tells the audience that he can kill anyone anywhere in the world with his bare hands, and he has access to all of the United States military's weapons.


## Model description

A fine-tuned version of [google/long-t5-tglobal-base](https://huggingface.co/google/long-t5-tglobal-base) on the `kmfoda/booksum` dataset:

- 30+ epochs of fine-tuning from the base model on V100/A100 GPUs
- all training used 16384 token input / 1024 max output

Read the paper by Guo et al. here: [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/pdf/2112.07916.pdf) 

## How-To in Python

Install/update transformers `pip install -U transformers`

Summarize text with pipeline:

```
from transformers import pipeline

summarizer = pipeline(
    'summarization',
    'pszemraj/long-t5-tglobal-base-16384-book-summary',
    )
long_text = "Here is a lot of text I don't want to read. Replace me"

result = summarizer(long_text)
print(result[0]['summary_text'])
```

Pass [other parameters related to beam search textgen](https://huggingface.co/blog/how-to-generate) when calling `summarizer` to get even higher quality results.

## Intended uses & limitations

- At the time of writing, the model is not _fully converged_ despite training for 20+ epochs. This checkpoint is serviceable enough (see examples).
  - I plan to update this page with newer checkpoints and post some metrics over time.
  - Compare performance to [LED-base](https://huggingface.co/pszemraj/led-base-book-summary) trained on the same dataset (API gen parameters are the same).
- while this model seems to improve upon factual consistency, **do not take summaries to be foolproof and check things that seem odd**.

## Training and evaluation data

`kmfoda/booksum` dataset on HuggingFace - read [the original paper here](https://arxiv.org/abs/2105.08209). Summaries longer than 1024 LongT5 tokens were filtered out to prevent the model from learning to generate "partial" summaries.

_NOTE: early checkpoints of this model were trained on a "smaller" subsection of the dataset as it was filtered for summaries of **1024 characters**. This was subsequently caught and adjusted to **1024 tokens** and then trained further for 10+ epochs._

## Training procedure

### Updates:

- July 3, 2022: Added a new version with several epochs of additional training that is more performant in general.

### Training hyperparameters

The following hyperparameters were used during the **most recent** training round\*:

- learning_rate: 0.0006
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 64
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.01
- num_epochs: 2


\*_Prior training sessions used roughly similar parameters; multiple sessions were required as this takes aeons to train_

### Training results



### Framework versions

- Transformers 4.20.1
- Pytorch 1.10.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1