t5x

No description available

2,656

302

2,656

181

View on GitHub

Top Related Projects

tensorflow

185,446

An Open Source Machine Learning Framework for Everyone

transformers

131,798

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

pytorch

82,049

Tensors and Dynamic neural networks in Python with strong GPU acceleration

DeepSpeed

34,658

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

Megatron-LM

10,275

Ongoing research training transformer models at scale

gpt-neox

6,886

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

Quick Overview

T5X is a modular, composable, and extensible framework for training and evaluating large language models, particularly focused on the T5 family of models. It is designed to be flexible and scalable, allowing researchers and practitioners to easily experiment with different model architectures, training regimes, and evaluation methods.

Pros

Highly modular and extensible architecture, allowing for easy customization and experimentation
Supports distributed training across multiple devices and machines
Integrates well with other Google Research projects and libraries
Provides pre-trained models and checkpoints for quick start and fine-tuning

Cons

Steep learning curve for those unfamiliar with JAX and Flax
Documentation can be sparse or outdated in some areas
Primarily focused on T5 models, which may limit its applicability for other model architectures
Requires significant computational resources for training large models

Code Examples

Loading a pre-trained T5 model:

import t5x

model = t5x.models.EncoderDecoderModel.load_from_checkpoint(
    "/path/to/checkpoint",
    batch_size=32
)

Defining a custom training task:

from t5x import tasks

class CustomTask(tasks.Task):
    def get_dataset(self, split):
        # Custom dataset loading logic
        pass

    def preprocess_example(self, example):
        # Custom preprocessing logic
        return example

custom_task = CustomTask()

Running inference with a trained model:

from t5x import inference

results = inference.predict(
    model,
    dataset=test_dataset,
    output_dir="/path/to/output",
    checkpoint_path="/path/to/checkpoint",
    beam_size=4
)

Getting Started

To get started with T5X, follow these steps:

Install the required dependencies:

pip install t5x

Clone the T5X repository:

git clone https://github.com/google-research/t5x.git
cd t5x

Set up your environment variables:

export PYTHONPATH="$PYTHONPATH:$(pwd)"

Run a simple training job:

from t5x import train

train.main(
    model_dir="/path/to/model_dir",
    gin_file="t5x/examples/t5/t5_1_1/small.gin",
    gin_bindings=[
        "MIXTURE_OR_TASK_NAME = 'c4_v220_span_corruption'",
        "TASK_FEATURE_LENGTHS = {'inputs': 512, 'targets': 114}",
        "TRAIN_STEPS = 1_000_000",
        "SAVING_PERIOD = 10_000",
        "EVAL_PERIOD = 100",
    ]
)

This will start training a small T5 model on the C4 dataset for span corruption.

Competitor Comparisons

tensorflow

185,446

An Open Source Machine Learning Framework for Everyone

Pros of TensorFlow

Broader ecosystem with extensive tools and libraries
More mature and stable, with longer development history
Wider industry adoption and community support

Cons of TensorFlow

Can be more complex and harder to learn for beginners
Less focused on specific tasks like T5X's emphasis on text-to-text transfer

Code Comparison

T5X example:

import t5x

model = t5x.models.EncoderDecoderModel(...)
trainer = t5x.trainer.Trainer(...)
trainer.train(...)

TensorFlow example:

import tensorflow as tf

model = tf.keras.Sequential([...])
model.compile(...)
model.fit(...)

Key Differences

T5X is specifically designed for text-to-text transfer tasks, while TensorFlow is a general-purpose machine learning framework
T5X builds on top of JAX and Flax, whereas TensorFlow has its own ecosystem
T5X focuses on research and experimentation, while TensorFlow caters to both research and production environments

Use Cases

Choose T5X for text-to-text transfer tasks and research in natural language processing
Opt for TensorFlow for a wide range of machine learning applications and production-ready deployments

transformers

131,798

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

Pros of Transformers

Broader model support: Includes a wide range of transformer-based models beyond T5
Extensive documentation and community support
Easier integration with popular deep learning frameworks like PyTorch and TensorFlow

Cons of Transformers

Less specialized for T5 models compared to T5X
May have slower performance for specific T5-related tasks
Larger codebase, potentially more complex for T5-focused projects

Code Comparison

T5X:

import t5x

model = t5x.models.EncoderDecoderModel(...)
trainer = t5x.trainer.Trainer(...)
trainer.train(...)

Transformers:

from transformers import T5ForConditionalGeneration, Trainer

model = T5ForConditionalGeneration.from_pretrained("t5-base")
trainer = Trainer(model=model, ...)
trainer.train()

Both repositories provide powerful tools for working with transformer models, particularly T5. T5X is more specialized for T5 models and may offer better performance in specific scenarios, while Transformers offers broader model support and easier integration with popular frameworks. The choice between them depends on the specific requirements of your project and your familiarity with each library.

pytorch

82,049

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Pros of PyTorch

Wider adoption and larger community support
More flexible and intuitive for research and experimentation
Extensive ecosystem of tools and libraries

Cons of PyTorch

Less optimized for production deployment compared to T5X
May require more manual optimization for large-scale training

Code Comparison

T5X (JAX-based):

import jax.numpy as jnp
from t5x import models

model = models.EncoderDecoderModel(...)
outputs = model(inputs)
loss = jnp.mean(outputs.loss)

PyTorch:

import torch
import torch.nn as nn

model = nn.TransformerEncoderDecoder(...)
outputs = model(inputs)
loss = outputs.loss.mean()

Both frameworks offer similar high-level APIs for building and training models, but T5X is more specialized for T5-like models, while PyTorch provides a more general-purpose deep learning framework. T5X leverages JAX for efficient compilation and parallelism, which can be advantageous for large-scale training. PyTorch, on the other hand, offers more flexibility and ease of use, making it popular for research and prototyping.

DeepSpeed

34,658

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

Pros of DeepSpeed

Broader compatibility with various deep learning frameworks (PyTorch, TensorFlow, etc.)
More extensive optimization techniques, including ZeRO-Offload and 3D parallelism
Active community support and frequent updates

Cons of DeepSpeed

Steeper learning curve due to more complex configuration options
May require more fine-tuning to achieve optimal performance for specific use cases

Code Comparison

T5x:

import t5x

model = t5x.models.EncoderDecoderModel(...)
trainer = t5x.trainer.Trainer(...)
trainer.train(...)

DeepSpeed:

import deepspeed

model = MyModel(...)
engine = deepspeed.initialize(model=model, config=ds_config)
engine.train(...)

Key Differences

T5x is specifically designed for T5 models, while DeepSpeed is more versatile
DeepSpeed offers more advanced parallelism and optimization techniques
T5x integrates seamlessly with JAX and Flax, while DeepSpeed primarily focuses on PyTorch

Use Cases

T5x: Ideal for projects specifically working with T5 models and JAX/Flax ecosystem
DeepSpeed: Better suited for large-scale training across various model architectures and frameworks

Megatron-LM

10,275

Ongoing research training transformer models at scale

Pros of Megatron-LM

Optimized for NVIDIA GPUs, offering better performance on supported hardware
Supports a wider range of model architectures, including GPT, BERT, and T5
Provides more advanced distributed training capabilities, including model and data parallelism

Cons of Megatron-LM

Steeper learning curve and more complex setup compared to T5X
Less focus on T5-specific optimizations and features
May require more manual configuration for certain tasks

Code Comparison

T5X example:

import t5x

model = t5x.models.EncoderDecoderModel(...)
trainer = t5x.trainer.Trainer(...)
trainer.train(...)

Megatron-LM example:

import megatron

args = megatron.get_args()
model = megatron.model.GPTModel(...)
optimizer = megatron.initialize_megatron(args)
train(model, optimizer, ...)

Both repositories offer powerful tools for training large language models, but they cater to different use cases. T5X is more focused on the T5 architecture and provides a simpler interface, while Megatron-LM offers more flexibility and optimization for NVIDIA hardware at the cost of increased complexity.

gpt-neox

6,886

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

Pros of gpt-neox

Designed specifically for training large language models, with optimizations for scale
Includes tools for dataset preparation and tokenization
Supports distributed training across multiple GPUs and nodes

Cons of gpt-neox

Less flexible for general-purpose machine learning tasks
May require more computational resources for training
Limited to autoregressive language models

Code Comparison

gpt-neox:

from megatron.neox_arguments import NeoXArgs
from megatron.global_vars import set_global_variables, get_args
from megatron.training import pretrain

args_defaults = {
    'num_layers': 12,
    'hidden_size': 768,
    'num_attention_heads': 12,
}

t5x:

from t5x import models
from t5x import partitioning
from t5x import train

model = models.EncoderDecoderModel(
    module=my_network_def,
    input_vocabulary=vocabulary,
    output_vocabulary=vocabulary,
    optimizer_def=optimizers.Adam(0.001),
)

Both repositories provide frameworks for training large language models, but they differ in their focus and implementation. gpt-neox is tailored for training GPT-style models at scale, while t5x offers more flexibility for various encoder-decoder architectures. The code snippets illustrate the different approaches to model configuration and initialization in each framework.

Convert designs to code with AI

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot

README

T5X

Go to T5X ReadTheDocs Documentation Page.

T5X is a modular, composable, research-friendly framework for high-performance, configurable, self-service training, evaluation, and inference of sequence models (starting with language) at many scales.

It is essentially a new and improved implementation of the T5 codebase (based on Mesh TensorFlow) in JAX and Flax. To learn more, see the T5X Paper.

Below is a quick start guide for training models with TPUs on Google Cloud. For additional tutorials and background, see the complete documentation.

Quickstart (Recommended)

T5X can be run with XManager on Vertex AI. Vertex AI is a platform for training that creates TPU instances and runs code on the TPUs. Vertex AI will also shut down the TPUs when the jobs terminate. This is signifcantly easier than managing GCE VMs and TPU VM instances.

Follow the pre-requisites and directions to install XManager.
Request TPU quota as required. GCP projects come with 8 cores by default, which is enough to run one training experiment on a single TPU host. If you want to run multi-host training or run multiple trials in parallel, you will need more quota. Navigate to Quotas.

The quota you want is:

Service: Vertex AI API
Dimensions (location): us-central1
If you want to run single-host experiments:
- Custom model training TPU V2 cores per region
- Custom model training TPU V3 cores per region
If you want to run multi-host experiments:
- Custom model training TPU V2 pod cores per region
- Custom model training TPU V3 pod cores per region

TIP: You won't be able to run single-host experiments with multi-host quota. (i.e. you can't run tpu_v2=8 using TPU V2 pod)

Launch the xmanager script located at t5x/scripts/xm_launch.py.

As a running example, we use the WMT14 En-De translation which is described in more detail in the Examples section below.

export GOOGLE_CLOUD_BUCKET_NAME=...
export TFDS_DATA_DIR=gs://$GOOGLE_CLOUD_BUCKET_NAME/t5x/data
export MODEL_DIR=gs://$GOOGLE_CLOUD_BUCKET_NAME/t5x/$(date +%Y%m%d)

# Pre-download dataset in multi-host experiments.
tfds build wmt_t2t_translate --data_dir=$TFDS_DATA_DIR

git clone https://github.com/google-research/t5x
cd ./t5x/

python3 ./t5x/scripts/xm_launch.py \
  --gin_file=t5x/examples/t5/t5_1_1/examples/base_wmt_from_scratch.gin \
  --model_dir=$MODEL_DIR \
  --tfds_data_dir=$TFDS_DATA_DIR

Check gs://$GOOGLE_CLOUD_BUCKET_NAME/t5x/ for the output artifacts, which can be read by TensorBoard.

GPU Usage

Note: NVIDIA has released an updated version of this repository with H100 FP8 support and broad GPU performance improvements. Please visit the NVIDIA Rosetta repository for more details and usage instructions.

T5X can be run easily on GPUs either in single-node configurations or multi-node configurations with a SLURM+pyxis cluster. Further instructions at t5x/contrib/gpu. The t5x/contrib/gpu/scripts_gpu folder contains example scripts for pretraining T5X on The Pile and for finetuning on SQuAD and MNLI. These scripts and associated gin configurations also contain additional GPU optimizations for better throughput. More examples and instructions can be found in the NVIDIA Rosetta repository maintained by NVIDIA with H100 FP8 support and broad GPU performance improvements.

Installation

Note that all the commands in this document should be run in the commandline of the TPU VM instance unless otherwise stated.

Follow the instructions to set up a Google Cloud Platform (GCP) account and enable the Cloud TPU API.

Note: T5X also works with GPU, please follow instructions in t5x/contrib/gpu if you'd like to use GPU version.
Create a Cloud TPU VM instance following this instruction. We recommend that you develop your workflow in a single v3-8 TPU (i.e., --accelerator-type=v3-8) and scale up to pod slices once the pipeline is ready. In this README, we focus on using a single v3-8 TPU. See here to learn more about TPU architectures.
With Cloud TPU VMs, you ssh directly into the host machine of the TPU VM. You can install packages, run your code run, etc. in the host machine. Once the TPU instance is created, ssh into it with
```
gcloud alpha compute tpus tpu-vm ssh ${TPU_NAME} --zone=${ZONE}
```
where TPU_NAME and ZONE are the name and the zone used in step 2.

Install T5X and the dependencies.

git clone --branch=main https://github.com/google-research/t5x
cd t5x

python3 -m pip install -e '.[tpu]' -f \
  https://storage.googleapis.com/jax-releases/libtpu_releases.html

Create Google Cloud Storage (GCS) bucket to store the dataset and model checkpoints. To create a GCS bucket, see these instructions.
(optional) If you prefer working with Jupyter/Colab style environment you can setup a custom Colab runtime by following steps from t5x/notebooks.

Example: English to German translation

As a running example, we use the WMT14 En-De translation. The raw dataset is available in TensorFlow Datasets as "wmt_t2t_translate".

T5 casts the translation task such as the following

{'en': 'That is good.', 'de': 'Das ist gut.'}

to the form called "text-to-text":

{'inputs': 'translate English to German: That is good.', 'targets': 'Das ist gut.'}

This formulation allows many different classes of language tasks to be expressed in a uniform manner and a single encoder-decoder architecture can handle them without any task-specific parameters. For more detail, refer to the T5 paper (Raffel et al. 2019).

For a scalable data pipeline and an evaluation framework, we use SeqIO, which was factored out of the T5 library. A seqio.Task packages together the raw dataset, vocabulary, preprocessing such as tokenization and evaluation metrics such as BLEU and provides a tf.data instance.

The T5 library provides a number of seqio.Tasks that were used in the T5 paper. In this example, we use wmt_t2t_ende_v003.

Before training or fine-tuning you need to download ["wmt_t2t_translate"] (https://www.tensorflow.org/datasets/catalog/wmt_t2t_translate) dataset first.

# Data dir to save the processed dataset in "gs://data_dir" format.
TFDS_DATA_DIR="..."

# Make sure that dataset package is up-to-date.
python3 -m pip install --upgrade tfds-nightly

# Pre-download dataset.
tfds build wmt_t2t_translate ${TFDS_DATA_DIR}

Training

To run a training job, we use the t5x/train.py script.

# Model dir to save logs, ckpts, etc. in "gs://model_dir" format.
MODEL_DIR="..."
T5X_DIR="..."  # directory where the T5X repo is cloned.
TFDS_DATA_DIR="..."

python3 ${T5X_DIR}/t5x/train.py \
  --gin_file="t5x/examples/t5/t5_1_1/examples/base_wmt_from_scratch.gin" \
  --gin.MODEL_DIR=\"${MODEL_DIR}\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

The configuration for this training run is defined in the Gin file base_wmt_from_scratch.gin. Gin-config is a library to handle configurations based on dependency injection. Among many benefits, Gin allows users to pass custom components such as a custom model to the T5X library without having to modify the core library. The custom components section shows how this is done.

While the core library is independent of Gin, it is central to the examples we provide. Therefore, we provide a short introduction to Gin in the context of T5X. All the configurations are written to a file "config.gin" in MODEL_DIR. This makes debugging as well as reproducing the experiment much easier.

In addition to the config.json, model-info.txt file summarizes the model parameters (shape, names of the axes, partitioning info) as well as the optimizer states.

TensorBoard

To monitor the training in TensorBoard, it is much easier (due to authentification issues) to launch the TensorBoard on your own machine and not in the TPU VM. So in the commandline where you ssh'ed into the TPU VM, launch the TensorBoard with the logdir pointing to the MODEL_DIR.

# NB: run this on your machine not TPU VM!
MODEL_DIR="..."  # Copy from the TPU VM.
tensorboard --logdir=${MODEL_DIR}

Or you can launch the TensorBoard inside a Colab. In a Colab cell, run

from google.colab import auth
auth.authenticate_user()

to authorize the Colab to access the GCS bucket and launch the TensorBoard.

%load_ext tensorboard
model_dir = "..."  # Copy from the TPU VM.
%tensorboard --logdir=model_dir

Fine-tuning

We can leverage the benefits of self-supervised pre-training by initializing from one of our pre-trained models. Here we use the T5.1.1 Base checkpoint.

# Model dir to save logs, ckpts, etc. in "gs://model_dir" format.
MODEL_DIR="..."

# Data dir to save the processed dataset in "gs://data_dir" format.
TFDS_DATA_DIR="..."
T5X_DIR="..."  # directory where the T5X repo is cloned.

python3 ${T5X_DIR}/t5x/train.py \
  --gin_file="t5x/examples/t5/t5_1_1/examples/base_wmt_finetune.gin" \
  --gin.MODEL_DIR=\"${MODEL_DIR}\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

Note: when supplying a string, dict, list, tuple value, or a bash variable via a flag, you must put it in quotes. In the case of strings, it requires escaped quotes (\"<string>\"). For example: --gin.utils.DatasetConfig.split=\"validation\" or --gin.MODEL_DIR=\"${MODEL_DIR}\".

Gin makes it easy to change a number of configurations. For example, you can change the partitioning.PjitPartitioner.num_partitions (overriding the value in base_wmt_from_scratch.gin) to chanage the parallelism strategy and pass it as a commandline arg.

--gin.partitioning.PjitPartitioner.num_partitions=8

Evaluation

To run the offline (i.e. without training) evaluation, you can use t5x/eval.py script.

EVAL_OUTPUT_DIR="..."  # directory to write eval output
T5X_DIR="..."  # directory where the t5x is cloned, e.g., ${HOME}"/t5x".
TFDS_DATA_DIR="..."
CHECKPOINT_PATH="..."

python3 ${T5X_DIR}/t5x/eval.py \
  --gin_file="t5x/examples/t5/t5_1_1/examples/base_wmt_eval.gin" \
  --gin.CHECKPOINT_PATH=\"${CHECKPOINT_PATH}\" \
  --gin.EVAL_OUTPUT_DIR=\"${EVAL_OUTPUT_DIR}\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

Inference

To run inference, you can use t5x/infer.py script. Here we use the same seqio.Task, but for inference we do not use the targets features other than logging them alongside the prediction in a JSON file.

INFER_OUTPUT_DIR="..."  # directory to write infer output
T5X_DIR="..."  # directory where the t5x is cloned, e.g., ${HOME}"/t5x".
TFDS_DATA_DIR="..."
CHECKPOINT_PATH="..."

python3 ${T5X_DIR}/t5x/infer.py \
  --gin_file="t5x/examples/t5/t5_1_1/examples/base_wmt_infer.gin" \
  --gin.CHECKPOINT_PATH=\"${CHECKPOINT_PATH}\" \
  --gin.INFER_OUTPUT_DIR=\"${INFER_OUTPUT_DIR}\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

Exporting as TensorFlow Saved Model

Pretrained model can be exported as TensorFlow Saved Model, and deployed to Vertex AI Prediction service using [Optimized TensorFlow Runtime] (https://cloud.google.com/vertex-ai/docs/predictions/optimized-tensorflow-runtime). Please note that exported model won't work on OSS based TensorFlow Model Server.

T5X_DIR="..."  # directory where the t5x is cloned, e.g., ${HOME}"/t5x".
CHECKPOINT_PATH="..."

BATCH_SIZE=None
BEAM_SIZE=1

# Use 'bfloat16' if you plan to run exported model on NVIDIA A100 or newer GPUs,
# for other GPUs use 'float32'.
ACTIVATION_DTYPE=bfloat16

# Version numbers must be numeric. We generate one based on datetime.
VERSION=$(date +%Y%m%d%H%M%S)

NAME=t5x_base_${ACTIVATION_DTYPE}  # Model name.

# Path to export model to. Note that export script is going to add _cpu suffix
# after model name.
OUTPUT=${CHECKPOINT_PATH}/saved_model.${NAME}/${VERSION}

declare -a ARGS=(
--gin_file=t5x/examples/t5/t5_1_1/base.gin
--gin_file=t5x/t5x/configs/runs/export.gin
--gin.TASK_FEATURE_LENGTHS="{'inputs': 256, 'targets': 256}"
--gin.CHECKPOINT_PATH=\"${CHECKPOINT_PATH}\"
--gin.MODEL_NAME=\"/ml/${USER}/t5x_base\"
--gin.MODEL_OUTPUT_DIR=\"${OUTPUT}\"
--gin.BEAM_SIZE=${BEAM_SIZE}
--gin.BATCH_SIZE=${BATCH_SIZE}
--gin.export_lib.save.partitioner=None
--gin.export_lib.save.warmup_examples="['hello world']"
--gin.export_lib.ExportableModule.use_batch_function=False
--gin.export_lib.ExportableModule.use_gpu=False
--gin.export_lib.ExportableModule.jit_compile=False
--gin.ACTIVATION_DTYPE=\"${ACTIVATION_DTYPE}\"
--gin.network.T5Config.dtype=\"${ACTIVATION_DTYPE}\"
--gin.utils.RestoreCheckpointConfig.dtype=\"${ACTIVATION_DTYPE}\"
--gin.DROPOUT_RATE=0.0
)

(python3 ${T5X_DIR}/t5x/export.py "${ARGS[@]}")

For detailed arguments definition refer to [export.gin] (t5x/configs/runs/export.gin).

You can run XL and smaller models on NVIDIA A100 40GB, and XXL models on NVIDIA A100 80GB.

Custom components

The translation example uses the encoder-decoder model that T5X provides as well as the dataset from the T5 library. This section shows how you can use your own dataset and a model and pass via Gin.

Example: custom dataset in a user directory

For this example, we have the following directory structure with ${HOME}/dir1/user_dir representing a user directory with custom components.

${HOME}
âââ dir1
 Â Â  âââ user_dir
 Â Â      âââ t5_1_1_base_de_en.gin
 Â Â      âââ tasks.py

As an example, let's define a new dataset. Here we use the same Translation dataset but we define the translation task in the opposite direction, i.e., German to English intead of English to German. We define this task in tasks.py

# ${HOME}/dir1/user_dir/tasks.py

import functools
import seqio
import tensorflow_datasets as tfds
from t5.evaluation import metrics
from t5.data import preprocessors

vocabulary = seqio.SentencePieceVocabulary(
    'gs://t5-data/vocabs/cc_all.32000/sentencepiece.model', extra_ids=100)
output_features = {
    'inputs': seqio.Feature(vocabulary=vocabulary),
    'targets': seqio.Feature(vocabulary=vocabulary)
}

seqio.TaskRegistry.add(
    'wmt_t2t_de_en_v003',
    source=seqio.TfdsDataSource(tfds_name='wmt_t2t_translate/de-en:1.0.0'),
    preprocessors=[
        functools.partial(
            preprocessors.translate,
            source_language='de', target_language='en'),
        seqio.preprocessors.tokenize,
        seqio.CacheDatasetPlaceholder(),
        seqio.preprocessors.append_eos_after_trim,
    ],
    metric_fns=[metrics.bleu],
    output_features=output_features)

In the Gin file, most of the settings are equivalent to those used in the En->De example. So we include the Gin file from that example. To use "wmt_t2t_de_en_v003" task we just defined, we need to import the task module "tasks.py". Note that we use a relative path defined with respect to the user directory. This will be specified as a flag.

# ${HOME}/dir1/user_dir/t5_1_1_base_de_en.gin
from __gin__ import dynamic_registration
import tasks  # This imports the task defined in dir1/user_dir/tasks.py.

include "t5x-tmp/t5x/examples/t5/t5_1_1/examples/base_wmt_from_scratch.gin"
MIXTURE_OR_TASK_NAME = "wmt_t2t_de_en_v003"

Finally, we launch training passing the user directory as a flag gin_search_paths such that the Gin file and python modules can be specified with relative paths.

PROJECT_DIR=${HOME}"/dir1/user_dir"
T5X_DIR="..."  # directory where the t5x is cloned.
TFDS_DATA_DIR="..."
MODEL_DIR="..."
export PYTHONPATH=${PROJECT_DIR}

python3 ${T5X_DIR}/t5x/train.py \
  --gin_search_paths=${PROJECT_DIR} \
  --gin_file="t5_1_1_base_de_en.gin" \
  --gin.MODEL_DIR=\"${MODEL_DIR}\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

Checkpoints

Native Checkpoints

We have released the checkpoints of many of the original T5 models and their variants a native T5X format for maximal efficiency. See the complete list including the matching Gin configuration files.

These are converted from the public Mesh TensorFlow checkpoints .

Compatibility with the Mesh TensorFlow checkpoints

The Mesh TensorFlow checkpoints trained using the T5 library can be directly loaded into T5X. For example, we can rerun the fine-tuning example initializing from the MTF checkpoint by changing the INIT_CHECKPOINT Gin macro.

# Model dir to save logs, ckpts, etc. in "gs://model_dir" format.
MODEL_DIR="..."

# Data dir to save the processed dataset in "gs://data_dir" format.
TFDS_DATA_DIR="..."
T5X_DIR="..."  # directory where the T5X repo is cloned.

python3 ${T5X_DIR}/t5x/train.py \
  --gin_file="t5x/examples/t5/t5_1_1/examples/base_wmt19_ende_train.gin" \
  --gin.MODEL_DIR=\"${MODEL_DIR}\" \
  --gin.MIXTURE_OR_TASK_NAME=\"wmt_t2t_ende_v003\" \
  --gin.INIT_CHECKPOINT=\"gs://t5-data/pretrained_models/t5.1.1.base/model.ckpt-1000000\" \
  --tfds_data_dir=${TFDS_DATA_DIR}

Note that restoring directly from the Mesh TensorFlow checkpoints can be inefficient if heavy model parallelism is used for large models. This is because each host loads the entire copy of the model first and then keep only the relevant slices dictated by the model parallelism specification. If you have Mesh TensorFlow checkpoints that you run often, we recommend converting the checkpoints to T5X native format using the convert_tf_checkpoint script.

Citing T5X

Please use the following bibtex entry to cite T5X.

@article{roberts2022t5x,
  url = {https://arxiv.org/abs/2203.17189},
  author = {Roberts, Adam and Chung, Hyung Won and Levskaya, Anselm and Mishra, Gaurav and Bradbury, James and Andor, Daniel and Narang, Sharan and Lester, Brian and Gaffney, Colin and Mohiuddin, Afroz and Hawthorne, Curtis and Lewkowycz, Aitor and Salcianu, Alex and van Zee, Marc and Austin, Jacob and Goodman, Sebastian and Soares, Livio Baldini and Hu, Haitang and Tsvyashchenko, Sasha and Chowdhery, Aakanksha and Bastings, Jasmijn and Bulian, Jannis and Garcia, Xavier and Ni, Jianmo and Chen, Andrew and Kenealy, Kathleen and Clark, Jonathan H. and Lee, Stephan and Garrette, Dan and Lee-Thorp, James and Raffel, Colin and Shazeer, Noam and Ritter, Marvin and Bosma, Maarten and Passos, Alexandre and Maitin-Shepard, Jeremy and Fiedel, Noah and Omernick, Mark and Saeta, Brennan and Sepassi, Ryan and Spiridonov, Alexander and Newlan, Joshua and Gesmundo, Andrea},
  title = {Scaling Up Models and Data with $\texttt{t5x}$ and $\texttt{seqio}$},
  journal={arXiv preprint arXiv:2203.17189},
  year = {2022},
}

Note

This is not an officially supported Google product

Top Related Projects

Convert designs to code with AI

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot