keras/gpt2_extra_large_en

Model Overview

GPT-2 is a language model published by OpenAI. Models are fine tuned on WebText, and range in size from 125 million to 1.5 billion parameters. See the model card below for benchmarks, data sources, and intended use cases.

Weights are released under the MIT License. Keras model code is released under the Apache 2 License.

Links

• GPT-2 Quickstart Notebook
• GPT-2 API Documentation
• GPT-2 Model Card
• KerasHub Beginner Guide
• KerasHub Model Publishing Guide

Installation

Keras and KerasHub can be installed with:

pip install -U -q keras-hub
pip install -U -q keras

Jax, TensorFlow, and Torch come preinstalled in Kaggle Notebooks. For instruction on installing them in another environment see the Keras Getting Started page.

Presets

The following model checkpoints are provided by the Keras team. Full code examples for each are available below.

Preset name	Parameters	Description
gpt2_base_en	124.44M	12-layer GPT-2 model where case is maintained. Trained on WebText.
gpt2_medium_en	354.82M	24-layer GPT-2 model where case is maintained. Trained on WebText.
gpt2_large_en	774.03M	36-layer GPT-2 model where case is maintained. Trained on WebText.
gpt2_extra_large_en	1.56B	48-layer GPT-2 model where case is maintained. Trained on WebText.
gpt2_base_en_cnn_dailymail	124.44M	12-layer GPT-2 model where case is maintained. Finetuned on the CNN/DailyMail summarization dataset.

Prompts

GPT-2 models are fine tuned on WebText. Prompting should follow text completion formatting. See the following for an example:

prompt = "Keras is a "

would have GPT-2 aim to complete the sentence.

Example Usage

import keras
import keras_hub
import numpy as np

Use generate() to do text generation.

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("gpt2_extra_large_en")
gpt2_lm.generate("I want to say", max_length=30)

# Generate with batched prompts.
gpt2_lm.generate(["This is a", "Where are you"], max_length=30)

Compile the generate() function with a custom sampler.

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("gpt2_extra_large_en")
gpt2_lm.compile(sampler="greedy")
gpt2_lm.generate("I want to say", max_length=30)

gpt2_lm.compile(sampler=keras_hub.samplers.BeamSampler(num_beams=2))
gpt2_lm.generate("I want to say", max_length=30)

Use generate() without preprocessing.

# Prompt the model with `5338, 318` (the token ids for `"Who is"`).
# Use `"padding_mask"` to indicate values that should not be overridden.
prompt = {
    "token_ids": np.array([[5338, 318, 0, 0, 0]] * 2),
    "padding_mask": np.array([[1, 1, 0, 0, 0]] * 2),
}

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset(
    "gpt2_extra_large_en",
    preprocessor=None,
)
gpt2_lm.generate(prompt)

Call fit() on a single batch.

features = ["The quick brown fox jumped.", "I forgot my homework."]
gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("gpt2_extra_large_en")
gpt2_lm.fit(x=features, batch_size=2)

Call fit() without preprocessing.

x = {
    "token_ids": np.array([[50256, 1, 2, 3, 4]] * 2),
    "padding_mask": np.array([[1, 1, 1, 1, 1]] * 2),
}
y = np.array([[1, 2, 3, 4, 50256]] * 2)
sw = np.array([[1, 1, 1, 1, 1]] * 2)

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset(
    "gpt2_extra_large_en",
    preprocessor=None,
)
gpt2_lm.fit(x=x, y=y, sample_weight=sw, batch_size=2)

Example Usage with Hugging Face URI

import keras
import keras_hub
import numpy as np

Use generate() to do text generation.

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("hf://keras/gpt2_extra_large_en")
gpt2_lm.generate("I want to say", max_length=30)

# Generate with batched prompts.
gpt2_lm.generate(["This is a", "Where are you"], max_length=30)

Compile the generate() function with a custom sampler.

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("hf://keras/gpt2_extra_large_en")
gpt2_lm.compile(sampler="greedy")
gpt2_lm.generate("I want to say", max_length=30)

gpt2_lm.compile(sampler=keras_hub.samplers.BeamSampler(num_beams=2))
gpt2_lm.generate("I want to say", max_length=30)

Use generate() without preprocessing.

# Prompt the model with `5338, 318` (the token ids for `"Who is"`).
# Use `"padding_mask"` to indicate values that should not be overridden.
prompt = {
    "token_ids": np.array([[5338, 318, 0, 0, 0]] * 2),
    "padding_mask": np.array([[1, 1, 0, 0, 0]] * 2),
}

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset(
    "hf://keras/gpt2_extra_large_en",
    preprocessor=None,
)
gpt2_lm.generate(prompt)

Call fit() on a single batch.

features = ["The quick brown fox jumped.", "I forgot my homework."]
gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset("hf://keras/gpt2_extra_large_en")
gpt2_lm.fit(x=features, batch_size=2)

Call fit() without preprocessing.

x = {
    "token_ids": np.array([[50256, 1, 2, 3, 4]] * 2),
    "padding_mask": np.array([[1, 1, 1, 1, 1]] * 2),
}
y = np.array([[1, 2, 3, 4, 50256]] * 2)
sw = np.array([[1, 1, 1, 1, 1]] * 2)

gpt2_lm = keras_hub.models.GPT2CausalLM.from_preset(
    "hf://keras/gpt2_extra_large_en",
    preprocessor=None,
)
gpt2_lm.fit(x=x, y=y, sample_weight=sw, batch_size=2)