Hugging Face

Join the Hugging Face community

and get access to the augmented documentation experience

Collaborate on models, datasets and Spaces

Faster examples with accelerated inference

Switch between documentation themes

Sign Up

to get started

This model was contributed to Hugging Face Transformers on 2022-09-14.

GPT-NeoX-Japanese

GPT-NeoX-Japanese, a Japanese language model based on GPT-NeoX. Japanese uses three types of characters (hiragana, katakana, kanji) and has a huge vocabulary. This model uses BPEEncoder V2, a sub-word tokenizer to handle the different characters.

The model also removes some bias parameters for better performance.

You can find all the original GPT-NeoX-Japanese checkpoints under the ABEJA organization.

This model was contributed by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori from ABEJA, Inc..

Click on the GPT-NeoX-Japanese models in the right sidebar for more examples of how to apply GPT-NeoX-Japanese to different language tasks.

The example below demonstrates how to generate text with Pipeline or the AutoModel, and from the command line.

from transformers import pipeline


pipeline = pipeline(task="text-generation",
                    model="abeja/gpt-neox-japanese-2.7b", device=0)
pipeline("人とAIが協調するためには、")

Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the Quantization overview for more available quantization backends.

The example below uses bitsandbytes to only quantize the weights to 4-bits.

from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig


quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype="float16"
)
model = AutoModelForCausalLM.from_pretrained(
    "abeja/gpt-neox-japanese-2.7b",
    quantization_config=quantization_config,
    device_map="auto"
)

tokenizer = AutoTokenizer.from_pretrained("abeja/gpt-neox-japanese-2.7b")
input_ids = tokenizer.encode("人とAIが協調するためには、", return_tensors="pt").to(model.device)
output = model.generate(input_ids)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Use the AttentionMaskVisualizer to better understand what tokens the model can and cannot attend to.

from transformers.utils.attention_visualizer import AttentionMaskVisualizer


visualizer = AttentionMaskVisualizer("abeja/gpt-neox-japanese-2.7b")
visualizer("<img>What is shown in this image?")

Resources

Refer to the Training a better GPT model: Learnings from PaLM blog post for more details about how ABEJA trained GPT-NeoX-Japanese.

GPTNeoXJapaneseConfig

>>> from transformers import GPTNeoXJapaneseConfig, GPTNeoXJapaneseModel

>>> # Initializing a GPTNeoXJapanese gpt-neox-japanese-2.7b style configuration
>>> configuration = GPTNeoXJapaneseConfig()

>>> # Initializing a model (with random weights) from the gpt-neox-japanese-2.7b style configuration
>>> model = GPTNeoXJapaneseModel(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config

GPTNeoXJapaneseTokenizer

>>> from transformers import GPTNeoXJapaneseTokenizer

>>> tokenizer = GPTNeoXJapaneseTokenizer.from_pretrained("abeja/gpt-neox-japanese-2.7b")
>>> # You can confirm both 慶応 and 慶應 are encoded to 17749
>>> tokenizer("吾輩は猫である🐯。実は慶応(慶應)大学出身")["input_ids"]
[30014, 26883, 26638, 27228, 25, 26650, 31732, 31679, 27809, 26638, 17749, 31592, 17749, 31593, 321, 1281]

>>> # Both 慶応 and 慶應 are decoded to 慶応
>>> tokenizer.decode(tokenizer("吾輩は猫である🐯。実は慶応(慶應)大学出身")["input_ids"])
'吾輩は猫である🐯。実は慶応(慶応)大学出身'

GPTNeoXJapaneseModel

Parameters

Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

The model will output the same cache format that is fed as input.

If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

A BaseModelOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (GPTNeoXJapaneseConfig) and inputs.

The GPTNeoXJapaneseModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

• last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size)) — Sequence of hidden-states at the output of the last layer of the model.

  If past_key_values is used only the last hidden-state of the sequences of shape (batch_size, 1, hidden_size) is output.

• past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

  Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if config.is_encoder_decoder=True in the cross-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

• hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

• attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

  Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

>>> from transformers import AutoTokenizer, GPTNeoXJapaneseModel
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("abeja/gpt-neox-japanese-2.7b")
>>> model = GPTNeoXJapaneseModel.from_pretrained("abeja/gpt-neox-japanese-2.7b")

>>> inputs = tokenizer("日本語のGPT-neoxがHugging Faceで使えます😀", return_tensors="pt")
>>> outputs = model(**inputs)

>>> last_hidden_states = outputs.last_hidden_state

GPTNeoXJapaneseForCausalLM

Parameters

Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

The model will output the same cache format that is fed as input.

If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

A CausalLMOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (GPTNeoXJapaneseConfig) and inputs.

The GPTNeoXJapaneseForCausalLM forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

• loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss (for next-token prediction).

• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

• past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

  Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

• hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

• attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

  Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

>>> from transformers import AutoTokenizer, GPTNeoXJapaneseForCausalLM, GPTNeoXJapaneseConfig
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("abeja/gpt-neox-japanese-2.7b")
>>> config = GPTNeoXJapaneseConfig.from_pretrained("abeja/gpt-neox-japanese-2.7b")
>>> config.is_decoder = True
>>> model = GPTNeoXJapaneseForCausalLM.from_pretrained("abeja/gpt-neox-japanese-2.7b", config=config)

>>> inputs = tokenizer("日本語のGPT-neoxがHugging Faceで使えます😀", return_tensors="pt")
>>> outputs = model(**inputs)

>>> prediction_logits = outputs.logits