Llama2
개요
Llama2 모델은 Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Ya1smine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushkar Mishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom의 논문 LLaMA: Open Foundation and Fine-Tuned Chat Models에서 제안되었습니다. 채팅 어플리케이션에 맞게 미세 조정된 체크포인트를 포함된 7B에서 70B 범위의 매개변수를 가진 기초 언어 모델 모음입니다!
논문의 초록은 다음과 같습니다:
이 연구에서 우리는 70억에서 700억 파라미터의 범위에서 사전 훈련 및 미세 조정된 대규모 언어 모델(LLMs)의 모음인 Llama 2를 개발 및 공개합니다. Llama 2-Chat라고 불리는 미세 조정된 LLMs은 대화 사용 사례에 최적화되었습니다. 우리의 모델은 테스트한 대부분의 벤치마크에서 오픈 소스 채팅 모델보다 성능이 뛰어나며, 유용성과 안전성에 대한 인적 평가를 바탕으로 비공개 소스 모델을 대체할 수 있는 적절한 대안이 될 수 있습니다. 우리는 Llama 2-Chat의 미세 조정 및 안전성 향상의 접근 방식에 대한 자세한 설명을 제공하여 커뮤니티가 우리의 작업을 기반으로 LLMs의 책임있는 개발에 기여할 수 있도록 합니다.
여기에서 모든 Llama2 모델을 확인할 수 있습니다.
Llama2 모델은 bfloat16을 사용하여 훈련되었지만, 원래 추론은 float16을 사용합니다. 허브에 업로드된 체크포인트는 torch_dtype = 'float16'을 사용하며, 이는 AutoModel API에 의해 체크포인트를 torch.float32에서 torch.float16으로 캐스팅하는 데 사용됩니다.
온라인 가중치의 dtype은 model = AutoModelForCausalLM.from_pretrained("path", torch_dtype = "auto")를 사용하여 모델을 초기화할 때 torch_dtype="auto"를 사용하지 않는 한 대부분 관련이 없습니다. 그 이유는 모델이 먼저 다운로드될 것이고 (온라인 체크포인트의 dtype을 사용하여) 그다음에 기본 dtype인 torch로 캐스팅하고(torch.float32가 됨), 마지막으로 구성(configuration)에서 제공된 torch_dtype이 있는 경우 이를 사용하기 때문입니다.
모델을 float16에서 훈련하는 것은 권장되지 않으며 nan을 생성하는 것으로 알려져 있습니다. 따라서 모델은 bfloat16에서 훈련되어야 합니다.
🍯 팁:
- Llama2 모델의 가중치는 이 양식을 작성하여 얻을 수 있습니다.
- 아키텍처는 처음 버전의 Llama와 매우 유사하며, 이 논문의 내용에 따라 Grouped Query Attention (GQA)이 추가되었습니다.
config.pretraining_tp를 1과 다른 값으로 설정하면 더 정확하지만 느린 선형 레이어 계산이 활성화되어 원본 로짓과 더 잘 일치하게 됩니다.- 원래 모델은
pad_id = -1을 사용하는데, 이는 패딩 토큰이 없음을 의미합니다. 동일한 로직을 사용할 수 없으므로tokenizer.add_special_tokens({"pad_token":"<pad>"})를 사용하여 패딩 토큰을 추가하고 이에 따라 토큰 임베딩 크기를 조정해야 합니다. 또한model.config.pad_token_id를 설정해야 합니다. 모델의embed_tokens레이어는self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.config.padding_idx)로 초기화되어, 패딩 토큰 인코딩이 0을 출력하도록 합니다. 따라서 초기화 시에 전달하는 것을 권장합니다. - 양식을 작성하고 모델 체크포인트 접근 권한을 얻은 후에는 이미 변환된 체크포인트를 사용할 수 있습니다. 그렇지 않고 자신의 모델을 직접 변환하려는 경우, 변환 스크립트를 자유롭게 사용하세요. 스크립트는 다음과 같은 예시의 명령어로 호출할 수 있습니다:
python src/transformers/models/llama/convert_llama_weights_to_hf.py \
--input_dir /path/to/downloaded/llama/weights --model_size 7B --output_dir /output/path- 변환 후 모델과 토크나이저는 다음과 같이 로드할 수 있습니다:
from transformers import LlamaForCausalLM, LlamaTokenizer
tokenizer = LlamaTokenizer.from_pretrained("/output/path")
model = LlamaForCausalLM.from_pretrained("/output/path")스크립트를 실행하려면 모델을 float16 정밀도로 전부 호스트할 수 있을 만큼 충분한 CPU RAM이 필요합니다 (가장 큰 버전이 여러 체크포인트로 제공되더라도 각 체크포인트는 모델 가중치의 일부만을 포함하므로 모두 RAM에 로드해야 합니다). 75B 모델의 경우, 총 145GB의 RAM이 필요합니다.
- LLaMA 토크나이저는 sentencepiece를 기반으로 한 BPE 모델입니다. sentencepiece의 특징 중 하나는 시퀀스를 디코딩할 때 첫 번째 토큰이 단어의 시작이면 (예: “Banana”) 토크나이저는 문자열 앞에 접두사 공간을 추가하지 않는 것입니다.
이 모델은 Arthur Zucker가 Lysandre Debut의 도움을 받아 제공하였습니다. Hugging Face에서의 구현 코드는 여기의 GPT-NeoX 를 기반으로 합니다. 저자의 원래 코드는 여기에서 찾을 수 있습니다.
리소스
LLaMA2를 시작하는 데 도움이 될 Hugging Face의 공식 및 커뮤니티(🌎로 표시) 리소스 목록입니다. 여기에 새로운 리소스를 추가하기 위해서 Pull Request를 열어 주시면 검토하겠습니다! 리소스는 기존 리소스와 중복되지 않는 새로운 것을 보여주는 것이 이상적입니다.
- Llama 2 is here - get it on Hugging Face, Llama 2에 관한 블로그 포스트와 🤗 Transformers 및 🤗 PEFT와 함께 사용하는 방법에 대한 내용입니다.
- LLaMA 2 - Every Resource you need, LLaMA 2에 대해 알아보고 빠르게 시작하는 데 필요한 관련 리소스의 모음입니다.
- Google Colab에서 QLoRA와 4-bit 정밀도를 사용하여 Llama 2를 미세 조정하는 방법에 대한 노트북입니다. 🌎
- “Llama-v2-7b-guanaco” 모델을 4-bit QLoRA로 미세 조정하고 PDF에서 Q&A 데이터셋을 생성하는 방법에 대한 노트북입니다. 🌎
⚗️ 최적화
- Llama 2를 DPO로 미세 조정하기, TRL 라이브러리의 DPO 방법을 사용하여 특정 데이터셋에서 Llama 2를 미세 조정하는 방법을 안내하는 가이드입니다.
- 확장 가이드: Llama 2 명령어 조정, 입력에서 명령어를 생성하도록 Llama 2를 훈련시키는 방법을 안내하는 가이드로, 명령어를 따르는 모델에서 명령어를 주는 모델로 변환합니다.
- 개인 컴퓨터에서 QLoRA와 TRL을 사용하여 Llama 2 모델을 미세 조정하는 방법에 대한 노트북입니다. 🌎
⚡️ 추론
- AutoGPTQ 라이브러리의 GPTQ를 사용하여 Llama 2 모델을 양자화하는 방법에 대한 노트북입니다. 🌎
- 로컬 컴퓨터나 Google Colab에서 4-bit 양자화로 Llama 2 채팅 모델을 실행하는 방법에 대한 노트북입니다. 🌎
🚀 배포
- Amazon SageMaker에서 LLaMA 2 (7-70B) 미세 조정하기, Amazon SageMaker에서 QLoRA 미세 조정 및 배포에 이르기까지의 완전한 가이드입니다.
- Amazon SageMaker에서 Llama 2 7B/13B/70B 배포하기, 안전하고 확장 가능한 배포를 위해 Hugging Face의 LLM DLC 컨테이너를 사용하는 방법에 대한 가이드입니다.
LlamaConfig
class transformers.LlamaConfig
< source >( vocab_size = 32000 hidden_size = 4096 intermediate_size = 11008 num_hidden_layers = 32 num_attention_heads = 32 num_key_value_heads = None hidden_act = 'silu' max_position_embeddings = 2048 initializer_range = 0.02 rms_norm_eps = 1e-06 use_cache = True pad_token_id = None bos_token_id = 1 eos_token_id = 2 pretraining_tp = 1 tie_word_embeddings = False rope_theta = 10000.0 rope_scaling = None attention_bias = False attention_dropout = 0.0 mlp_bias = False head_dim = None **kwargs )
Parameters
- vocab_size (
int, optional, defaults to 32000) — Vocabulary size of the LLaMA model. Defines the number of different tokens that can be represented by theinputs_idspassed when calling LlamaModel - hidden_size (
int, optional, defaults to 4096) — Dimension of the hidden representations. - intermediate_size (
int, optional, defaults to 11008) — Dimension of the MLP representations. - num_hidden_layers (
int, optional, defaults to 32) — Number of hidden layers in the Transformer decoder. - num_attention_heads (
int, optional, defaults to 32) — Number of attention heads for each attention layer in the Transformer decoder. - num_key_value_heads (
int, optional) — This is the number of key_value heads that should be used to implement Grouped Query Attention. Ifnum_key_value_heads=num_attention_heads, the model will use Multi Head Attention (MHA), ifnum_key_value_heads=1the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details checkout this paper. If it is not specified, will default tonum_attention_heads. - hidden_act (
strorfunction, optional, defaults to"silu") — The non-linear activation function (function or string) in the decoder. - max_position_embeddings (
int, optional, defaults to 2048) — The maximum sequence length that this model might ever be used with. Llama 1 supports up to 2048 tokens, Llama 2 up to 4096, CodeLlama up to 16384. - initializer_range (
float, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices. - rms_norm_eps (
float, optional, defaults to 1e-06) — The epsilon used by the rms normalization layers. - use_cache (
bool, optional, defaults toTrue) — Whether or not the model should return the last key/values attentions (not used by all models). Only relevant ifconfig.is_decoder=True. - pad_token_id (
int, optional) — Padding token id. - bos_token_id (
int, optional, defaults to 1) — Beginning of stream token id. - eos_token_id (
int, optional, defaults to 2) — End of stream token id. - pretraining_tp (
int, optional, defaults to 1) — Experimental feature. Tensor parallelism rank used during pretraining. Please refer to this document to understand more about it. This value is necessary to ensure exact reproducibility of the pretraining results. Please refer to this issue. - tie_word_embeddings (
bool, optional, defaults toFalse) — Whether to tie weight embeddings - rope_theta (
float, optional, defaults to 10000.0) — The base period of the RoPE embeddings. - rope_scaling (
Dict, optional) — Dictionary containing the scaling configuration for the RoPE embeddings. NOTE: if you apply new rope type and you expect the model to work on longermax_position_embeddings, we recommend you to update this value accordingly. Expected contents:rope_type(str): The sub-variant of RoPE to use. Can be one of [‘default’, ‘linear’, ‘dynamic’, ‘yarn’, ‘longrope’, ‘llama3’], with ‘default’ being the original RoPE implementation.factor(float, optional): Used with all rope types except ‘default’. The scaling factor to apply to the RoPE embeddings. In most scaling types, afactorof x will enable the model to handle sequences of length x original maximum pre-trained length.original_max_position_embeddings(int, optional): Used with ‘dynamic’, ‘longrope’ and ‘llama3’. The original max position embeddings used during pretraining.attention_factor(float, optional): Used with ‘yarn’ and ‘longrope’. The scaling factor to be applied on the attention computation. If unspecified, it defaults to value recommended by the implementation, using thefactorfield to infer the suggested value.beta_fast(float, optional): Only used with ‘yarn’. Parameter to set the boundary for extrapolation (only) in the linear ramp function. If unspecified, it defaults to 32.beta_slow(float, optional): Only used with ‘yarn’. Parameter to set the boundary for interpolation (only) in the linear ramp function. If unspecified, it defaults to 1.short_factor(List[float], optional): Only used with ‘longrope’. The scaling factor to be applied to short contexts (<original_max_position_embeddings). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2long_factor(List[float], optional): Only used with ‘longrope’. The scaling factor to be applied to long contexts (<original_max_position_embeddings). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2low_freq_factor(float, optional): Only used with ‘llama3’. Scaling factor applied to low frequency components of the RoPEhigh_freq_factor(float, optional*): Only used with ‘llama3’. Scaling factor applied to high frequency components of the RoPE - attention_bias (
bool, optional, defaults toFalse) — Whether to use a bias in the query, key, value and output projection layers during self-attention. - attention_dropout (
float, optional, defaults to 0.0) — The dropout ratio for the attention probabilities. - mlp_bias (
bool, optional, defaults toFalse) — Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers. - head_dim (
int, optional) — The attention head dimension. If None, it will default to hidden_size // num_heads
This is the configuration class to store the configuration of a LlamaModel. It is used to instantiate an LLaMA model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the LLaMA-7B.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the
documentation from PretrainedConfig for more information.
>>> from transformers import LlamaModel, LlamaConfig
>>> # Initializing a LLaMA llama-7b style configuration
>>> configuration = LlamaConfig()
>>> # Initializing a model from the llama-7b style configuration
>>> model = LlamaModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.configLlamaTokenizer
class transformers.LlamaTokenizer
< source >( vocab_file unk_token = '<unk>' bos_token = '<s>' eos_token = '</s>' pad_token = None sp_model_kwargs: Optional = None add_bos_token = True add_eos_token = False clean_up_tokenization_spaces = False use_default_system_prompt = False spaces_between_special_tokens = False legacy = None add_prefix_space = True **kwargs )
Parameters
- vocab_file (
str) — Path to the vocabulary file. - unk_token (
strortokenizers.AddedToken, optional, defaults to"<unk>") — The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. - bos_token (
strortokenizers.AddedToken, optional, defaults to"<s>") — The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token. - eos_token (
strortokenizers.AddedToken, optional, defaults to"</s>") — The end of sequence token. - pad_token (
strortokenizers.AddedToken, optional) — A special token used to make arrays of tokens the same size for batching purpose. Will then be ignored by attention mechanisms or loss computation. - sp_model_kwargs (
Dict[str, Any],Optional, optional) — Will be passed to theSentencePieceProcessor.__init__()method. The Python wrapper for SentencePiece can be used, among other things, to set:-
enable_sampling: Enable subword regularization. -
nbest_size: Sampling parameters for unigram. Invalid for BPE-Dropout.nbest_size = {0,1}: No sampling is performed.nbest_size > 1: samples from the nbest_size results.nbest_size < 0: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm.
-
alpha: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout.
-
- add_bos_token (
bool, optional, defaults toTrue) — Whether or not to add anbos_tokenat the start of sequences. - add_eos_token (
bool, optional, defaults toFalse) — Whether or not to add aneos_tokenat the end of sequences. - clean_up_tokenization_spaces (
bool, optional, defaults toFalse) — Whether or not to cleanup spaces after decoding, cleanup consists in removing potential artifacts like extra spaces. - use_default_system_prompt (
bool, optional, defaults toFalse) — Whether or not the default system prompt for Llama should be used. - spaces_between_special_tokens (
bool, optional, defaults toFalse) — Whether or not to add spaces between special tokens. - legacy (
bool, optional) — Whether or not thelegacybehavior of the tokenizer should be used. Legacy is before the merge of #24622 and #25224 which includes fixes to properly handle tokens that appear after special tokens. Make sure to also setfrom_slowtoTrue. A simple example:legacy=True:
Construct a Llama tokenizer. Based on byte-level Byte-Pair-Encoding. The default padding token is unset as there is no padding token in the original model.
get_special_tokens_mask
< source >( token_ids_0: List token_ids_1: Optional = None already_has_special_tokens: bool = False ) → List[int]
Parameters
- token_ids_0 (
List[int]) — List of IDs. - token_ids_1 (
List[int], optional) — Optional second list of IDs for sequence pairs. - already_has_special_tokens (
bool, optional, defaults toFalse) — Whether or not the token list is already formatted with special tokens for the model.
Returns
List[int]
A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer prepare_for_model method.
create_token_type_ids_from_sequences
< source >( token_ids_0: List token_ids_1: Optional = None ) → List[int]
Parameters
- token_ids_0 (
List[int]) — List of ids. - token_ids_1 (
List[int], optional) — Optional second list of IDs for sequence pairs.
Returns
List[int]
List of token type IDs according to the given sequence(s).
Creates a mask from the two sequences passed to be used in a sequence-pair classification task. An ALBERT
sequence pair mask has the following format:
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence | second sequence |if token_ids_1 is None, only returns the first portion of the mask (0s).
save_vocabulary
< source >( save_directory filename_prefix: Optional = None ) → Tuple(str)
Save the vocabulary and special tokens file to a directory.
LlamaTokenizerFast
class transformers.LlamaTokenizerFast
< source >( vocab_file = None tokenizer_file = None clean_up_tokenization_spaces = False unk_token = '<unk>' bos_token = '<s>' eos_token = '</s>' add_bos_token = True add_eos_token = False use_default_system_prompt = False legacy = None add_prefix_space = None **kwargs )
Parameters
- vocab_file (
str, optional) — SentencePiece file (generally has a .model extension) that contains the vocabulary necessary to instantiate a tokenizer. - tokenizer_file (
str, optional) — tokenizers file (generally has a .json extension) that contains everything needed to load the tokenizer. - clean_up_tokenization_spaces (
bool, optional, defaults toFalse) — Whether or not to cleanup spaces after decoding, cleanup consists in removing potential artifacts like extra spaces. - unk_token (
strortokenizers.AddedToken, optional, defaults to"<unk>") — The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. - bos_token (
strortokenizers.AddedToken, optional, defaults to"<s>") — The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token. - eos_token (
strortokenizers.AddedToken, optional, defaults to"</s>") — The end of sequence token. - add_bos_token (
bool, optional, defaults toTrue) — Whether or not to add anbos_tokenat the start of sequences. - add_eos_token (
bool, optional, defaults toFalse) — Whether or not to add aneos_tokenat the end of sequences. - use_default_system_prompt (
bool, optional, defaults toFalse) — Whether or not the default system prompt for Llama should be used - legacy (
bool, optional) — Whether or not thelegacybehavior of the tokenizer should be used. Legacy is before the merge of #24622 and #25224 which includes fixes to properly handle tokens that appear after special tokens. Make sure to also setfrom_slowtoTrue. A simple example:legacy=True:
Construct a Llama tokenizer. Based on byte-level Byte-Pair-Encoding.
This uses notably ByteFallback and no normalization.
>>> from transformers import LlamaTokenizerFast
>>> tokenizer = LlamaTokenizerFast.from_pretrained("hf-internal-testing/llama-tokenizer")
>>> tokenizer.encode("Hello this is a test")
[1, 15043, 445, 338, 263, 1243]If you want to change the bos_token or the eos_token, make sure to specify them when initializing the model, or
call tokenizer.update_post_processor() to make sure that the post-processing is correctly done (otherwise the
values of the first token and final token of an encoded sequence will not be correct). For more details, checkout
[post-processors] (https://huggingface.co/docs/tokenizers/api/post-processors) documentation.
This tokenizer inherits from PreTrainedTokenizerFast which contains most of the main methods. Users should
refer to this superclass for more information regarding those methods.
get_special_tokens_mask
< source >( token_ids_0: List token_ids_1: Optional = None already_has_special_tokens: bool = False ) → A list of integers in the range [0, 1]
Parameters
- token_ids_0 (
List[int]) — List of ids of the first sequence. - token_ids_1 (
List[int], optional) — List of ids of the second sequence. - already_has_special_tokens (
bool, optional, defaults toFalse) — Whether or not the token list is already formatted with special tokens for the model.
Returns
A list of integers in the range [0, 1]
1 for a special token, 0 for a sequence token.
Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer prepare_for_model or encode_plus methods.
create_token_type_ids_from_sequences
< source >( token_ids_0: List token_ids_1: Optional = None ) → List[int]
Create the token type IDs corresponding to the sequences passed. What are token type IDs?
Should be overridden in a subclass if the model has a special way of building those.
Updates the underlying post processor with the current bos_token and eos_token.
LlamaModel
class transformers.LlamaModel
< source >( config: LlamaConfig )
Parameters
- config (LlamaConfig) —
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
from_pretrained()method to load the model weights. config — LlamaConfig
The bare LLaMA Model outputting raw hidden-states without any specific head on top.
This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
Transformer decoder consisting of config.num_hidden_layers layers. Each layer is a LlamaDecoderLayer
forward
< source >( input_ids: LongTensor = None attention_mask: Optional = None position_ids: Optional = None past_key_values: Union = None inputs_embeds: Optional = None use_cache: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None cache_position: Optional = None )
Parameters
- input_ids (
torch.LongTensorof shape(batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it.Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details. - attention_mask (
torch.Tensorof shape(batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in[0, 1]:- 1 for tokens that are not masked,
- 0 for tokens that are masked.
Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details.If
past_key_valuesis used, optionally only the lastinput_idshave to be input (seepast_key_values).If you want to change padding behavior, you should read
modeling_opt._prepare_decoder_attention_maskand modify to your needs. See diagram 1 in the paper for more information on the default strategy.- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- position_ids (
torch.LongTensorof shape(batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range[0, config.n_positions - 1]. - past_key_values (
Cacheortuple(tuple(torch.FloatTensor)), optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in thepast_key_valuesreturned by the model at a previous stage of decoding, whenuse_cache=Trueorconfig.use_cache=True.Two formats are allowed:
- a
Cacheinstance, see our kv cache guide; - Tuple of
tuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)). This is also known as the legacy cache format.
The model will output the same cache format that is fed as input. If no
past_key_valuesare passed, the legacy cache format will be returned.If
past_key_valuesare used, the user can optionally input only the lastinput_ids(those that don’t have their past key value states given to this model) of shape(batch_size, 1)instead of allinput_idsof shape(batch_size, sequence_length). - a
- inputs_embeds (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passinginput_idsyou can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_idsindices into associated vectors than the model’s internal embedding lookup matrix. - use_cache (
bool, optional) — If set toTrue,past_key_valueskey value states are returned and can be used to speed up decoding (seepast_key_values). - output_attentions (
bool, optional) — Whether or not to return the attentions tensors of all attention layers. Seeattentionsunder returned tensors for more detail. - output_hidden_states (
bool, optional) — Whether or not to return the hidden states of all layers. Seehidden_statesunder returned tensors for more detail. - return_dict (
bool, optional) — Whether or not to return aModelOutputinstead of a plain tuple. - cache_position (
torch.LongTensorof shape(sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily toposition_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length.
The LlamaModel 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.
LlamaForCausalLM
forward
< source >( input_ids: LongTensor = None attention_mask: Optional = None position_ids: Optional = None past_key_values: Union = None inputs_embeds: Optional = None labels: Optional = None use_cache: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None cache_position: Optional = None num_logits_to_keep: int = 0 ) → transformers.modeling_outputs.CausalLMOutputWithPast or tuple(torch.FloatTensor)
Parameters
- input_ids (
torch.LongTensorof shape(batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it.Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details. - attention_mask (
torch.Tensorof shape(batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in[0, 1]:- 1 for tokens that are not masked,
- 0 for tokens that are masked.
Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details.If
past_key_valuesis used, optionally only the lastinput_idshave to be input (seepast_key_values).If you want to change padding behavior, you should read
modeling_opt._prepare_decoder_attention_maskand modify to your needs. See diagram 1 in the paper for more information on the default strategy.- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- position_ids (
torch.LongTensorof shape(batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range[0, config.n_positions - 1]. - past_key_values (
Cacheortuple(tuple(torch.FloatTensor)), optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in thepast_key_valuesreturned by the model at a previous stage of decoding, whenuse_cache=Trueorconfig.use_cache=True.Two formats are allowed:
- a
Cacheinstance, see our kv cache guide; - Tuple of
tuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)). This is also known as the legacy cache format.
The model will output the same cache format that is fed as input. If no
past_key_valuesare passed, the legacy cache format will be returned.If
past_key_valuesare used, the user can optionally input only the lastinput_ids(those that don’t have their past key value states given to this model) of shape(batch_size, 1)instead of allinput_idsof shape(batch_size, sequence_length). - a
- inputs_embeds (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passinginput_idsyou can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_idsindices into associated vectors than the model’s internal embedding lookup matrix. - use_cache (
bool, optional) — If set toTrue,past_key_valueskey value states are returned and can be used to speed up decoding (seepast_key_values). - output_attentions (
bool, optional) — Whether or not to return the attentions tensors of all attention layers. Seeattentionsunder returned tensors for more detail. - output_hidden_states (
bool, optional) — Whether or not to return the hidden states of all layers. Seehidden_statesunder returned tensors for more detail. - return_dict (
bool, optional) — Whether or not to return aModelOutputinstead of a plain tuple. - cache_position (
torch.LongTensorof shape(sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily toposition_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length.Args — labels (
torch.LongTensorof shape(batch_size, sequence_length), optional): Labels for computing the masked language modeling loss. Indices should either be in[0, ..., config.vocab_size]or -100 (seeinput_idsdocstring). Tokens with indices set to-100are ignored (masked), the loss is only computed for the tokens with labels in[0, ..., config.vocab_size].num_logits_to_keep (
int, optional): Calculate logits for the lastnum_logits_to_keeptokens. If0, calculate logits for allinput_ids(special case). Only last token logits are needed for generation, and calculating them only for that token can save memory, which becomes pretty significant for long sequences or large vocabulary size.
Returns
transformers.modeling_outputs.CausalLMOutputWithPast or tuple(torch.FloatTensor)
A transformers.modeling_outputs.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 (LlamaConfig) and inputs.
-
loss (
torch.FloatTensorof shape(1,), optional, returned whenlabelsis provided) — Language modeling loss (for next-token prediction). -
logits (
torch.FloatTensorof 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 (
tuple(tuple(torch.FloatTensor)), optional, returned whenuse_cache=Trueis passed or whenconfig.use_cache=True) — Tuple oftuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head))Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
past_key_valuesinput) to speed up sequential decoding. -
hidden_states (
tuple(torch.FloatTensor), optional, returned whenoutput_hidden_states=Trueis passed or whenconfig.output_hidden_states=True) — Tuple oftorch.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 whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.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.
The LlamaForCausalLM 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.
Example:
>>> from transformers import AutoTokenizer, LlamaForCausalLM
>>> model = LlamaForCausalLM.from_pretrained("meta-llama/Llama-2-7b-hf")
>>> tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf")
>>> prompt = "Hey, are you conscious? Can you talk to me?"
>>> inputs = tokenizer(prompt, return_tensors="pt")
>>> # Generate
>>> generate_ids = model.generate(inputs.input_ids, max_length=30)
>>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
"Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."LlamaForSequenceClassification
class transformers.LlamaForSequenceClassification
< source >( config )
Parameters
- config (LlamaConfig) —
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
from_pretrained()method to load the model weights.
The LLaMa Model transformer with a sequence classification head on top (linear layer).
LlamaForSequenceClassification uses the last token in order to do the classification, as other causal models (e.g. GPT-2) do.
Since it does classification on the last token, it requires to know the position of the last token. If a
pad_token_id is defined in the configuration, it finds the last token that is not a padding token in each row. If
no pad_token_id is defined, it simply takes the last value in each row of the batch. Since it cannot guess the
padding tokens when inputs_embeds are passed instead of input_ids, it does the same (take the last value in
each row of the batch).
This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
forward
< source >( input_ids: Optional = None attention_mask: Optional = None position_ids: Optional = None past_key_values: Union = None inputs_embeds: Optional = None labels: Optional = None use_cache: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None )
Parameters
- input_ids (
torch.LongTensorof shape(batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it.Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details. - attention_mask (
torch.Tensorof shape(batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in[0, 1]:- 1 for tokens that are not masked,
- 0 for tokens that are masked.
Indices can be obtained using
AutoTokenizer. SeePreTrainedTokenizer.encode()andPreTrainedTokenizer.__call__()for details.If
past_key_valuesis used, optionally only the lastinput_idshave to be input (seepast_key_values).If you want to change padding behavior, you should read
modeling_opt._prepare_decoder_attention_maskand modify to your needs. See diagram 1 in the paper for more information on the default strategy.- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- position_ids (
torch.LongTensorof shape(batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range[0, config.n_positions - 1]. - past_key_values (
Cacheortuple(tuple(torch.FloatTensor)), optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in thepast_key_valuesreturned by the model at a previous stage of decoding, whenuse_cache=Trueorconfig.use_cache=True.Two formats are allowed:
- a
Cacheinstance, see our kv cache guide; - Tuple of
tuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)). This is also known as the legacy cache format.
The model will output the same cache format that is fed as input. If no
past_key_valuesare passed, the legacy cache format will be returned.If
past_key_valuesare used, the user can optionally input only the lastinput_ids(those that don’t have their past key value states given to this model) of shape(batch_size, 1)instead of allinput_idsof shape(batch_size, sequence_length). - a
- inputs_embeds (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passinginput_idsyou can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_idsindices into associated vectors than the model’s internal embedding lookup matrix. - use_cache (
bool, optional) — If set toTrue,past_key_valueskey value states are returned and can be used to speed up decoding (seepast_key_values). - output_attentions (
bool, optional) — Whether or not to return the attentions tensors of all attention layers. Seeattentionsunder returned tensors for more detail. - output_hidden_states (
bool, optional) — Whether or not to return the hidden states of all layers. Seehidden_statesunder returned tensors for more detail. - return_dict (
bool, optional) — Whether or not to return aModelOutputinstead of a plain tuple. - cache_position (
torch.LongTensorof shape(sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily toposition_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length. - labels (
torch.LongTensorof shape(batch_size,), optional) — Labels for computing the sequence classification/regression loss. Indices should be in[0, ..., config.num_labels - 1]. Ifconfig.num_labels == 1a regression loss is computed (Mean-Square loss), Ifconfig.num_labels > 1a classification loss is computed (Cross-Entropy).
The LlamaForSequenceClassification 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.