o ߥi0@sddlZddlmZddlZddlmZddlmZddlmZddl m Z ddl m Z ddl mZddlmZd d lmZmZd d lmZeZGd d d ejZGdddejZGdddejZGdddejZGdddejZe jejej dGdddeZ!dS)N)CrossEntropyLoss)ACT2FN)Models)MODELS)AttentionFillMaskModelOutput)logger)Tasks)MegatronBertModelMegatronBertPreTrainedModel)MegatronBertConfigc2eZdZfddZdejdejfddZZS)#MegatronBertPredictionHeadTransformcsVtt|j|j|_t|jtrt |j|_ n|j|_ tj |j|j d|_ dS)N)eps) super__init__nnLinear hidden_sizedense isinstance hidden_actstrrtransform_act_fn LayerNormlayer_norm_epsselfconfig __class__a/home/ubuntu/.local/lib/python3.10/site-packages/modelscope/models/nlp/megatron_bert/fill_mask.pyr&s   z,MegatronBertPredictionHeadTransform.__init__ hidden_statesreturncCs"||}||}||}|SN)rrrrr#r!r!r"forward0s   z+MegatronBertPredictionHeadTransform.forward__name__ __module__ __qualname__rtorchTensorr' __classcell__r!r!rr"r$s  rc$eZdZfddZddZZS)MegatronBertLMPredictionHeadcsLtt||_tj|j|jdd|_t t |j|_ |j |j_ dS)NF)bias) rrr transformrrr vocab_sizedecoder Parameterr,zerosr1rrr!r"r:s   z%MegatronBertLMPredictionHead.__init__cCs||}||}|Sr%)r2r4r&r!r!r"r'Hs  z$MegatronBertLMPredictionHead.forwardr)r*r+rr'r.r!r!rr"r08s r0cr )MegatronBertOnlyMLMHeadcstt||_dSr%)rrr0 predictionsrrr!r"rQs z MegatronBertOnlyMLMHead.__init__sequence_outputr$cC||}|Sr%)r9)rr:prediction_scoresr!r!r"r'U zMegatronBertOnlyMLMHead.forwardr(r!r!rr"r8Os r8cr/)MegatronBertOnlyNSPHeadcstt|jd|_dSN)rrrrrseq_relationshiprrr!r"r]s z MegatronBertOnlyNSPHead.__init__cCr;r%)rA)r pooled_outputseq_relationship_scorer!r!r"r'ar=zMegatronBertOnlyNSPHead.forwardr7r!r!rr"r>[s r>cr/)MegatronBertPreTrainingHeadscs(tt||_t|jd|_dSr?)rrr0r9rrrrArrr!r"ris  z%MegatronBertPreTrainingHeads.__init__cCs||}||}||fSr%)r9rA)rr:rBr<rCr!r!r"r'ns  z$MegatronBertPreTrainingHeads.forwardr7r!r!rr"rDgs rD) module_namecspeZdZddgZddgZdeffdd Zdd Zd d Z dd dZ dddZ Z S)MegatronBertForMaskedLMpoolerrA position_idszpredictions.decoder.biasrc s@t||jrtdt|dd|_t||_| dS)NzsIf you want to use `MegatronBertForMaskedLM` make sure `config.is_decoder=False` for bi-directional self-attention.F)add_pooling_layer) rr is_decoderrwarningr bertr8cls post_init)rrkwargsrr!r"r|s   z MegatronBertForMaskedLM.__init__cCs |jjjSr%rMr9r4)rr!r!r"get_output_embeddingss z-MegatronBertForMaskedLM.get_output_embeddingscCs||jj_dSr%rP)rnew_embeddingsr!r!r"set_output_embeddingssz-MegatronBertForMaskedLM.set_output_embeddingsNc  Cs| dur| n|jj} |j||||||||| | | d } | d}||}d}| dur:t}||d|jj| d}| sP|f| dd}|durN|f|S|St||| j| j |dS)a  Args: input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`): Indices of input sequence tokens in the vocabulary. Indices can be obtained using :class:`~modelscope.models.nlp.structbert.SbertTokenizer`. See :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for details. `What are input IDs? <../glossary.html#input-ids>`__ attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(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**. `What are attention masks? <../glossary.html#attention-mask>`__ token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0, 1]``: - 0 corresponds to a `sentence A` token, - 1 corresponds to a `sentence B` token. `What are token type IDs? <../glossary.html#token-type-ids>`_ position_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,config.max_position_embeddings - 1]``. `What are position IDs? <../glossary.html#position-ids>`_ head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`): Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``: - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`): Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert :obj:`input_ids` indices into associated vectors than the model's internal embedding lookup matrix. output_attentions (:obj:`bool`, `optional`): Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned tensors for more detail. output_hidden_states (:obj:`bool`, `optional`): Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for more detail. return_dict (:obj:`bool`, `optional`): Whether or not to return a :class:`~transformers.ModelOutput` instead of a plain tuple. labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]` Returns: Returns `modelscope.outputs.AttentionFillMaskModelOutput` Examples: >>> from modelscope.models import Model >>> from modelscope.preprocessors import Preprocessor >>> model = Model.from_pretrained('damo/nlp_megatronbert_backbone_base_std') >>> preprocessor = Preprocessor.from_pretrained('damo/nlp_megatronbert_backbone_base_std') >>> print(model(**preprocessor(('This is a test', 'This is also a test')))) N) attention_masktoken_type_idsrH head_mask inputs_embedsencoder_hidden_statesencoder_attention_maskoutput_attentionsoutput_hidden_states return_dictrr@)losslogitsr# attentions input_ids) ruse_return_dictrLrMrviewr3rr#r`)rrarTrUrHrVrWrXrYlabelsrZr[r\outputsr:r<masked_lm_lossloss_fctoutputr!r!r"r'sLT zMegatronBertForMaskedLM.forwardcKs|j}|d}|jjdurtd||jddf}tj||gdd}tj|df|jjtj|j d}tj||gdd}||dS)Nrz.The PAD token should be defined for generationr r])dim)dtypedevice)rarT) shaper pad_token_id ValueError new_zerosr,catfulllongrk)rrarT model_kwargs input_shapeeffective_batch_sizeattention_mask_new_zeros dummy_tokenr!r!r"prepare_inputs_for_generation s"     z5MegatronBertForMaskedLM.prepare_inputs_for_generation) NNNNNNNNNNNNr%) r)r*r+"_keys_to_ignore_on_load_unexpected_keys_to_ignore_on_load_missingr rrQrSr'rxr.r!r!rr"rFts, }rF)"r,torch.nnrtorch.utils.checkpointrtransformers.activationsrmodelscope.metainformodelscope.models.builderrmodelscope.outputsrmodelscope.utilsrloggingmodelscope.utils.constantrbackboner r configurationr get_loggerModulerr0r8r>rDregister_module fill_mask megatron_bertrFr!r!r!r"s(