o ۷if%@sdZddlmZddlmZddlmZmZddlm Z m Z m Z ddl m Z ddlmZmZdd lmZeeZGd d d e ZGd d d eZd d gZdS)zCodeGen model configuration) OrderedDict)Mapping)AnyOptional)PreTrainedTokenizer TensorTypeis_torch_available)PretrainedConfig)OnnxConfigWithPast PatchingSpec)loggingcsXeZdZdZdZdddddZ            dfdd ZZS) CodeGenConfiga This is the configuration class to store the configuration of a [`CodeGenModel`]. It is used to instantiate a CodeGen 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 CodeGen [Salesforce/codegen-2B-mono](https://huggingface.co/Salesforce/codegen-2B-mono) architecture. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: vocab_size (`int`, *optional*, defaults to 50400): Vocabulary size of the CodeGen model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`CodeGenModel`]. n_positions (`int`, *optional*, defaults to 2048): The maximum sequence length that this model might ever be used with. Typically set this to something large just in case (e.g., 512 or 1024 or 2048). n_ctx (`int`, *optional*, defaults to 2048): This attribute is used in `CodeGenModel.__init__` without any real effect. n_embd (`int`, *optional*, defaults to 4096): Dimensionality of the embeddings and hidden states. n_layer (`int`, *optional*, defaults to 28): Number of hidden layers in the Transformer encoder. n_head (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. rotary_dim (`int`, *optional*, defaults to 64): Number of dimensions in the embedding that Rotary Position Embedding is applied to. n_inner (`int`, *optional*): Dimensionality of the inner feed-forward layers. `None` will set it to 4 times n_embd activation_function (`str`, *optional*, defaults to `"gelu_new"`): Activation function, to be selected in the list `["relu", "silu", "gelu", "tanh", "gelu_new"]`. resid_pdrop (`float`, *optional*, defaults to 0.0): The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. embd_pdrop (`int`, *optional*, defaults to 0.0): The dropout ratio for the embeddings. attn_pdrop (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention. layer_norm_epsilon (`float`, *optional*, defaults to 1e-05): The epsilon to use in the layer normalization layers. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). bos_token_id (`int`, *optional*, defaults to 50256): Beginning of stream token id. eos_token_id (`int`, *optional*, defaults to 50256): End of stream token id. tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether the model's input and output word embeddings should be tied. Note that this is only relevant if the model has a output word embedding layer. Example: ```python >>> from transformers import CodeGenConfig, CodeGenModel >>> # Initializing a CodeGen 6B configuration >>> configuration = CodeGenConfig() >>> # Initializing a model (with random weights) from the configuration >>> model = CodeGenModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```codegen n_positionsn_embdn_headn_layer)max_position_embeddings hidden_sizenum_attention_headsnum_hidden_layers@Ngelu_newh㈵>{Gz?TPFc s||_||_||_||_||_||_||_||_| |_| |_ | |_ | |_ | |_ ||_ ||_||_||_tjd|||d|dS)N) bos_token_id eos_token_idtie_word_embeddings) vocab_sizen_ctxrrrrn_inner rotary_dimactivation_function resid_pdrop embd_pdrop attn_pdroplayer_norm_epsiloninitializer_range use_cacher#r$super__init__)selfr'rr(rrrr*r)r+r,r-r.r/r0r1r#r$r%kwargs __class__r&g/home/ubuntu/vllm_env/lib/python3.10/site-packages/transformers/models/codegen/configuration_codegen.pyr3gs, zCodeGenConfig.__init__)rrrrrrrNrrrrr r!Tr"r"F)__name__ __module__ __qualname____doc__ model_type attribute_mapr3 __classcell__r&r&r6r8rs6@ rcseZdZ   ddededeeedeffdd Z e d e ee e efffd d Z e d e fd dZe d e fddZ    ddede de dedeed e eeff fdd Ze d e fddZZS)CodeGenOnnxConfigdefaultNFconfigtaskpatching_specsuse_pastcs2tj||||dt|jddsd|j_dSdS)N)rCrDrE pad_token_idr)r2r3getattr_configrF)r4rBrCrDrEr6r&r8r3s zCodeGenOnnxConfig.__init__returncCsJtddddi}|jr|j|ddddd|d<|Sddd|d<|S) N input_idsbatchsequence)rinputs) directionzpast_sequence + sequenceattention_mask)rrEfill_with_past_key_values_)r4 common_inputsr&r&r8rNszCodeGenOnnxConfig.inputscC|jjSN)rHrr4r&r&r8 num_layerszCodeGenOnnxConfig.num_layerscCrSrT)rHrrUr&r&r8rrWz%CodeGenOnnxConfig.num_attention_heads tokenizer batch_size seq_lengthis_pair frameworkc stt|j|||||d}td|di}|jrItstdddl|dj\}} | d} ||j | |j j |j ffddt |j D|d<|d |d <|jrj|d j} j|d j|| | d gd d |d <|S) N)rZr[r\r]rJzACannot generate dummy past_keys inputs without PyTorch installed.rcs g|] }fqSr&)zeros).0_ past_shapetorchr&r8 sz;CodeGenOnnxConfig.generate_dummy_inputs..past_key_valuesrP)dtyperM)dim)r2r generate_dummy_inputsrrEr ValueErrorrdshaperrHrrangerVrgcatones) r4rYrZr[r\r]rRordered_inputsrKseqlenpast_key_values_length mask_dtyper6rbr8ris2       z'CodeGenOnnxConfig.generate_dummy_inputscCsdS)N r&rUr&r&r8default_onnx_opsetsz$CodeGenOnnxConfig.default_onnx_opset)rANF)rXrXFN)r9r:r;r strrlistr boolr3propertyrintrNrVrrrrrirtr?r&r&r6r8r@sL    ,r@N)r< collectionsrcollections.abcrtypingrrrrr configuration_utilsr onnxr r utilsr get_loggerr9loggerrr@__all__r&r&r&r8s     x Q