h @ddlmZmZddlZddlmZddlmZmZddl m Z ddl m Z ddl mZdd lmZdd lmZdd lmZd d lmZmZmZmZmZmZmZmZmZd dlm Z ejBe"Z#GddeZ$GddeZ%GddeZ&GddeZ'Gdde Z(GddeZ)GddeZ*GddeZ+Gdd eZ,gd!Z-y)")CallableOptionalN)Cache)PretrainedConfiglayer_type_validation)FlashAttentionKwargs)rope_config_validation)ALL_ATTENTION_FUNCTIONS)Unpack)logging)deprecate_kwarg) LlamaAttentionLlamaDecoderLayerLlamaForCausalLMLlamaForQuestionAnsweringLlamaForSequenceClassificationLlamaForTokenClassificationLlamaPreTrainedModelapply_rotary_pos_embeager_attention_forward) Qwen2ModelceZdZdZdZdgZddddddddZdgdgfd d gd gfd gd gfd Z d fd ZxZ S) SmolLM3Configa This is the configuration class to store the configuration of a [`SmolLM3Model`]. It is used to instantiate a SmolLM3 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 SmolLM3 3B. e.g. [HuggingFaceTB/SmolLM3-3B](https://huggingface.co/HuggingFaceTB/SmolLM3-3B) 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 128256): Vocabulary size of the SmolLM3 model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`SmolLM3Model`] hidden_size (`int`, *optional*, defaults to 2048): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 11008): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 36): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. num_key_value_heads (`int`, *optional*, defaults to 4): This is the number of key_value heads that should be used to implement Grouped Query Attention. If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if `num_key_value_heads=1` the 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](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `16`. hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): The non-linear activation function (function or string) in the decoder. max_position_embeddings (`int`, *optional*, defaults to 32768): The maximum sequence length that this model might ever be used with. 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 to `True`): Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True`. pad_token_id (`int`, *optional*, defaults to 128004): The id of the padding token. bos_token_id (`int`, *optional*, defaults to 128000): The id of the beginning of sentence token. eos_token_id (`int`, *optional*, defaults to 128001): The id of the end of sentence token. rope_theta (`float`, *optional*, defaults to 2000000.0): 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 longer `max_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, a `factor` of 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 the `factor` field 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 2 `long_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 2 `low_freq_factor` (`float`, *optional*): Only used with 'llama3'. Scaling factor applied to low frequency components of the RoPE `high_freq_factor` (`float`, *optional*): Only used with 'llama3'. Scaling factor applied to high frequency components of the RoPE use_sliding_window (`bool`, *optional*, defaults to `False`): Whether to use sliding window attention. sliding_window (`int`, *optional*): Sliding window attention (SWA) window size. If not specified, will default to `None`. no_rope_layers (`List[int]`, *optional*): List with at least the same length as the number of layers in the model. A `1` at an index position indicates that the corresponding layer will use RoPE, while a `0` indicates that it's a NoPE layer. no_rope_layer_interval (`int`, *optional*, defaults to 4): If `no_rope_layers` is `None`, it will be created using a NoPE layer every `no_rope_layer_interval` layers. layer_types (`list`, *optional*): Attention pattern for each layer. Automatically computed based on sliding window and NoPE settings. attention_bias (`bool`, *optional*, defaults to `False`): 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. ```python >>> from transformers import SmolLM3Model, SmolLM3Config >>> # Initializing a SmolLM3 style configuration >>> configuration = SmolLM3Config() >>> # Initializing a model from the SmolLM3 style configuration >>> model = SmolLM3Model(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```smollm3past_key_valuescolwiserowwise)zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projzlayers.*.mlp.gate_projzlayers.*.mlp.up_projzlayers.*.mlp.down_proj input_ids inputs_embeds hidden_statesattention_mask) embed_tokenslayersnormc t|d| | |d|||_||_||_||_||_||_||_||_ ||_ ||}||_ ||_ | |_ | |_| |_||_||_||_||_|1t)|Dcgc]}t+|dz|zdk7c}|_n||_||_|Jg}t)|D]:}|j,|}|r||s|j1d*|j1d<||_t5|j2|j"*d|j"vr|j"d|j"d<t7|ycc}w) N) pad_token_id bos_token_id eos_token_idrsliding_attentionfull_attentiontype rope_type)super__init__ vocab_sizemax_position_embeddingsmlp_bias hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headsuse_sliding_windowsliding_windownum_key_value_heads hidden_actinitializer_range rms_norm_eps use_cache rope_theta rope_scalingattention_biasattention_dropoutrangeintno_rope_layersno_rope_layer_intervalappend layer_typesrr )selfr3r6r7r8r9r<r=r4r>r?r@r(r)r*rArBr:r;rGrHrJrCrDr5kwargs layer_idxhas_rope __class__s i/var/www/html/eduruby.in/venv/lib/python3.12/site-packages/transformers/models/smollm3/modular_smollm3.pyr2zSmolLM3Config.__init__s8  %%%   %'>$  &!2!2#6 "4,  &"5 #6 $!2("$(,!2  !TYZkTl#GPY]&<<AB#D #1D &<#  K"#45 9 ..y9%.*DX&&':;&&'78  9'd../    (Vt7H7H-H-1->->v-FD  k *t$3#s&F)iii+$siluig{Gz?gư>Tiiig>ANFNNrSNFF) __name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inferencebase_model_tp_planbase_model_pp_planr2 __classcell__rOs@rPrr-sq fJ#4"5&/%.%.%."+ )"+&(9:#%568IJ!"_$56 %  3T%T%rc,eZdZdedeffd Zeddd ddejd e ejejfd e ejde e d e ejd e ed e eje ejffdZxZS)SmolLM3AttentionconfigrMct||||j||_|jr$|j |dk(r|j |_yd|_y)Nr,)r1r2rGuse_roper:rJr;rKrcrMrOs rPr2zSmolLM3Attention.__init__ sb +--i8 ((V-?-? -JNa-a  ! !  r`past_key_valuerz4.58)new_nameversionr"position_embeddingsr#cache_positionrLreturnc ^|jdd}g|d|j}|j|j|j dd} |j |j|j dd} |j |j|j dd} |jr|\} } t| | | | \} } |%d|i}|j| | |j|\} } t}|jjdk7rt|jj}||| | | |f|jsdn |j |j"|j$d|\}}|j&g|dj)}|j+|}||fS)Nr+rrkeagerrU)dropoutscalingr;)shapehead_dimq_projview transposek_projv_projrerupdaterMrrc_attn_implementationr trainingrDrqr;reshape contiguouso_proj)rKr"rjr#rrkrL input_shape hidden_shape query_states key_states value_statescossin cache_kwargsattention_interface attn_output attn_weightss rPforwardzSmolLM3Attention.forwards$))#2. 88b8$--8 {{=166|DNNqRST [[/44\BLLQPQR {{=166|DNNqRST ==*HC';L*VY[^'_ $L*  &,n=L'6'='=j,X\XfXfht'u $J (? ;; + +w 6"9$++:Z:Z"[ $7      % $}}C$2H2HLL.. %  % ! \*k));;;;FFH kk+. L((r`)NN)rVrWrXrrFr2rtorchTensortuplerr LongTensorr r rr^r_s@rPrbrb s }  %0A6R ,059 *)||*)#5<<#=>*)!. *) "% *) !!1!12 *)-.*) u||Xell33 4*)S*)r`rbc(eZdZdedeffd ZxZS)SmolLM3DecoderLayerrcrMcNt||||j||_y)N)r1r2rJattention_typerfs rPr2zSmolLM3DecoderLayer.__init__Ds% +$00;r`)rVrWrXrrFr2r^r_s@rPrrCs<}<<rs & JB95&0   .   H %Z%$Z%z6)~6)r<+<  1  :  )  'E  $?  ";  r`