+ j*:Rt^RIt^RIHt^RIHt]P !]4t!RR]4t !RR]4t !RR ]4t !R R ]4t R#) z#BARK model generation configurationN)GenerationConfig)loggingc:aa]tRt^toRtRV3RlltRtVtV;t#)BarkSemanticGenerationConfigsemanticc <\SV`!R RVRV RVRVRVRVRVRVR V/ VBWnWnWnWnWnWnWnVVn R #) a Class that holds a generation configuration for [`BarkSemanticModel`]. This configuration inherit from [`GenerationConfig`] and can be used to control the model generation. Read the documentation from [`GenerationConfig`] for more information. Args: eos_token_id (`int`, *optional*, defaults to 10_000): The id of the *end-of-sequence* token. renormalize_logits (`bool`, *optional*, defaults to `True`): Whether to renormalize the logits after applying all the logits processors (including the custom ones). It's highly recommended to set this flag to `True` as the search algorithms suppose the score logits are normalized but some logit processors break the normalization. max_new_tokens (`int`, *optional*, defaults to 768): The maximum numbers of tokens to generate, ignoring the number of tokens in the prompt. output_scores (`bool`, *optional*, defaults to `False`): Whether or not to return the prediction scores. See `scores` under returned tensors for more details. return_dict_in_generate (`bool`, *optional*, defaults to `False`): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. output_hidden_states (`bool`, *optional*, defaults to `False`): Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for more details. output_attentions (`bool`, *optional*, defaults to `False`): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more details. temperature (`float`, *optional*, defaults to 1.0): The value used to modulate the next token probabilities. do_sample (`bool`, *optional*, defaults to `False`): Whether or not to use sampling ; use greedy decoding otherwise. text_encoding_offset (`int`, *optional*, defaults to 10_048): Text encoding offset. text_pad_token (`int`, *optional*, defaults to 129_595): Text pad token. semantic_infer_token (`int`, *optional*, defaults to 129_599): Semantic infer token. semantic_vocab_size (`int`, *optional*, defaults to 10_000): Semantic vocab size. max_input_semantic_length (`int`, *optional*, defaults to 256): Max length of semantic input vector. semantic_rate_hz (`float`, *optional*, defaults to 49.9): Semantic rate in Hertz. min_eos_p (`float`, *optional*): Minimum threshold of the probability of the EOS token for it to be sampled. This is an early stopping strategy to mitigate potential unwanted generations at the end of a prompt. The original implementation suggests a default value of 0.2. temperature do_sample eos_token_idrenormalize_logitsmax_new_tokens output_scoresreturn_dict_in_generateoutput_hidden_statesoutput_attentionsN) super__init__text_encoding_offsettext_pad_tokensemantic_pad_tokensemantic_infer_tokensemantic_vocab_sizemax_input_semantic_lengthsemantic_rate_hz min_eos_p)selfr r r r rrrrr rrrrrrrkwargs __class__s&&&&&&&&&&&&&&&&&,w/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/bark/generation_configuration_bark.pyr%BarkSemanticGenerationConfig.__init__sB  #  &  2 * ( %< "6 0  %9!,".$8!#6 )B& 0")rrrrrrrr)'TiFFFF?Fi@'i;i?r"g33333H@N __name__ __module__ __qualname____firstlineno__ model_typer__static_attributes____classdictcell__ __classcell__r __classdict__s@@rrrsJU#U#r!rcFaa]tRt^ttoRtRV3RlV3RllltRtVtV;t#)BarkCoarseGenerationConfigcoarse_acousticsc&<V^8dQhRS[RS[/#)max_coarse_historysliding_window_len)int)formatr/s"r __annotate__'BarkCoarseGenerationConfig.__annotate__ws%J5J5 J5 J5r!c <\SV`!R RVRVRVRVRVRVRV/VBWnWnWnWnWnWnWnR#) aS Class that holds a generation configuration for [`BarkCoarseModel`]. This configuration inherit from [`GenerationConfig`] and can be used to control the model generation. Read the documentation from [`GenerationConfig`] for more information. Args: renormalize_logits (`bool`, *optional*, defaults to `True`): Whether to renormalize the logits after applying all the logits processors (including the custom ones). It's highly recommended to set this flag to `True` as the search algorithms suppose the score logits are normalized but some logit processors break the normalization. output_scores (`bool`, *optional*, defaults to `False`): Whether or not to return the prediction scores. See `scores` under returned tensors for more details. return_dict_in_generate (`bool`, *optional*, defaults to `False`): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. output_hidden_states (`bool`, *optional*, defaults to `False`): Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for more details. output_attentions (`bool`, *optional*, defaults to `False`): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more details. temperature (`float`, *optional*, defaults to 1.0): The value used to modulate the next token probabilities. do_sample (`bool`, *optional*, defaults to `False`): Whether or not to use sampling ; use greedy decoding otherwise. coarse_semantic_pad_token (`int`, *optional*, defaults to 12_048): Coarse semantic pad token. coarse_rate_hz (`int`, *optional*, defaults to 75): Coarse rate in Hertz. n_coarse_codebooks (`int`, *optional*, defaults to 2): Number of coarse codebooks. coarse_infer_token (`int`, *optional*, defaults to 12_050): Coarse infer token. max_coarse_input_length (`int`, *optional*, defaults to 256): Max length of input coarse vector. max_coarse_history (`int`, *optional*, defaults to 630): Max length of the output of the coarse acoustics model used in the fine generation step. sliding_window_len (`int`, *optional*, defaults to 60): The coarse generation step uses a sliding window to generate raw audio. rr r r rrrNr) rrcoarse_semantic_pad_tokencoarse_rate_hzn_coarse_codebookscoarse_infer_tokenmax_coarse_input_lengthr5r6)rr r rrrrr r<r=r>r?r@r5r6rrs&&&&&&&&&&&&&&&,rr#BarkCoarseGenerationConfig.__init__wsr  #   2 ( %< "6 0  *C&,"4"4'>$"4"4r!)r?r=r<r5r@r>r6)TFFFFr#Fi/Kr4i/r$iv<r%r.s@@rr1r1ts#JJ5J5J5r!r1c@aa]tRt^toRtRV3RlltRtRtVtV;t #)BarkFineGenerationConfigfine_acousticsc L<\SV`VR7W nW0nW@nR#)a:Class that holds a generation configuration for [`BarkFineModel`]. [`BarkFineModel`] is an autoencoder model, so should not usually be used for generation. However, under the hood, it uses `temperature` when used by [`BarkModel`] This configuration inherit from [`GenerationConfig`] and can be used to control the model generation. Read the documentation from [`GenerationConfig`] for more information. Args: temperature (`float`, *optional*): The value used to modulate the next token probabilities. max_fine_history_length (`int`, *optional*, defaults to 512): Max length of the fine history vector. max_fine_input_length (`int`, *optional*, defaults to 1024): Max length of fine input vector. n_fine_codebooks (`int`, *optional*, defaults to 8): Number of codebooks used. )rN)rrmax_fine_history_lengthmax_fine_input_lengthn_fine_codebooks)rrrHrIrJrrs&&&&&,rr!BarkFineGenerationConfig.__init__s'4 [1'>$%:" 0r!c R#)zt Overrides GenerationConfig.validate because BarkFineGenerationConfig don't use any parameters outside temperature. Nr)rrs&,rvalidate!BarkFineGenerationConfig.validatesr!)rHrIrJ)r#i) r&r'r(r)r*rrMr+r,r-r.s@@rrErEs!J1@  r!rEc\a]tRt^toRtR V3RlRllt]V3RlRl4tRtRt Vt R#) BarkGenerationConfigbarkNcV<V^8dQhRS[R,RS[R,RS[R,/#)r4semantic_configNcoarse_acoustics_configfine_acoustics_config)dict)r8r/s"rr9!BarkGenerationConfig.__annotate__s5/+/+/+"&/+ $d{ /+r!c Vf/p\PR4Vf/p\PR4Vf/p\PR4\R/VBVn\ R/VBVn\ R/VBVnW@nWPn R#)aClass that holds a generation configuration for [`BarkModel`]. The [`BarkModel`] does not have a `generate` method, but uses this class to generate speeches with a nested [`BarkGenerationConfig`] which uses [`BarkSemanticGenerationConfig`], [`BarkCoarseGenerationConfig`], [`BarkFineGenerationConfig`]. This configuration inherit from [`GenerationConfig`] and can be used to control the model generation. Read the documentation from [`GenerationConfig`] for more information. Args: semantic_config (`Dict`, *optional*): Semantic generation configuration. coarse_acoustics_config (`Dict`, *optional*): Coarse generation configuration. fine_acoustics_config (`Dict`, *optional*): Fine generation configuration. sample_rate (`int`, *optional*, defaults to 24_000): Sample rate. codebook_size (`int`, *optional*, defaults to 1024): Vector length for each codebook. NzMsemantic_config is None. initializing the semantic model with default values.zScoarse_acoustics_config is None. initializing the coarse model with default values.zOfine_acoustics_config is None. initializing the fine model with default values.r) loggerinforrUr1rVrErW sample_rate codebook_size)rrUrVrWr]r^rs&&&&&&,rrBarkGenerationConfig.__init__s<  " O KKg h " *&( # KKm n ($& ! KKi j;NoN'A'\D['\$%=%V@U%V"&*r!c,<V^8dQhRS[RS[RS[/#)r4rUrVrW)rr1rE)r8r/s"rr9rY%s)  5 "<  8  r!c rV!RRVP4RVP4RVP4/VB#)z Instantiate a [`BarkGenerationConfig`] (or a derived class) from bark sub-models generation configuration. Returns: [`BarkGenerationConfig`]: An instance of a configuration object rUrVrWr)to_dict)clsrUrVrWrs&&&&,rfrom_sub_model_configs+BarkGenerationConfig.from_sub_model_configs$sP +335 $;$C$C$E #8"?"?"A    r!c &\P!VP4pVPP 4VR&VP P 4VR&VP P 4VR&VPPVR&V#)z Serializes this instance to a Python dictionary. Override the default [`~PreTrainedConfig.to_dict`]. Returns: `dict[str, any]`: Dictionary of all the attributes that make up this configuration instance, rUrVrWr*) copydeepcopy__dict__rUrbrVrWrr*)routputs& rrbBarkGenerationConfig.to_dict9s}t}}-$($8$8$@$@$B !,0,H,H,P,P,R()*.*D*D*L*L*N&'#~~88| r!)rVr^rWr]rU)NNNi]rO) r&r'r(r)r*r classmethodrdrbr+r,)r/s@rrRrRs3J/+/+b  (r!rR) __doc__rggeneration.configuration_utilsrutilsr get_loggerr&r[rr1rErRrr!rrqsb* >   H %X##3X#vM5!1M5`' /' TY+Yr!