謜i dZddlmZddlmZmZer ddlZddlmZeje Z dde e dzfdZy) z?AQLM (Additive Quantization of Language Model) integration file)should_convert_module)is_torch_availableloggingNmodules_to_not_convertc .ddlm}d}|jD]\}}t||st j d5t |tjr||j|j|jdu|j|j|j|j}t!||_|j%d|j'||d}ddd|st(j+d|S#1swYxYw) ai Public method that recursively replaces the Linear layers of the given model with AQLM quantized layers. Args: model (`torch.nn.Module`): The model to convert, can be any `torch.nn.Module` instance. modules_to_not_convert (`list[str]`, *optional*, defaults to `None`): A list of nn.Linear weights to not convert. If a parameter path is in the list (e.g. `lm_head.weight`), the corresponding module will not be converted. quantization_config (`AqlmConfig`): The quantization config object that contains the quantization parameters. r)QuantizedLinearFmetaN)bias in_group_sizeout_group_size num_codebooksnbits_per_codebookTzYou are loading your model using eetq but no linear modules were found in your model. Please double check your model architecture, or submit an issue on github if you think this is a bug.)aqlmr named_modulesrtorchdevice isinstancennLinear in_features out_featuresr r r rrtype source_clsrequires_grad_ set_submoduleloggerwarning)modelrquantization_configr has_been_replaced module_namemodule new_modules Z/mnt/e/genesis-system/.venv/lib/python3.12/site-packages/transformers/integrations/aqlm.pyreplace_with_aqlm_linearr&s%$224) V$[2HI  \\& ! )&")),,&&''D0"5"C"C#6#E#E"5"C"C':'M'M )-V %))%0##K<$(! ) ))&   L/ ) )s B'D  D )NN)__doc__quantizers.quantizers_utilsrutilsrrrtorch.nnr get_logger__name__rliststrr&r%r1sFB?/   H %+DI