System Info

all libraries are latest versions
python 3.11

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder
• My own task or dataset (give details below)

Reproduction

def get_linear_layers(model):
    """Extract linear layers from the model for LoRA configuration."""
    model_modules = str(model.modules)
    pattern = r'\((\w+)\): Linear'
    return list(set(re.findall(pattern, model_modules)))
    
     def _apply_lora(self, model):
        """Apply LoRA configuration to the model."""
        lora_config = LoraConfig(
            r=16,
            lora_alpha=32,
            lora_dropout=0.05,
            bias="none",
            target_modules=get_linear_layers(model),
            task_type="CAUSAL_LM",
        )
        model = get_peft_model(model, lora_config)

when i run the ft script the first warning is this :

UserWarning: Model with tie_word_embeddings=True and the tied_target_modules=['lm_head'] are part of the adapter. This can lead to complications, for example when merging the adapter or converting your model to formats other than safetensors. See for example #2018.

but it works fine and i get adapters in which i can run later using Peftmodel.from_pretrained(model,adapter)...

but when i want to run convert adapters to gguf so i merge them with the LLm to get a final version of gguf model that i can run for inference i run the following script using llama.cpp:

python convert_lora_to_gguf.py --base "/models/gemma-2-2b-it" --outfile "/gemma-2b/adapters.gguf" --outtype auto "/ft_models/gemma-2b/adapters"

INFO:lora-to-gguf:Loading base model: gemma-2-2b-it
INFO:hf-to-gguf:choosing --outtype bf16 from first tensor type (torch.float32)
INFO:gguf.gguf_writer:gguf: This GGUF file is for Little Endian only
INFO:lora-to-gguf:Exporting model...
Traceback (most recent call last):
  File "/home/ec2-user/abdullah/projects/llama.cpp/convert_lora_to_gguf.py", line 432, in <module>
    model_instance.write()
  File "/home/ec2-user/abdullah/projects/llama.cpp/convert_hf_to_gguf.py", line 434, in write
    self.prepare_tensors()
  File "/home/ec2-user/abdullah/projects/llama.cpp/convert_hf_to_gguf.py", line 298, in prepare_tensors
    for new_name, data_torch in (self.modify_tensors(data_torch, name, bid)):
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ec2-user/abdullah/projects/llama.cpp/convert_lora_to_gguf.py", line 408, in modify_tensors
    raise ValueError("lm_head is present in adapter, but is ignored in base model")
ValueError: lm_head is present in adapter, but is ignored in base model

Expected behavior

what should i do for this kind of issue?