train.sh in the helpsteer and DPO examples errs with AttributeError: 'Linear' object has no attribute 'name'

[kylewadegrove]

train.sh loads models correctly but then throws an AttributeError when the linear variable as no field 'name'. Below is for helpsteer example but encountered this on DPO as well. This is with sillm-mx 0.1.2 installed via pip.

Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in _run_code
File "/opt/anaconda3/envs/llm/lib/python3.12/site-packages/sillm/lora.py", line 78, in
model.init_lora(**lora_config)
File "/opt/anaconda3/envs/llm/lib/python3.12/site-packages/sillm/training/lora.py", line 234, in init_lora
(key, LoRALinear.from_linear(module, rank=rank, alpha=alpha, dropout=dropout, scale=scale))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/anaconda3/envs/llm/lib/python3.12/site-packages/sillm/training/lora.py", line 60, in from_linear
lora_lin.name = linear.name
^^^^^^^^^^^
File "/opt/anaconda3/envs/llm/lib/python3.12/site-packages/mlx/nn/layers/base.py", line 137, in getattr
super(Module, self).getattribute(key)
AttributeError: 'Linear' object has no attribute 'name'