Zhincore

I'm all for it. Makes sense

Duplicate of #255

I get the following error when loading a model: ``` Missing key(s) in state_dict: "model.layers.0.self_attn.k_proj.bias", "model.layers.0.self_attn.o_proj.bias", "model.layers.0.self_attn.q_proj.bias", "model.layers.0.self_attn.v_proj.bias", "model.layers.0.mlp.down_proj.bias", "model.layers.0.mlp.gate_proj.bias", "model.layers.0.mlp.up_proj.bias", "model.layers.1.self_attn.k_proj.bias", "model.layers.1.self_attn.o_proj.bias", "model.layers.1.self_attn.q_proj.bias", "model.layers.1.self_attn.v_proj.bias", "model.layers.1.mlp.down_proj.bias", "model.layers.1.mlp.gate_proj.bias", "model.layers.1.mlp.up_proj.bias", "model.layers.2.self_attn.k_proj.bias", "model.layers.2.self_attn.o_proj.bias",...

Now it loads! Pretty quick too without triton doing it's thing. But I get `Offload_LlamaModel.forward() got an unexpected keyword argument 'position_ids'` when trying to generate, though.

> the moment I use monkey-patch, it seems to ignore pre_layer Did you not have that issue on other branches? I thought that's just the monkey-patch bypassing these features

I think we have to wait until monkey-patch is no longer required once LoRA is supported by the upstream libraries or something

I have the same issues :(

I don't think it's dependent on parameters, the real error is `AttributeError: 'dict' object has no attribute 'get_control'` (at least for me and OP), which seems to be a programming...

Still same issue EDIT: This was happening because I was trying to use my own script. RVC has 3 levels of try-except that re-try the `_path_audio_` method again, and now...

I have this issue with `HP5_only_main_vocal` model. This is how the resulting vocals look in audacity:  Right channel is totally f*cked, not only crackling