Comparison between different instantiation of Peftmodel

[shedding-ash]

Is there any difference between

config = {
            "peft_type": "LORA",
            "task_type": "SEQ_2_SEQ_LM",
            "r": 8,
            "target_modules": ["q", "v"],
            "lora_dropout": 0.1,
         }
peft_config = get_peft_config(config)
model = AutoModelForSeq2SeqLM.from_pretrained("t5-base")
peft_model = PeftModelForSeq2SeqLM(model, peft_config)

and

config = LoraConfig(
            task_type="SEQ_2_SEQ_LM",
            r=8,
            target_modules=["q", "v"],
            lora_dropout=0.1,
        )
model = AutoModelForSeq2SeqLM.from_pretrained("t5-base")
peft_model = LoraModel(model, config)

[BenjaminBossan]

No, those two should be the same.