79498175

Date: 2025-03-10 13:49:26
Score: 1
Natty:
Report link

The problem is using .cuda() to move the model to the GPU when loading the model using the BitsAndBytesConfig. I was able to get the error to go away by using the device_map argument when loading the model, e.g.,

model = AutoModelForCausalLM.from_pretrained(
    "ThetaCursed/Ovis1.6-Gemma2-9B-bnb-4bit",
    trust_remote_code=True,
    device_map='auto',
    **kwargs
)
Reasons:
  • Has code block (-0.5):
  • Unregistered user (0.5):
  • Low reputation (1):
Posted by: Tom