Some self transactions have input data for Ethscriptions and/or Facet that looks like trace data, so models try to interprets this as a normal ERC20 transfer or mint. Example:
data:,{"p":"erc-20","op":"mint","tick":"shte","id":"2667","amt":"1000"}
Transaction
This is problematic because the model is not very accurate interpreting this data and just makes additional details from thin air.
Example transaction:
0x8379e92c883d86cf581449c0ba7aa8214f6e4c6ae6685d153f8d9da79105e27f
Proposed solution:
Instruct model in the prompt to ignore input data in self transfers or fine tune another model on this specific input data type.
Some self transactions have input data for Ethscriptions and/or Facet that looks like trace data, so models try to interprets this as a normal ERC20 transfer or mint. Example: data:,{"p":"erc-20","op":"mint","tick":"shte","id":"2667","amt":"1000"} Transaction
This is problematic because the model is not very accurate interpreting this data and just makes additional details from thin air.
Example transaction: 0x8379e92c883d86cf581449c0ba7aa8214f6e4c6ae6685d153f8d9da79105e27f
Proposed solution: Instruct model in the prompt to ignore input data in self transfers or fine tune another model on this specific input data type.
Frequency: Low Severity: Medium