22-hours / cabrita

Finetuning InstructLLaMA with portuguese data
Apache License 2.0
558 stars 68 forks source link

Tokenizer bug on dictionary #15

Open vinidiol opened 11 months ago

vinidiol commented 11 months ago

When we tried to run the line: tokenizer = AutoTokenizer.from_pretrained("nlpcloud/instruct-gpt-j-fp16")

We got the error: TypeError: unhashable type: 'dict'

It could be a library bug? How to fix this problem?