[tyfon]

GPT-4 will also happily invent python libraries that doesn't exist to enable some functionality in the code it produces.

I don't think this is a Bard only feature.

It's not as bad as the 65B llama model I run on my (amd) PC though, especially with quantized weights it tends to stop coding at some point and repeat the last line over and over. The 30B unquantized model seems better in this particular thing.

[AviationAtom]

With LLaMA it's all about the tuning. Using an Alpaca-tunes 7B yields pretty reasonable results, considering it's constraints.

[BryceSchroeder]

You might want to consider the sampler you're using. Consider using a repetition penalty.