bump burn: 0.13.0 -> 0.16.0

[HerrMuellerluedenscheid]

Fixes issue mentioned here: tracel-ai/burn#2845 (comment)

[HerrMuellerluedenscheid]

This fixes the example using feature: weight_embeded. However I'm running into the following error when using feature weights_file:

thread 'main' panicked at /Users/mariuskriegerowski/src/rust/burn-models/squeezenet-burn/target/release/build/squeezenet-burn-7fe83ccf42e0f30c/out/model/squeezenet1.rs:67:14:
Record file to exist.: Unknown("Unable to load record.\nMetadata has a different float type: Actual \"half::binary16::f16\", Expected \"f32\"\nMetadata has a different int type: Actual \"i16\", Expected \"i32\"\nMetadata has a different format: Actual \"burn_core::record::file::NamedMpkGzFileRecorder<burn_core::record::settings::HalfPrecisionSettings>\", Expected \"burn_core::record::file::NamedMpkFileRecorder<burn_core::record::settings::FullPrecisionSettings>\"\nMetadata has a different Burn version: Actual \"0.13.2\", Expected \"0.16.0\"\nError: Unknown(\"Syntax(\\\"missing field `bytes`\\\")\\nThe internal data format has changed since version 0.14.0. If you are trying to load a record saved in a previous version, use the `record-backward-compat` feature flag. Once you have saved the record in the new format, you can disable the feature flag.\\n\")")
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Not sure, yet what the issue is. Hints are appreciated :)

[antimora]

@HerrMuellerluedenscheid try cleaning the build like cargo clean. It's trying to load model for the previous version.

[HerrMuellerluedenscheid]

@HerrMuellerluedenscheid try cleaning the build like cargo clean. It's trying to load model for the previous version.

Thanks! yeah, don't know why I didn't do that. That fixed it :)

So, this is ready. I separated the actual bugfix in #58 but actually now you can also merge this and have both fixed together. As you like.

[antimora]

LGTM

Thanks for upgrading and fixing the issues.

[antimora]

I will merge once the CI passes.