Did you use ghost gradients? (gradients that tend to reactivate features that are at zero)

You should train both a feedforward network and a CNN on image classification on imagenet, to see if we see that the hessian of the CNN is more similar to the identity after training rather than the feedforward because of the image understanding priors.

Given that this method returns a numeric matrix, then it must be an Hessian evaluated at a point or the average Hessian of many points. Is the result the Hessian averaged over all training data? Is this average useful rather than just cancelling out high and low Hessian values

You can use the "mp-net2" model from sentence transformers for zero-shot classification (scalar product between the text and the embeddings of "sex" and "violence") decide a cut-off and you are done.

Thanks for the quick response, have you tried fine-tuning the new llama2 models on the data gathered so far to see if there is any interesting results? QLORA is pretty efficient for this.

It looks pretty cool! Adding a Google sign-in option would greatly broaden the reach of the game as most non-technical people do not have a Github account.

We could not find a "speak in French" vector after about an hour of effort, but it's possible we missed something straightforward

Did you try 10 or 20 simple French phrases with a positive sign and their translations with a negative sign?

Also try 1000 english words and 1000 french translations in case scale is the problem.

Also try:

"The following text is in English: ' "

"The following text is in French: ' "

with the second phrase written itself in French.

Very interesting, could you elaborate or give some links?

In my opinion Wearable health is highly neglected because older people are less tech savy than young people, so they use it less than younger people, but they would also benefit much more from the technology. If a 20 year old wears a smart watch that measures and records heart-rate it is almost only for fun, if a 60 year old does it, it could prevent and inform about important issues, but the 20 year old is much more likely to actually use it than the 60 year old.

I also asked ChatGPT, here are the six best ideas that it had (excluding electric bikes, as it was already my idea ;P) (cherry picked by me over 21):

Online education: Online education platforms like Coursera and Khan Academy were mature, widely available, intuitive, and cost no money for basic usage. They also had no regulatory barriers or moral issues and could be used by mutual agreement among one or a few people. Online education also saved a lot of time and played relatively well with the existing format of learning and education.

Digital wallets: Digital wallet technologies like Apple Pay and Google Pay were mature, widely available, intuitive, and cost no money for basic usage. They also had no regulatory barriers or moral issues and could be used by mutual agreement among one or a few people. Digital wallets also saved a lot of time and played relatively well with the existing format of financial transactions and management.

Wearable health monitoring devices: These devices, such as Fitbit and Apple Watch, are mature, widely available, intuitive, and cost no money for basic usage. They also have no regulatory barriers or moral issues and can be used by mutual agreement among one or a few people. They save time and money by allowing users to track their health metrics and stay motivated to maintain a healthy lifestyle.

Online language learning platforms: Platforms like Duolingo and Rosetta Stone are mature, widely available, intuitive, and cost no money for basic usage. They have no regulatory barriers or moral issues and can be used by mutual agreement among one or a few people. They save time and money by providing an accessible and effective way to learn a new language.

Home energy management systems: Devices like Nest Thermostat and Sense Home Energy Monitor are mature, widely available, intuitive, and cost no money for basic usage. They have no regulatory barriers or moral issues and can be used by mutual agreement among one or a few people. They save money and energy by providing users with insights into their home energy usage and allowing them to make informed decisions about energy conservation.

Collaborative writing and editing tools: Platforms like Google Docs and Microsoft Teams are mature, widely available, intuitive, and cost no money for basic usage. They have no regulatory barriers or moral issues and can be used by mutual agreement among one or a few people. They save time and make collaboration more efficient and effective, whether for school projects, business proposals, or creative writing.

Both of them are very reasonable, online education is accessible, almost free, and makes it possible to study even while holding a full time job, from a quick glance a great deal of your requirements are satisfied. 

Digital wallets I am less sure about, I never used one, but they look really convenient and easy to use, but I would need more info on how secure they are before starting to use them.

Overall, I think all of these ideas kind of fit your point.