ck_one

Congrats on the launch!

In what way is "prefix-aware embedding models trained with contrastive loss" better than the standard embedding model provided by OpenAI?

"added in learning from feedback and time based decay" => Sounds interesting! Have you seen significant gains in precision and recall here?

It looks like you are using NextJS app dir + external backend. Why did you decide against NextJs for frontend and backend? Are you happy with your choice?

Can you share some insights about your dev process? App locks pretty slick! Congrats!

What allows you to ship quickly and in high quality?

What’s your use case for it?

Why do you need ChatGPT for the API calls?

Biggest project to restore hair cells and ultimately hearing is probably this one: https://hearinghealthfoundation.org/hearing-restoration-proj...

It was shown in a couple of papers that we can restore hair cells in mammals. Damaged hair cells are the root cause for the majority of people with hearing loss & tinnitus. The most promising path seems to use so called supporting cells in the inner ear and convert them into hair cells. Researchers are getting closer and closer every year. I think we are now at a point where it's not a question of if but rather when.

Here is a quote from one of the leading scientist in the field:

What is needed to help make HRP goals happen? Frankly, funding to keep our research moving forward. A postdoctoral fellow with five to six years of training starts out on a modest salary of about $45,000, plus $12,000 in benefits. So that’s $57,000 before they even pick up a test tube in the lab. Each person will typically use between $15,000- $20,000 a year in supplies and chemicals. Simply maintaining a single cage of mice for one year costs $210, and my lab can use between 300-500 cages of mice for our experiments! HHF and its donors have been extremely generous in their support, however with additional funding the output from the consortium could be significantly greater and accelerate the pace to a cure.

Link: https://hearinghealthfoundation.org/spotlight/groves

Overall the field of hearing restoration still only receives tiny amounts of funding (<200 Mio). The research is in a vacuum phase. It's not proven out enough for Big Pharma to come. Relying on small government grants makes it difficult to get the research to a stage where it's attractive enough for Big Pharma.

Best bet at this point is probably when a former big tech executives would get hearing loss/tinnitus and then decides to put real money behind the problem. Bryan Johnson who created the Blueprint program has hearing loss but I guess he is not wealthy enough to make a difference.

EDIT (to put numbers into perspective):

The size of the problem: Sensorineural hearing loss disables over 360 million people worldwide. Irrespective of its cause and severity, hearing loss can have a large impact on people’s health and well-being. The treatment of hearing loss is currently limited to the use of hearing aids or devices surgically implanted in the middle or inner ear. These devices often perform poorly in noisy environments and can be very costly. It has been estimated that the costs of untreated hearing loss are €213 billion in Europe alone each year.

The funding (EU): An international consortium of 7 partners has been awarded a €5,8 million European Commission Horizon 2020 grant to develop and test a new drug to treat hearing loss caused by the loss of sensory hair cells.

Do you have a waitlist?

I am in the same situation. Would love for somebody with more knowledge to jump in and give some hints.

Does your Auth solution support Server Components from NextJS?

Have you tried MMS on real world data or is it just assumption?

When I first starting doing more research I was surprised by the low funding and the fairly big number of affected patients. Why is there not more funding? Research is making steady progress so the field doesn't appear to be a money sink with nothing to show for.

Thanks for following up.

I think there are different forms of hearing loss (hidden hearing loss, noice induced hearing loss, middle ear hearing loss, ...). I think noice induced hearing loss is most common nowadays and is caused by structural damage to hair cells. Regenerating hair cells might cure this particular form of hearing loss but not others. I think the one you mentioned has a different root cause for which new hair cells are only part of the puzzle.

Are both cell types functional?

Can you point me to some sources that back up your claim that neurons are damaged first and only later hair cells are lost as well. I always thought it’s the other way round.

Code generation works better for me compared to GPT-4 and it's a lot faster.

Interesting, thank you! Have you played around with different prompts to generate these questions?

How do you generate the questions and how do you make sure to not lose information?

E.g. Today I woke up at 9.am, had a light breakfast and then went on a run in Golden Gate Park.

What questions do you generate from this sentence?

I am doing something similar. How well does the search for you? Do you have any best practices for creating embeddings e.g. creating hierarchical embeddings? How much do you clean the data from videos/podcasts?

For niche topics you probably can.

It looks like you are using the bing api to search the web and then somehow integrate the result into the answer. Will do more digging ;)

Cool project and it seems like you and your team have working on it long before the hype began.

It's a good step up from using pure GPT4.

What do you think how they built it? How do you access content from other pages so quickly?

My guess is that they have crawled a lot of popular developer docs pages (Mozilla, Stackoverflow, Youtube, etc) and created embeddings for all paragraphs on these sites. Then for each search query they use a clever prompt + use the knowledge from the embeddings look up.

How does it work?

Why is the historical dataset needed?

Oh you are right. Do you know why there is the 10% threshold?

The idea was to incorporate the main company in the US as a Delaware C and the holding company in Germany.

Do you know how you would get penalised?

Is there no similar instrument in the US which allows you to not withdraw money & delay the taxation event?

I am still in team Europe and believe that taxes are important and that rich individuals can't take better care of public infra. I just don't want to make stupid mistakes at the beginning with my company structure.

Interesting, thanks for your comment. The question is then if the US will also tax my German holding company even if I don't withdraw any money from it to my personal account. If the US also taxes my holding company as if it were my personal income then there is no point in setting up a holding company for which I have to fly to Germany to make the signature ;)

That's a good question! I don't see a reason why it shouldn't work. There shouldn't be a difference between a founder holding shares and an employee holding shares. Not sure how options play into it though.

Do we learn anything interesting about Tether through the Celsius bankruptcy? Will Celsius be forced to reveal how their relationship with Tether worked?

Sounds super useful!

Where did you get the data from? Do you still have the code somewhere?