#openeurollm — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #openeurollm, aggregated by home.social.
-
@waldschnecke I am really with you on this one: That would be a very dumb move.
But I have to repeat myself, I use open source models. Even if their training data were scraped illegally and/or against my moral principles, still the company who did that has no material benefit from me using the OSS model they released. Maybe the OSS models are "appetizers" they release for free in hope to make me want to use their paid subscription models which are "even better" but that won't happen. Incidentally, there are European non-commercial efforts such as #OpenEuroLLM which train on legal training data only . Once their model will work sufficiently well, I will switch over.
On the other hand, I cannot afford to ignore and reject the technology outright because others in my business have started using it and I have the choice to use it as well or get left behind. In this scenario, I prefer advocating for responsible use (including not paying the tech brows, knowing what you are doing, and keeping the human in charge) over outright rejection. I am not alone, and Joplin tries to cater for both kinds of responsible users: Those of your type who outright reject it (they get the default setting), and those of my type who try to use it responsibly.
Rejecting the software for giving users the choice feels kind of overdoing it if you ask me.
-
@waldschnecke I am really with you on this one: That would be a very dumb move.
But I have to repeat myself, I use open source models. Even if their training data were scraped illegally and/or against my moral principles, still the company who did that has no material benefit from me using the OSS model they released. Maybe the OSS models are "appetizers" they release for free in hope to make me want to use their paid subscription models which are "even better" but that won't happen. Incidentally, there are European non-commercial efforts such as #OpenEuroLLM which train on legal training data only . Once their model will work sufficiently well, I will switch over.
On the other hand, I cannot afford to ignore and reject the technology outright because others in my business have started using it and I have the choice to use it as well or get left behind. In this scenario, I prefer advocating for responsible use (including not paying the tech brows, knowing what you are doing, and keeping the human in charge) over outright rejection. I am not alone, and Joplin tries to cater for both kinds of responsible users: Those of your type who outright reject it (they get the default setting), and those of my type who try to use it responsibly.
Rejecting the software for giving users the choice feels kind of overdoing it if you ask me.
-
@waldschnecke I am really with you on this one: That would be a very dumb move.
But I have to repeat myself, I use open source models. Even if their training data were scraped illegally and/or against my moral principles, still the company who did that has no material benefit from me using the OSS model they released. Maybe the OSS models are "appetizers" they release for free in hope to make me want to use their paid subscription models which are "even better" but that won't happen. Incidentally, there are European non-commercial efforts such as #OpenEuroLLM which train on legal training data only . Once their model will work sufficiently well, I will switch over.
On the other hand, I cannot afford to ignore and reject the technology outright because others in my business have started using it and I have the choice to use it as well or get left behind. In this scenario, I prefer advocating for responsible use (including not paying the tech brows, knowing what you are doing, and keeping the human in charge) over outright rejection. I am not alone, and Joplin tries to cater for both kinds of responsible users: Those of your type who outright reject it (they get the default setting), and those of my type who try to use it responsibly.
Rejecting the software for giving users the choice feels kind of overdoing it if you ask me.
-
@waldschnecke I am really with you on this one: That would be a very dumb move.
But I have to repeat myself, I use open source models. Even if their training data were scraped illegally and/or against my moral principles, still the company who did that has no material benefit from me using the OSS model they released. Maybe the OSS models are "appetizers" they release for free in hope to make me want to use their paid subscription models which are "even better" but that won't happen. Incidentally, there are European non-commercial efforts such as #OpenEuroLLM which train on legal training data only . Once their model will work sufficiently well, I will switch over.
On the other hand, I cannot afford to ignore and reject the technology outright because others in my business have started using it and I have the choice to use it as well or get left behind. In this scenario, I prefer advocating for responsible use (including not paying the tech brows, knowing what you are doing, and keeping the human in charge) over outright rejection. I am not alone, and Joplin tries to cater for both kinds of responsible users: Those of your type who outright reject it (they get the default setting), and those of my type who try to use it responsibly.
Rejecting the software for giving users the choice feels kind of overdoing it if you ask me.
-
@waldschnecke I am really with you on this one: That would be a very dumb move.
But I have to repeat myself, I use open source models. Even if their training data were scraped illegally and/or against my moral principles, still the company who did that has no material benefit from me using the OSS model they released. Maybe the OSS models are "appetizers" they release for free in hope to make me want to use their paid subscription models which are "even better" but that won't happen. Incidentally, there are European non-commercial efforts such as #OpenEuroLLM which train on legal training data only . Once their model will work sufficiently well, I will switch over.
On the other hand, I cannot afford to ignore and reject the technology outright because others in my business have started using it and I have the choice to use it as well or get left behind. In this scenario, I prefer advocating for responsible use (including not paying the tech brows, knowing what you are doing, and keeping the human in charge) over outright rejection. I am not alone, and Joplin tries to cater for both kinds of responsible users: Those of your type who outright reject it (they get the default setting), and those of my type who try to use it responsibly.
Rejecting the software for giving users the choice feels kind of overdoing it if you ask me.