Par pitié, arrêtons de confondre Arabe et musulmans. Je ne sais même pas si la majorité des Arabes vivant en France sont religieux.
Tu as des subsides pour la LAMAL. Pour vivre à 1300 tu es obligé de les utiliser.
Je confirme 1300 c'est obligatoirement une colocation ou en couple et tu ne vas jamais au restaurant.
Mais à 4000 tu n'as plus toutes ces contraintes. Quand j'ai touché 3600 CHF pour la première fois j'avais vraiment l'impression d'être riche. Je pouvais aller au restaurant tous les jours, partir à l'autre bout de l'Europe sur un coup de tête etc...
4000 balles et c’est la galère avec ça
Faut pas abuser, j'ai vécu il y a 3 ans avec 1300 balles par mois à Lausanne et c'était pas si juste que ça.
4000 tu as une assez bonne qualité de vie.
J'ajouterai que ça dépend des cantons, je connais des gens dans le Valais qui ne gagnent que 2800 CHF par mois en travaillant a plein temps (41h semaine).
Thank you, handling these subjects is already annoying at family dinners. I can't imagine handling this on an online forum.
Tu trouves la forme d'exploitation faite par une entreprise aussi mauvaise que celle faite par ces trafiquants de drogues?
It's great for making images to associate with language learning flashcards.
L'imminence des Jeux a conduit la SNCF à revoir ses règles de sécurité, notamment en cas d'intrusion sur les voies. Désormais les trains ne s'arrêtent plus, mais circulent à "marche prudente".
They've got thunderbird which is as far as I know the only serious alternative to outlook.
Ancien ministre de François Hollande, l’eurodéputé refuse le «pacte législatif» proposé par Laurent Wauquiez, qu’il juge «très à droite» et prisonnier du «soutien tacite» du RN.
Pas forcément d'accord pour les 1,4 milliards pour la Seine. A ma connaissance il n'y a pas eu de mesures d'urgence et temporaires. Toute l'infrastructure va rester en place.
Après c'est vrai que ça coûte 140 euros par habitant d'Ile de France, ce qui n'est pas négligeable. Mais de la a appeler ça un caprice.
Ton article parle d'autre chose. Il parle de l'année 2023 par rapport à l'année 2022. La on compare le premier semestre 2024 au premier semestre 2023.
J'ai répondu par rapport au secteur de l'industrie dans un autre commentaire, pour ce secteur là ça semble être une bonne chose. Pour les autres secteurs en forte baisse, je n'ai pas fait d'analyse, mais il n'y a pas de raison que cette baisse soit liée à des choses négatives. Par exemple une grosse partie des réductions sont dues à un hiver clément.
Le seul point noir, c'est les transports. On sait très bien comment réduire les émissions, vu que la demande est élastique par rapport au prix. Mais plus personne n'osera mettre en place une taxe CO2 (merci les gilets jaunes). Le RN promet même de retirer les taxes sur l'essence (quitte même à ne pas respecter les règles de l'UE), histoire qu'on soit encore plus dépendant des dictatures.
Peut-être pour la baisse de 5% des émissions de l'industrie. Mais pas pour les autres secteurs.
C'est aussi possible qu'il n'y ait pas eu de baisse dans l'industrie. Ton article parle de faillite d'entreprises, ce qui ne veut pas dire une cessation d'activité économique, juste que le propriétaire a perdu son argent.
Le premier exemple de ton article, je la connais bien. C'est Ascometal, j'ai suivi la saga avec Marsactu. L'usine continuera de produire après sa reprise avec juste l'arrêt de l'activité du fil. L'entreprise racheteuse va aussi investir pour ajouter un four électrique pour réduire l'impact carbone de l'entreprise.
Ce qu'il faudrait regarder c'est la production industriel, qui est assez stable :
Le ministre de la Transition écologique a annoncé ce mercredi 26 juin que les émissions de gaz à effet de serre ont continué de diminuer en ce début d’année 2024. Une baisse à la fois structurelle, mais aussi conjoncturelle.
Sauf que les gens qui ne font pas d'études ont une espérance de vie plus faible. C'est normal qu'il puissent prendre leur retraite plus tôt.
C'est ce qui me gênait le plus dans la réforme de Macron. L'augmentation de l'âge minimum pour la retraite touche plus les précaires qu'une augmentation des années de cotisations.
En tant qu'ingénieur, j'ai un travail bien moins pénible qu'un ouvrier, c'est normal que je prenne une retraite plus tardive.
including the Slogan: 'From the river to the sea – Palestine shall be free'
As it should, this phrase and it's Israeli counterpart "between the sea and the Jordan there will be only Israeli sovereignty" are often accompanied by calls for mass deportation at best and genocide at worst.
These sentences are not bad on their own, but the parties from which they originate (Hamas and Likud) have transparent desires for war crimes and genocide.
“They had Polish refugee status certificates for the adults and Polish ID cards for the children. These people were taken to the police station for further investigation,” said von Vegesack. He added that, after their legal situation was confirmed, the family had to be sent back to Poland.
According to the spokesman, German officials informed the Polish border guard about the situation through the Polish-German Centre for Cooperation of Border, Police and Customs Services in Świecko.
“Since there was no reaction from the Polish side for several hours, the officers decided to take the family to the Polish-German border…in order to release them to Poland,” said von Vegesack.
The article above should be more "polish refugees left alone in Poland by German police" rather than the current one which suggests that Germany is getting rid of its migrants illegally.
Huh, your comment made look if there were personalized recommendation algorithms running on lemmy. From what I found, it appears that Lemmy does not use personalized recommendation algorithms : https://join-lemmy.org/docs/contributors/07-ranking-algo.html
The specific function used for ranking is here : https://github.com/LemmyNet/lemmy/blob/4ba6221e04ab3e186669aeaa890d23b1e3f3d1a9/crates/db_schema/replaceable_schema/utils.sql#L18
I'm wondering how hard it would be to adapt the code to customize the score for every user, instead of it being global.
I makes it look like the advancement here is finding methods to efficiently use sets of graphs which are an order of magnitude larger than prior methods could use for training? They also seem to have used more sets of graphs than prior models across a wider set of domains. Am I reading this correctly?
I find it challenging to gauge the paper's impact fully, as this isn't my area of expertise. However, the ability to use diverse graphs in a single model surprised me and seemed worth sharing.
Separately, Jong has also alleged that Apple subjected her to a hostile work environment after a senior member of her team, Blaine Weilert, sexually harassed her. After she complained, Apple investigated and Weilert reportedly admitted to touching her "in a sexually suggestive manner without her consent," the complaint said. Apple then disciplined Weilert but ultimately would not allow Jong to escape the hostile work environment, requiring that she work with Weilert on different projects. Apple later promoted Weilert.
As a result of Weilert's promotion, the complaint said that Apple placed Weilert in a desk "sitting adjacent" to Jong's in Apple’s offices. Following a request to move her desk, a manager allegedly "questioned" Jong's "willingness to perform her job and collaborate" with Weilert, advising that she be “professional, respectful, and collaborative,” rather than honoring her request for a non-hostile workplace.
...
French here we use both the middle and the left. It depends on the group of friends.
Now instead of just querying the goddamn database, a one line fucking SQL statement, I have to deal with the user team
Exactly, you understand very well the purpose of microservices. You can submit a patch if you need that feature now.
Funnily enough I'm the technical lead of the team that handles the user service in an insurance company.
Due to direct access to our data without consulting us, we're getting legal issues as people were using addresses to guess where people lived instead of using our endpoints.
I guess some people really hate the validation that service layers have.
You are in a bubble. A neo nazi march was banned two weeks ago in France before being allowed again by the judicial system. The exact same scenario has been repeating for pro-palestine protests.
At least in France, the scenario seems to be that the government wants to ban any controversial march and is being kept under control by the justice system.
I also have a similar experience, I was mugged at knife point and spit on by two adolescents. After that I was jumpy around groups of teens.
That said , I do not think my fear of teens was rational, neither was it healthy. Only a small minority of teens will mug people. Fearing a whole group for the actions of the few is in human nature, but it is something we must fight against.
I mean what is the end goal if women are in fear of men ? You can probably reduce violent crime even more, but it remains a rare event. Only 31 out of 1000 people were victims of a violent crime in the UK in 2010. If that doesn't work, what remains? Sex segregation ?
cross-posted from: https://lemmy.one/post/13942290
> Abstract: We present Scallop, a language which combines the benefits of deep learning and logical reasoning. Scallop enables users to write a wide range of neurosymbolic applications and train them in a data- and compute-efficient manner. It achieves these goals through three key features: 1) a flexible symbolic representation that is based on the relational data model; 2) a declarative logic programming language that is based on Datalog and supports recursion, aggregation, and negation; and 3) a framework for automatic and efficient differentiable reasoning that is based on the theory of provenance semirings. We evaluate Scallop on a suite of eight neurosymbolic applications from the literature. Our evaluation demonstrates that Scallop is capable of expressing algorithmic reasoning in diverse and challenging AI tasks, provides a succinct interface for machine learning programmers to integrate logical domain knowledge, and yields solutions that are comparable or superior to state-of-the-art models in terms of accuracy. Furthermore, Scallop's solutions outperform these models in aspects such as runtime and data efficiency, interpretability, and generalizability.
Midjourney claims the alleged activity caused a 24-hour service outage.
cross-posted from: https://lemmy.ml/post/13088944
Insight into the hidden ecosystem of autonomous chatbots and data scrapers crawling across the web
abstract : > How do sequence models represent their decision-making process? Prior work suggests that Othello-playing neural network learned nonlinear models of the board state (Li et al., 2023). In this work, we provide evidence of a closely related linear representation of the board. In particular, we show that probing for "my colour" vs. "opponent's colour" may be a simple yet powerful way to interpret the model's internal state. This precise understanding of the internal representations allows us to control the model's behaviour with simple vector arithmetic. Linear representations enable significant interpretability progress, which we demonstrate with further exploration of how the world model is computed.
Paper here : https://arxiv.org/pdf/2312.00752.pdf
Abstract : > Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module. Many subquadratic-time architectures such as linear attention, gated convolution and recurrent models, and structured state space models (SSMs) have been developed to address Transformers’ computational inefficiency on long sequences, but they have not performed as well as attention on important modalities such as language. We identify that a key weakness of such models is their inability to perform content-based reasoning, and make several improvements. First, simply letting the SSM parameters be functions of the input addresses their weakness with discrete modalities, allowing the model to selectively propagate or forget information along the sequence length dimension depending on the current token. Second, even though this change prevents the use of efficient convolutions, we design a hardware-aware parallel algorithm in recurrent mode. We integrate these selective SSMs into a simplified end-to-end neural network architecture without attention or even MLP blocks (Mamba). Mamba enjoys fast inference (5× higher throughput than Transformers) and linear scaling in sequence length, and its performance improves on real data up to million-length sequences. As a general sequence model backbone, Mamba achieves state-of-the-art performance across several modalities such as language, audio, and genomics. On language modeling, our Mamba-3B model outperforms Transformers of the same size and matches Transformers twice its size, both in pretraining and downstream evaluation.
Update, July 13: AI Assistant is available in pre-release versions, but is not bundled with the stable releases of JetBrains IDEs v.2023.2. It can be installed as a separate plugin available for versi
Was looking at EAP6 release notes and was pleasantly surprised to see this there.
I'm quite happy that intellij provides on premise solutions, it gives a small chance of this coming to my job one day. I believe this will be quite useful for repetitive code and certain types of tests.
cross-posted from: https://kbin.social/m/machinelearning/t/98088
> Abstract: > > > > > > > Work on scaling laws has found that large language models (LMs) show predictable improvements to overall loss with increased scale (model size, training data, and compute). Here, we present evidence for the claim that LMs may show inverse scaling, or worse task performance with increased scale, e.g., due to flaws in the training objective and data. We present empirical evidence of inverse scaling on 11 datasets collected by running a public contest, the Inverse Scaling Prize, with a substantial prize pool. Through analysis of the datasets, along with other examples found in the literature, we identify four potential causes of inverse scaling: (i) preference to repeat memorized sequences over following in-context instructions, (ii) imitation of undesirable patterns in the training data, (iii) tasks containing an easy distractor task which LMs could focus on, rather than the harder real task, and (iv) correct but misleading few-shot demonstrations of the task. We release the winning datasets at https://inversescaling.com/data to allow for further investigation of inverse scaling. Our tasks have helped drive the discovery of U-shaped and inverted-U scaling trends, where an initial trend reverses, suggesting that scaling trends are less reliable at predicting the behavior of larger-scale models than previously understood. Overall, our results suggest that there are tasks for which increased model scale alone may not lead to progress, and that more careful thought needs to go into the data and objectives for training language models. > > > >
How many humans does it take to make tech seem human? Millions.
Hyena Hierarchy seems to aim to be a drop-in replacement for attention : https://arxiv.org/pdf/2302.10866.pdf
It looks good on paper, but I haven't been able to find anybody using it in a model. Does anyone have an example of a code or implementation ? Is there really a big improvement on long context lengths ?