Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models · HackerPapers discussions

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Sun, 15 Jun 2025 14:32:23 GMT

Your entire edit essentially walks back your earlier strong claims. None of your current points actually support your position. 1. No, it doesn't. That's a ridiculous claim. Are you seriously suggesting that statistics require reasoning? 2. If you map that language to tokens, it's obvious the model will follow that mapping. etc. Here are papers showing that these models can't reason: https://arxiv.org/abs/2311.00871 https://arxiv.org/abs/2309.13638 https://arxiv.org/abs/2311.09247 https://arxiv.org/abs/2305.18654 https://arxiv.org/abs/2309.01809 You're mistaking pattern matching and the…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Mon, 19 Aug 2024 12:26:29 GMT

> You can get them to solve unseen problems just fine Prove that the problem wasn't seen by them in other form. > Specify a grammar in BNF notation and tell it to generate or parse sentences for you. You can produce a more than random enough grammar that it it can't have derived the parsing of it from past text, but necessarily reasons about BNF notation sufficiently well to be able to use it to deduce the grammar, and use that to parse subsequent sentences. You can have it analyse them and tag them according to the grammar to. And generate sentences. Oh, come on. It's like rewriting the same…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Mon, 19 Aug 2024 12:17:28 GMT

> "generalize to its dataset" is a contradiction Not "to" but over, example the same code written in one language over the other language. > if you think being able to generalize in ways similar to the whole of the internet does not give your meaningful abilities to reason, I'm not sure what I can tell you If after reading papers below that show empirically that they can't reason, you will still think they can reason, then I don't know what I can tell you. https://arxiv.org/abs/2311.00871 https://arxiv.org/abs/2309.13638 https://arxiv.org/abs/2311.09247 https://arxiv.org/abs/2305.18654…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Thu, 07 Mar 2024 12:02:24 GMT

There is evidence that they actually can't reason: https://arxiv.org/abs/2311.00871 https://arxiv.org/abs/2309.13638 https://arxiv.org/abs/2311.09247 https://arxiv.org/abs/2305.18654 https://arxiv.org/abs/2309.01809

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Mon, 22 Jan 2024 18:41:00 GMT

For every N papers suggesting that LLM can "understand" you have X papers that suggests they can't. https://arxiv.org/abs/2311.00871 https://arxiv.org/abs/2309.13638 https://arxiv.org/abs/2311.09247 https://arxiv.org/abs/2305.18654 https://arxiv.org/abs/2309.01809

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Sun, 07 Jan 2024 21:20:19 GMT

There is also a lot of evidence lately that they do not generalize. https://arxiv.org/abs/2311.00871 https://arxiv.org/abs/2309.13638 https://arxiv.org/abs/2311.09247 https://arxiv.org/abs/2305.18654

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Wed, 27 Dec 2023 17:32:22 GMT

A few more interesting papers not mentioned in the article: "Faith and Fate: Limits of Transformers on Compositionality" https://arxiv.org/abs/2305.18654 "Comparing Humans, GPT-4, and GPT-4V On Abstraction and Reasoning Tasks": https://arxiv.org/abs/2311.09247 "Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solve" https://arxiv.org/abs/2309.13638 "Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models" https://arxiv.org/abs/2311.00871

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Thu, 14 Dec 2023 22:18:57 GMT

Weather and stock market are both chaotic systems. Increasing evidence suggests that AGI will not be attainable solely using LLMs/transformers/current architecture, as LLMs can't extrapolate beyond the patterns in their training data (according to a paper from DeepMind last month): "Together our results highlight that the impressive ICL abilities of high-capacity sequence models may be more closely tied to the coverage of their pretraining data mixtures than inductive biases that create fundamental generalization capabilities."[1] 1. https://arxiv.org/abs/2311.00871

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by lossolo

Wed, 06 Dec 2023 19:10:03 GMT

> we've been doing this ai stuff since you (other AI companies) were little babies Actually, they kind of did. What's interesting is that they still only match GPT-4's version but don't propose any architectural breakthroughs. From an architectural standpoint, not much has changed since 2017. The 'breakthroughs', in terms of moving from GPT to GPT-4, included: adding more parameters (GPT-2/3/4), fine-tuning base models following instructions (RLHF), which is essentially structured training (GPT-3.5), and multi-modality, which involves using embeddings from different sources in the same latent…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by optimalsolver

Mon, 20 Nov 2023 16:40:11 GMT

That's assuming that large language models or transformers are the key to AGI, which is looking increasingly unlikely: https://arxiv.org/abs/2311.00871

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by mempko

Thu, 09 Nov 2023 20:05:50 GMT

Google Deep Mind just demonstrated that transformers don't generalize well outside their training set. https://arxiv.org/abs/2311.00871 The question I have is, what if these agents are really good at doing things they learned about, but not in new things? How will that impact people's actions? Will we end up with a static world of people doing only things the AI can do? A kind of 1999 Matrix stasis world of 'the best we can do now'? What will happen to what Arthur Kessler called 'blue' thoughts in such an AI world? Because IMO, the purpose of computers is to make models and see what happens.…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by gardenfelder

Tue, 07 Nov 2023 18:17:18 GMT

The paper https://arxiv.org/abs/2311.00871

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — HN submission

Tue, 07 Nov 2023 16:51:41 GMT

Pretraining data enables narrow selection capabilities in transformer models

65 points; 109 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — HN submission

Mon, 06 Nov 2023 12:05:48 GMT

Evidence that transformers cannot generalize beyond their training data

3 points; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — mentioned by mark_l_watson

Mon, 06 Nov 2023 11:13:02 GMT

Painfully, years ago I wrote a book with Web 3.0 in the title. For me that meant semantic web and linked data. I am enjoying reading this entire thread. For AI skeptics: I am not trying to talk you into anything you don’t believe, but just as an experiment, whenever you think about LLMs, appreciate how well they generalize past things they have been trained on. Peter Norvig and a friend of his recently write an article arguing that LLMs probably meet our traditional criteria for AGI. EDIT: I am aware of the recent paper https://arxiv.org/abs/2311.00871 that argues that LLMs have limited…

Paper and discussion history. A mention is not an endorsement.

Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models — HN submission

Mon, 06 Nov 2023 00:29:37 GMT

Pretraining Data Mixtures Enable Narrow Model Selection Capabil. In Transformers

5 points; 2 discussion replies on HN. Paper and source history. A submission is not an endorsement.