Thinking Like Transformers · HackerPapers discussions

Thinking Like Transformers — mentioned by syllogism

Tue, 03 Nov 2026 13:23:49 GMT

Here's the work that shows how rule-based logic can be compiled into transformer weights: * Thinking Like Transformers” by Weiss, Goldberg & Yahav (ICML 2021).: https://arxiv.org/abs/2106.06981 * Tracr: Compiled Transformers as a Laboratory for Interpretability https://arxiv.org/abs/2301.05062 (code at https://github.com/deepmind/tracr) So it's not really an "LLMs can modify weights" thing. Rather it's "LLMs can write programs that can write or modify weights". Indeed LLMs can write many programs, it's impressive.

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

Thinking Like Transformers — mentioned by andy12_

Mon, 16 Feb 2026 10:35:42 GMT

You do understand that the mechanism through which an auto-regressive transformer works (predicting one token at a time) is completely unrelated to how a model with that architecture behaves or how it's trained, right? You can have both: - An LLM that works through completely different mechanisms, like predicting masked words, predicting the previous word, or predicting several words at a time. - A normal traditional program, like a calculator, encoded as an autoregressive transformer that calculates its output one word at a time (compiled neural networks) [1][2] So saying "it predicts the…

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

Thinking Like Transformers — mentioned by leogao

Sat, 22 Nov 2025 18:55:42 GMT

There's been some work (e.g RASP - https://arxiv.org/abs/2106.06981) on taking logical computations and compiling them into transformer weights.

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

Thinking Like Transformers — mentioned by astrange

Fri, 21 Nov 2025 05:18:54 GMT

The difference is there are exponentially more states than an n-gram model. It's really not the same thing at all. An LLM can perform nearly arbitrary computation inside its fixed-size memory. https://arxiv.org/abs/2106.06981 (An LLM with tool use isn't a Markov process at all of course.)

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

Thinking Like Transformers — mentioned by nikvaes

Mon, 08 Apr 2024 15:37:23 GMT

After trying to understand and implement some algorithms in RASP [1, 2], my take-way was that certain functions need a certain amount of transformer layers to operate. Following this logic, it should become apparent that the functions learned by transformers can be spread over multiple heads. Repeating these functions might be very valuable for understanding and solving a problem, but current inference does not allow (a set of subsequent) heads to be repeated. This paper indeed seems a promising direction. [1] https://arxiv.org/pdf/2106.06981.pdf [2] https://www.youtube.com/watch?v=t5LjgczaS80

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

Thinking Like Transformers — mentioned by shawntan

Fri, 22 Sep 2023 22:45:08 GMT

Something close exists: RASP https://arxiv.org/abs/2106.06981 python implementation: https://srush.github.io/raspy/

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

Thinking Like Transformers — mentioned by teddykoker

Fri, 22 Sep 2023 17:43:10 GMT

A related line of work is "Thinking Like Transformers" [1]. They introduce a primitive programming language, RASP, which is composed of operations capable of being modeled with transformer components, and demonstrate how different programs can be written with it, e.g. histograms, sorting. Sasha Rush and Gail Weiss have an excellent blog post on it as well [2]. Follow on work actually demonstrated how RASP-like programs could actually be compiled into model weights without training [3]. [1] https://arxiv.org/abs/2106.06981 [2] https://srush.github.io/raspy/ [3] https://arxiv.org/abs/2301.05062

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

Thinking Like Transformers — mentioned by jbay808

Thu, 24 Aug 2023 18:39:22 GMT

It is only difficult for a LLM to sort a list of numbers if the list is longer than half of the context window. (Source: I tested this myself[1]). The sorts are not error-free every time, but with sufficient training they become error-free the vast majority of the time, even for long lists. This is not especially surprising because transformers are capable of directly representing sorting programs.[2] [1] https://jbconsulting.substack.com/p/its-not-just-statistics-... [2] https://arxiv.org/abs/2106.06981

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

Thinking Like Transformers — mentioned by YeGoblynQueenne

Sat, 24 Jun 2023 18:24:31 GMT

>> Transformers are RASP programs, which includes sorting programs. See the Weiss paper (https://arxiv.org/pdf/2106.06981.pdf). That's one preprint on arxiv, that makes a wild claim about a new concept that they acronymise as "RASP". It's not any kind of established terminology, nor is it anything but a claim. What is certainly established is that a function, and an algorithm, are different objects. To clarify, a function is a mapping between the elements of two sets, whereas an algorithm is a sequence of operations that calculates the result of a function and is guaranteed to terminate.…

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

Thinking Like Transformers — mentioned by jbay808

Sat, 24 Jun 2023 15:52:43 GMT

> it sure has not learned any "algorithm". At best it has learned to approximate a sorting algorithm. That's what statistical machine learning models do, they are function approximators; not program learners. Transformers are RASP programs, which includes sorting programs. See the Weiss paper (https://arxiv.org/pdf/2106.06981.pdf). > Also, Machine Learning 101: you test your models on a test set that is disjoint to the training set. To clarify, we do this not because it's in the book and that's the rules, but because, by testing the model on held-out data, we can predict the error the model…

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

Thinking Like Transformers — HN submission

Tue, 13 Jun 2023 15:40:13 GMT

Thinking Like Transformers (2021) [pdf]

112 points; 20 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Thinking Like Transformers — mentioned by jbay808

Wed, 24 May 2023 23:35:53 GMT

Author here. Happy to see this discussion. Absolutely, GPT-4 sometimes has trouble reasoning and doesn't reason perfectly. I'm impressed by its successes, but I agree it's not at the human level yet, and I would not make the claim that it is. Counting is a task that transformers can do, per Weiss.[1] But it's not surprising that transformer networks in general have trouble counting characters -- the tokenizer replaces common sub-strings, so the number of characters will not in general be the number of tokens. The network might have little way of even knowing how many characters are in a given…

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

Thinking Like Transformers — mentioned by attemptone

Tue, 18 Oct 2022 11:38:12 GMT

https://arxiv.org/abs/2106.06981 - Thinking like Transformes The paper suggests that transformer work on a set of select, aggregate and element-wise operations. Which seems pretty close to the SQL statements i write from day to day.

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

Thinking Like Transformers — mentioned by lalaithion

Mon, 20 Dec 2021 17:43:33 GMT

Not sure if this is a good introduction, but a good second paper to read is https://arxiv.org/abs/2106.06981 You can think of finite state machines as being two functions: f(input, state) = output, and g(input, state) = next_state. (Traditional FSMs have 3 ‘output’ states, basically ‘terminated - success’, ‘terminated - failure’, and ‘still working’, but in theory it makes sense to fully generalize it). If you think about plain neural networks as approximating arbitrary functions f(input) = output, then recurrent neural networks are “continuous state machines”, where you have the same two…

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

Thinking Like Transformers — HN submission

Wed, 16 Jun 2021 12:59:19 GMT

Thinking Like Transformers

161 points; 42 discussion replies on HN. Paper and source history. A submission is not an endorsement.