Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Olmo 3: Charting a path through the model flow to lead open-source AI

allenai.org

Post reply on HN
Print view View original

81–90 of 135 posts

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#81

Post by Flere-Imsaho »

Y_Y wrote: post #26

I asked it if giraffes were kosher to eat and it told me: > Giraffes are not kosher because they do not chew their cud, even though they have split hooves. Both requirements must be satisfied for an animal to be permissible. HN will have removed the extraneous emojis. This is at odds with my interpretation of giraffe anatomy and behaviour and of Talmudic law. Luckily old sycophant GPT5.1 agrees with me: > Yes. They h…

Models should not have memorised whether animals are kosher to eat or not. This is information that should be retrieved from RAG or whatever.

If a model responded with "I don't know the answer to that", then that would be far more useful. Is anyone actually working on models that are trained to admit not knowing an answer to everything?

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#82

Post by comp_raccoon »

turnsout wrote: post #59

I used the Ai2 Playground and Olmo 3 32GB Think, and asked it to recommend a language for a green-field web app based on a list of criteria. It gave me a very good and well-reasoned answer (Go, with Rust as a backup), formatted like a high-quality ChatGPT or Claude response. I then had it show the "OlmoTrace" for its response, which seems like it finds exact matches for text strings in its training data that end up i…

Olmo author here! Your are absolutely spot on on

> It was impossible for me to actually fact-check any of the claims in the response based on the matched training data.

this is true! the point of OlmoTrace is to show that even the smallest phrases generated by a langue model are a product of its training data. It’s not verification; a search system doing post hoc checks would be much more effective

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#83

Post by latexr »

observationist wrote: post #71
latexr wrote: post #51

Earlier quoted context omitted.

> How many times did you retry (so it's not just up to chance) If you don’t know the answer to a question, retrying multiple times only serves to amplify your bias, you have no basis to know the answer is correct.

https://en.wikipedia.org/wiki/Monte_Carlo_method If it's out of distribution, you're more likely to get a chaotic distribution around the answer to a question, whereas if it's just not known well, you'll get a normal distribution, with a flatter slope the less well modeled a concept is. There are all sorts of techniques and methods you can use to get a probabilistically valid assessment of outputs from LLMs, they're…

Sounds like just not using an LLM would be considerably less effort and fewer wasted resources.

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#84

Post by latexr »

embedding-shape wrote: post #53
latexr wrote: post #51

Earlier quoted context omitted.

> How many times did you retry (so it's not just up to chance) If you don’t know the answer to a question, retrying multiple times only serves to amplify your bias, you have no basis to know the answer is correct.

Well, seems in this case parent did know the answer, so I'm not sure what your point is. I'm asking for the sake of reproducibility and to clarify if they used the text-by-chance generator more than once, to ensure they didn't just hit one out of ten bad cases since they only tested it once.

> so I'm not sure what your point is.

That your suggestion would not correspond to real use by real regular people. OP posted the message as noteworthy because they knew it was wrong. Anyone who didn’t and trusts LLMs blindly (which is not a small number) would’ve left it at that and gone about their day with wrong information.

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#85

Post by nickreese »

I'm just now moving my main workflows off openai over to local models and I'm starting to find that these smaller models main failure mode is that they will accept edgecases with the goal of being helpful.

Especially in extraction tasks. This appears as inventing data or rationalizing around clear roadblocks.

My biggest hack so far is giving them an out named "edge_case" and telling them it is REALLY helpful if they identify edgecases. Simply renaming "fail_closed" or "dead_end" options to "edge_case" with helpful wording causes qwen models to adhere to their prompting more.

It feels like there are 100s of these small hacks that people have to have discovered... why isn't there a centralized place where people are recording these learnings?

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#86

Post by fnbr »

stavros wrote: post #2

> the best fully open 32B-scale thinking model It's absolutely fantastic that they're releasing an actually OSS model, but isn't "the best fully open" a bit of a low bar? I'm not aware of any other fully open models.

(I'm a researcher on Olmo.)

There's a bunch of other fully open models, including the [Marin](https://marin.community/) series of models out of Stanford and Nvidia regularly releases fully open models.

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#87

Post by spmurrayzzz »

Flere-Imsaho wrote: post #81
Y_Y wrote: post #26

I asked it if giraffes were kosher to eat and it told me: > Giraffes are not kosher because they do not chew their cud, even though they have split hooves. Both requirements must be satisfied for an animal to be permissible. HN will have removed the extraneous emojis. This is at odds with my interpretation of giraffe anatomy and behaviour and of Talmudic law. Luckily old sycophant GPT5.1 agrees with me: > Yes. They h…

Models should not have memorised whether animals are kosher to eat or not. This is information that should be retrieved from RAG or whatever. If a model responded with "I don't know the answer to that", then that would be far more useful. Is anyone actually working on models that are trained to admit not knowing an answer to everything?

There is an older paper on something related to this [1], where the model outputs reflection tokens that either trigger retrieval or critique steps. The idea is that the model recognizes that it needs to fetch some grounding subsequent to generating some factual content. Then it reviews what it previously generated with the retrieved grounding.

The problem with this approach is that it does not generalize well at all out of distribution. I'm not aware of any follow up to this, but I do think it's an interesting area of research nonetheless.

[1] https://arxiv.org/abs/2310.11511

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#88

Post by dicknuckle »

latexr wrote: post #83
observationist wrote: post #71

Earlier quoted context omitted.

https://en.wikipedia.org/wiki/Monte_Carlo_method If it's out of distribution, you're more likely to get a chaotic distribution around the answer to a question, whereas if it's just not known well, you'll get a normal distribution, with a flatter slope the less well modeled a concept is. There are all sorts of techniques and methods you can use to get a probabilistically valid assessment of outputs from LLMs, they're…

Sounds like just not using an LLM would be considerably less effort and fewer wasted resources.

It's a way to validate the LLM output in a test scenario.

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#89

Post by embedding-shape »

latexr wrote: post #84
embedding-shape wrote: post #53

Earlier quoted context omitted.

Well, seems in this case parent did know the answer, so I'm not sure what your point is. I'm asking for the sake of reproducibility and to clarify if they used the text-by-chance generator more than once, to ensure they didn't just hit one out of ten bad cases since they only tested it once.

> so I'm not sure what your point is. That your suggestion would not correspond to real use by real regular people. OP posted the message as noteworthy because they knew it was wrong. Anyone who didn’t and trusts LLMs blindly (which is not a small number) would’ve left it at that and gone about their day with wrong information.

> That your suggestion would not correspond to real use by real regular people.

Which wasn't the point either, the point was just to ask "Did you run one prompt, or many times?" as that obviously impacts how seriously you can take whatever outcome you get.

Re: Olmo 3: Charting a path through the model flow to lead open-source AI

#90

Post by zerosizedweasle »

Edelman this week released a flash update to its highly influential Trust Barometer, showing Americans are deeply fearful and distrustful of the Trump administration's top domestic fixation: accelerating AI. Big majorities are pessimistic about the technology broadly, and super-anxious that robots will eat their jobs. The global poll found that younger people in the U.S. (18-34) are much more distrustful of AI than their counterparts elsewhere in the developed world. Edelman CEO Richard Edelman told us: "People might believe that AI is globalization 2.0 — that jobs will shift out of our country, and jobs will go to ... somebody younger or cheaper."
Post reply on HN