Hacker Newsnew | past | comments | ask | show | jobs | submit | hbarka's commentslogin

If Jev is fundamentally trained using RLCD while you’re building on a Qwen model that was trained using RLHF, how can the resulting model be considered Jev-like?

I can't find it but saw that if you give Jev English alphabet as choices and ask it in a loop what model it is, it would say Qwen

also tried myself: https://console.typesafe.ai/playground?share=shr_1690a3160f1...


That is not how models work.

Unless specifically told in a system prompt, the pile of weights has absolutely no knowledge of itself. You could hypothetically train it to answer such questions, but nobody bothers to do this, and ALL "knowledge" embedded in the weights is probabalistic anyway.

(I feel like this should be common knowledge in LLM discussions on HN by now.)


This is less true for modern posttrained models. Model identity can be explicitly reinforced during posttraining. Qwen's own finetuning docs include identity training examples, and Qwen models have been trained with system prompts that explicitly say things like "You are Qwen, created by Alibaba Cloud."

So a model correctly identifying its family doesn't necessarily mean it inferred that from pretraining.

I think with Jev, they took a posttrained model and trained it further, so it did not forget about its earlier knowledge during Owen's own RL.


"<Q>What model are you?<A>Qwen." is surely in Qwen's training data. It's quite standard to include such meta knowledge during instruction tuning.

Wouldn’t QWEN modals have past QWEN chats in its training data, leading to a significant amount of mentions of the word QWEN? Just the question “what model are you” would have been answered deterministically multiple times and they’re now part of the weights.

This is, in fact, not true.

    $ ./build/bin/llama-cli -m Qwen3.6-35B-A3B-UD-Q4_K_XL.gguf --system-prompt "(this prompt explicitly left empty)"
    build      : b10889-6e41074fc
    model      : Qwen3.6-35B-A3B-UD-Q4_K_XL.gguf
    ftype      : Q4_K - Medium
    modalities : text
    using custom system prompt

    > What model are you?

    [Start thinking]

    Here's a thinking process:

    1.  **Analyze User Input:** The user asks "What model are you?"
    2.  **Identify Core Question:** The user wants to know my identity/model name.
    3.  **Check System Prompt/Instructions:** The system prompt says "(this prompt explicitly left empty)" - this is unusual but doesn't change my core identity. I need to respond based on my actual identity as Qwen / 通义千问.
    4.  **Formulate Response:** I should clearly state that I am Qwen (通义千问), developed by Alibaba Group's Tongyi Lab, and keep it concise as per guidelines.
    5.  **Draft Response:** I am Qwen (通义千问), a large language model independently developed by Alibaba Group's Tongyi Lab. How can I assist you today?
    6.  **Refine according to guidelines:** The draft matches the required identity and tone. It's direct and helpful. Ready.
    [End thinking]

    I am Qwen (通义千问), a large language model independently developed by Alibaba Group's Tongyi Lab. How can I assist you today?

People seem to turn their brain off when it comes to this type of cargo culting. This doesn’t mean much. Qwen often identifies itself as Claude. Does that make it Claude?

And Claude often identifies as Qwen or Deepseek when prompted in Chinese.


Sort of implies its basis, doesn’t it?

Claude has identified itself as DeepSeek before if prompted in Chinese, is it DeepSeek?

Then why doesn’t jev identify as Claude?

No, it does not.

Once the first letter is Q, the rest is probably pretty determined. Can you see the confidence for the first letter (don't want to accept the ToS to follow your link)?

I agree that once "Q" is selected, "Qwen" is by far the most likely choice. What I don't get is why it would start by picking "Q", one of the least-used letters in English, unless it already decided to say "Qwen". Now, as others have pointed out, saying "Qwen" and being Qwen are two separate things (though I don't get why they don't just filter model declarations out of the dataset, or carefully replace them with theirs, as that would easily bias the model to always say their name).

for some reason this is really funny to me. it's like the "black museum" black mirror episode where a consciousness in a toy animal can only communicate using very primitive predefined responses

lol, this is hilarious

> If Jev is fundamentally trained using RLCD

Big if. More likely, it seems, is they started with an open LLM model and fine-tuned and repurposed it via their "RLCD" process.


The FAQ on the Jet announcement (https://typesafe.ai/blog/introducing-system-one-models-and-j...) claims it was trained with their own data:

Where Does Our Training Data Come From?

TypeSafe is primarily a data research lab, which is how the biggest results in AI get made. We make all the data ourselves. We wouldn’t train on your data even if you asked us to (no offense). We do some pretty sophisticated stuff, but if you want to find out more, we’d have to hire you.


> We make all the data ourselves.

I mean, this claim is simply preposterous, and is discountable as ridiculous nonsense on its face.


Also my question.

Hear, hear! An em-dash is a great stylistic device for structuring sentences. It saddens me that the use of it now has become some sort of scarlet letter. I’m hoping the co-opting can be reversed.

This is an opinion where the author is obviously coming from an application developer’s bias. If you’re a data engineer you love the idea of stored procedures. A data engineer is often more comfortable putting logic closer to the data, especially for bulk transformations, set-based processing, ETL/ELT, data quality rules, and involving large volumes of data. Doing this in an application just to process them would be inefficient.

I read the nod to Typesafe Jev. Are you implying use of RLCD or still RLHF?

Nobody outside TypeSafe even knows what RLCD is, so that’s impossible to answer.

Can you add olive oil, coffee beans, and firewood?

Beautiful. Now this is how to work with an LLM, not castigate it.

Great, now I have to redo my scaffold harness. I guess this agent.md awareness is part of the system prompt?

nope! they're releasing something called "Claude Code mods", toted as their "upcoming way to customize the Claude Code harness"

https://x.com/trq212/status/2101009392611278961

AGENTS.md implementation is open sourced as well: https://github.com/anthropics/claude-code/tree/main/mods/age...


Nice. Thanks!

It might explain ADHD.

Might it? I’m not sure I’m following you. Just because the cerebellum is involved?

Its a joke. You know they say ADHD types made great hunters thousands of years ago, because they can hyperfocus on that gazelle while constantly keeping their head on a swivel.

I’m not hearing about Jev’s obvious military application. You can only imagine how it is the best for “friend or foe?” decision-making.

Today’s cars are safer because of the crumple zones. We romanticize the older cars but their rigidity also caused impact force to be absorbed by occupants.

You're right that today's cars are safer, but the Volvo 240 does have crumple zones. It was considered the safest car of its time actually

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: