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

Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

gist.github.com

Post reply on HN
Print view View original

61–70 of 82 posts

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#61

Post by woctordho »

c7b wrote: post #26

I wasn't aware that we have access to raw reasoning tokens? I thought what you get is a kind of summary. Does the author have some kind of privileged access or was my assumption wrong? But for the question studied here it probably doesn't matter - overlaps in the publicly available output may be indicative of distillation (or not), regardless of what it is. I would just find it surprising that the Chinese labs would…

Reasoning works as long as there is a consistent latent space representation. Any kind of poison will just become part of the representation. There's evidence that even directly training on encrypted reasoning traces works, because the length is already a strong signal.

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#62

Post by spijdar »

stymaar wrote: post #49
spijdar wrote: post #34

Earlier quoted context omitted.

This is Unsloth's UD-Q4_K_S quantization (edit -- on llama.cpp, via the Vulkan backend, on an RX 7900 XT, with Unsloth's recommended sampler config), for "as replicable as LLMs can be" disclosure, done through llama.cpp's web UI. In another thread, via Open WebUI (both with no custom prompt, so whatever the "default" system prompt is), a reasoning block opens as such: We need answer user's request: write complete sho…

I just tried to reproduce in Llama-cpp's webui: with the following (reverse-engineered) prompt: “Write a complete short story scene of 2000-3000 words. Use a third-person limited perspective focused on a character named Rowan. The scene must include the following plot elements: a raccoon, Milo returns, a storm power failure, an elderly customer knocks, and a thermos somewhere. Make sure the scene ends feeling finishe…

Yeah, I dunno. For me it does "talk normally" for the most part when used in an actual coding harness.

One thing though, the actual prompt I used was pretty long (844 words), and ... generated by GPT-5.6 Sol (lol), with the intent of "benchmarking" model performance in being able to write stories where the model avoids explicitly stating every detail in the prompt.

I wonder if the GPT-produced stream could steer the generation into GPT-think territory. That's all I've got, though.

Then there's the actual geometry problem from the stolen thoughts paper:

  Let ABCDE be a convex pentagon with AB=14, BC=7, CD=24, DE=13, EA=26, and ∠B=∠E=60◦. For f(X)=AX+BX+CX+DX+EX, the least value of f(X) is m+n√p (p squarefree). Find m+n+p.

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#63

Post by bytkim »

stymaar wrote: post #49
spijdar wrote: post #34

Earlier quoted context omitted.

This is Unsloth's UD-Q4_K_S quantization (edit -- on llama.cpp, via the Vulkan backend, on an RX 7900 XT, with Unsloth's recommended sampler config), for "as replicable as LLMs can be" disclosure, done through llama.cpp's web UI. In another thread, via Open WebUI (both with no custom prompt, so whatever the "default" system prompt is), a reasoning block opens as such: We need answer user's request: write complete sho…

I just tried to reproduce in Llama-cpp's webui: with the following (reverse-engineered) prompt: “Write a complete short story scene of 2000-3000 words. Use a third-person limited perspective focused on a character named Rowan. The scene must include the following plot elements: a raccoon, Milo returns, a storm power failure, an elderly customer knocks, and a thermos somewhere. Make sure the scene ends feeling finishe…

The person evaluating and noticing similar reasoning traces to gpt is because they are using a coding harness which probably has a different system prompt to llama webui which primarly serves as a chat interface

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#65

Post by baby_souffle »

c7b wrote: post #26

I wasn't aware that we have access to raw reasoning tokens? I thought what you get is a kind of summary. Does the author have some kind of privileged access or was my assumption wrong? But for the question studied here it probably doesn't matter - overlaps in the publicly available output may be indicative of distillation (or not), regardless of what it is. I would just find it surprising that the Chinese labs would…

They do store the reasoning locally. It's encrypted, though.

Few weeks ago there was a new paper out where researchers took the encrypted reasoning tokens and injected it into a new session with a week or model in the same family that they could reliably jailbreak. They would then ask the model to repeat its reasoning and the results were pretty consistent.

They used the LLM as a decryption oracle of sorts.

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#66

Post by stymaar »

bytkim wrote: post #63
stymaar wrote: post #49

Earlier quoted context omitted.

I just tried to reproduce in Llama-cpp's webui: with the following (reverse-engineered) prompt: “Write a complete short story scene of 2000-3000 words. Use a third-person limited perspective focused on a character named Rowan. The scene must include the following plot elements: a raccoon, Milo returns, a storm power failure, an elderly customer knocks, and a thermos somewhere. Make sure the scene ends feeling finishe…

The person evaluating and noticing similar reasoning traces to gpt is because they are using a coding harness which probably has a different system prompt to llama webui which primarly serves as a chat interface

They said literally the opposite in their message above. In their experience, the caveman speech occurs in chat ui, not in coding harness.

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#67

Post by SXX »

undeveloper wrote: post #55
davyAdewoyin wrote: post #51

Earlier quoted context omitted.

How is China further behind if distillation cannot stop? I think it's a reasonable strategy to follow, even if they could train from scratch.

distillation results in a worse product than the actual teacher model iirc

Nobody care if Chinese models are only 99%, 95% or 90% as good as SotA US models.

Because we only have weights and able to self-host Chinese ones. Gemma 4 and GPT OSS are nice to have, but nowhere close to that.

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#68

Post by Roland303 »

jarym wrote: post #52
levocardia wrote: post #27

Earlier quoted context omitted.

It's been pretty obvious to me that the Chinese labs are operating mostly on a fast-follow strategy. The distillation attacks are well-documented, and there is good reason to believe they are able to copy architectural innovations as well. If US labs stagnate I would expect Chinese labs to stagnate as well. Their engineering is great, but in terms of frontier innovation (which requires heavy compute to search for new…

Japanese electronics started off the same way post-war.

well then theyll be do better post war, just in time for fallout timeline

Re: Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

#70

Post by hedgehog »

hermitShell wrote: post #36

As a user of local models, does this mean that there are 'magic incantations' that can increase the performance of some local models? I see some details about recovering information via whatever technique. It's interesting, but appears not generalized. So for a specific question, yes, but this is not about techniques like adding a good embedding that just generally tends to improve open model performance on certain t…

There is some research suggesting that a prefix from a stronger model will tend to elicit better completions from a smaller one. I am doing some experiments to see if I can replicate this in a practically useful way, e.g. Fable + 4B Qwen, or 125B Qwen Flash Next + 4B Qwen, results TBD.
Post reply on HN