Hacker News Reader

Comment by skohan | Hacker News Reader

I've had the exact opposite experience. I've been using 3.8 for my daily driver since last week, and I've gradually been giving it more and more complex tasks as it continues to deliver high quality results. Now I am basically handing off large complex features, and 3.8 is doing the planning, task breakdown, implementation and review with just a few notes from my side.

The tradeoff is time (especially on RDMA4 hardware) - it does take a long time and spend a lot of tokens to get to the result, but I've found I can trust the results enough that I can queue a lot of work, essentially have it running all the time and achieve a decent velocity.

It's the first small local model I've felt like I can do real work with.

View on HN

I can't get 3.8 to exit thinking loops. It will just think and think and think on the most trivial topics. I wanted it to port a speed test powershell script to c#. Claude opus 5 completes it under 60 seconds. I let 3.8 churn about 6 different times for 30+ minutes and it never wrote a single line of code to disk. It wrote lots of lines in thinking.

unsloth/Qwen3.8-27B-GGUF UD-Q3_K_XL DSH (pi)

Any tips?

set thinking to minimal and use these jinja templates: https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates

we went from 62% completion to 92% using a claude code harness

3.8 doesn't have a minimal thinking mode, only low, medium and xhigh.

That sounds like something is off - I'm using UD-Q4_K_XL on pi with xhigh thinking, and unless I'm vastly underestimating the complexity of the script that's the kind of task I would expect to take a couple of minutes (getting ~30t/s decode). What server are you running, and are you using the recommended parameters from qwen/unsloth?

I've found it tends toward long thinking loops even for simple tasks (and any quantization seems to increase their length), but those do exit eventually, unlike with Qwen 3.6.

I use the Unsloth UD_Q2_K_XL GGUF with default parameters, along with that custom template linked elsewhere in the thread, and no K/V cache quantization.

For smaller models, you'll probably find anything below Q4 will need handholding. Check Unsloth's graphs at the different quantisations VS error rates and you'll see why.

Thank you, I know about those. And I'll stick with this quant. Normally I'd be with yout there, but Qwen 3.8 is turning out to be good at self-correcting, and the free memory I can use for extra context is worthwhile.

My experience has been that anything less than a 4-bit quant has a tendency to go off the rails. There’s a threshold of coherency that is being crossed somewhere internal to the model I guess.

Try the same prompt with a larger quant (even if it runs very slowly because the model no longer fits in VRAM) & see if Qwen does better - if so, there’s your answer.

Use Muse Glimmer. It’s good.

What harness? I've had similar results as _Implicated_ said above - it's not done well in any of the tests I've tried with it. I currently have it hung off DS4Flash as a pseudo-vision tool and subagent only because of this.

I'm using pi inside a self-made harness. I've found going super lightweight with context (AGENTS.md is maybe 20 lines) and letting the model discover what it needs to gives the best results.

Both of you should mention what quant you're using. And as another comment said, what tasks you're doing, i.e. coding, classification, summarizing etc.

I'm using unsloth dynamic Q4_K_XL.

My use-case is coding, currently working on a project with a Rust backend and TS/React/Vite frontend, with probably tens of thousands of lines of code total (including tests).

Fascinating that after years of for me this for me that, literally no one writes 2-3 other words like “i do react frontend” or whatever just for us to know why the results are different

lol, exactly.

Your assumption is incorrect.

They also need to tell people what quantisation they're using. Because some 4 bit version is not the same as BF16, no matter what KLD suggests.

I'm using the unsloth dynamic Q4 and getting good results. I was running Q5, but Q4 gives more context headroom so I can run two agents in parallel with ~100k context each with 32GB vram.

Reply on news.ycombinator.com