this post was submitted on 23 Jul 2026

28 points (93.8% liked)

LocalLLaMA

5120 readers

46 users here now

Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.

Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.

As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.

Rules:

Rule 1 - No harassment or personal character attacks of community members. I.E no namecalling, no generalizing entire groups of people that make up our community, no baseless personal insults.

Rule 2 - No comparing artificial intelligence/machine learning models to cryptocurrency. I.E no comparing the usefulness of models to that of NFTs, no comparing the resource usage required to train a model is anything close to maintaining a blockchain/ mining for crypto, no implying its just a fad/bubble that will leave people with nothing of value when it burst.

Rule 3 - No comparing artificial intelligence/machine learning to simple text prediction algorithms. I.E statements such as "llms are basically just simple text predictions like what your phone keyboard autocorrect uses, and they're still using the same algorithms since <over 10 years ago>.

Rule 4 - No implying that models are devoid of purpose or potential for enriching peoples lives.

founded 3 years ago

MODERATORS

pax@sh.itjust.works

noneabove1182@sh.itjust.works

Smokeydope@lemmy.world

MonsterBug@sh.itjust.works

Grug (caveman compression) of Qwen3.6-35 and 27B is good! (aussie.zone)

submitted 1 month ago* (last edited 1 month ago) by SuspiciousCarrot78@aussie.zone to c/localllama@sh.itjust.works

8 comments fedilink hide all child comments

Hi all, if you've not heard, there's been a Caveman compression version of Qwen 3.6-27b and Qwen 3.6-35b-3ab.

https://huggingface.co/ProCreations/grug-27b-gguf

https://huggingface.co/ProCreations/grug-35b-v2

https://huggingface.co/ProCreations/grug-35b-v2-gguf

I've been playing around with 35B and I am able to run it on my ancient Quadro P1000 4gb at ~10tok/s.

But beyond that, the quality of the reasoning and the token discipline / output is actually higher than a stock, in my opinion.

You can read the benchmarks above; I am also uploading a HTML file here for your consideration.

(Sorry for the Limewire link; I dunno where else to share throw-away files. It's HTML, though the side by side probably works better if you actually download it)

Anyway...I'm doing more testing right now...but so far, this is a good cook.

Or -

Grug good. Me like.

top 8 comments

sorted by: hot top controversial new old

[–] SuspiciousCarrot78@aussie.zone 6 points 1 month ago* (last edited 1 month ago) (1 children)

Oh and if anyone is interested in my local settings

  -t 12 ^
  -tb 12 ^
  -ngl 28 ^
  --n-cpu-moe 41 ^
  --no-mmap ^
  --mlock ^
  --cache-type-k q4_0 ^
  --cache-type-v q4_0 ^
  -c 16384 ^
  -b 256 ^
  -ub 128 ^
  --host 0.0.0.0 ^
  --port %PORT% ^
  --ui-mcp-proxy

Turbo quant is not supported on this GPU (too old),

[–] SuspiciousCarrot78@aussie.zone 2 points 1 month ago* (last edited 1 month ago) (1 children)

Slight revision: on my rig (i7-8700, 32gb, Quadro p1000 4gb ddr5), this gives a nice 12 tok/s. Unfortunately, it's still hostile to my 1L box and very quickly thermally swamps the CPU (while gpu sits at 56 degrees, that little shit). C'est la vie.

 -t 8 ^
  -tb 8 ^
  -ngl 99 ^
  --n-cpu-moe 38 ^
  --flash-attn on ^
  --no-mmap ^
  --mlock ^
  --cache-type-k q4_0 ^
  --cache-type-v q4_0 ^
  -c 16384 ^
  -b 256 ^
  -ub 128 ^
  --host 0.0.0.0 ^
  --port %PORT% ^
  --ui-mcp-proxy

[–] brucethemoose@lemmy.world 2 points 1 month ago* (last edited 1 month ago)

Be aware, q4_0 KV quantization really borks models.

But the K is far more sensitive than the V. Try q5_1 for the K while leaving the V at q4_0 or q4_1; vram usage will be almost the same, but it should work dramatically better.

As for throttling, try disabling turbo on the 8700.

It doesn’t actually need turbo clocks for these models. The t/s loss I get from doing that on my rig is very modest.

And like others suggested, try the QAT release. You might try the ik_llama.cpp for while you’re at it, at it should be faster with MoEs like this.

[–] keepthepace@tarte.nuage-libre.fr 5 points 1 month ago (1 children)

The Model card is fun to read and the idea is solid. Thinking tokens always striked me as unoptimized and a bit forced.

[–] SuspiciousCarrot78@aussie.zone 3 points 1 month ago

I suspect the verbosity of Qwen 3.6 CoT token's is what causes the famous schitzo loop. Will be interesting to see if this holds.

The QAT versions also dropped today

[–] SuspiciousCarrot78@aussie.zone 2 points 1 month ago* (last edited 1 month ago)

Still fine tuning this. I'm finding that Grug has a particular affinity for larger -b and -ub sizes. I'm now able to hit 90-110 tok/s prefil (upto 155 tok/s fresh). Specific details below

155.42 tok/s fresh 2,009-token prefill
roughly 90–110 tok/s on larger incremental/agentic prefills
roughly 11.3–12.9 tok/s sustained generation
64°C peak CPU temperature
approximately 33 W peak CPU package power
no thermal throttling
ub and -b both 2048

All this on a Quadro P1000 4GB card.

In-sane.

    -m "%MODEL_PATH%" ^
    -t 8 ^
    -tb 8 ^
    -ngl 99 ^
    --n-cpu-moe 38 ^
    --flash-attn on ^
    --no-mmap ^
    --mlock ^
    -c 16384 ^
    -b 2048 ^
    -ub 2048 ^
    -np 1 ^
    --host 0.0.0.0 ^
    --port %PORT% ^
    --ui-mcp-proxy

PS: Interestingly, dropping cache precision away from FP16 reduced tok/s generation by 18%. I don't know why. I'm just fine tuning MTP now to see if I can eke out a few more tok/s, as MTP enabled grug-v2 just dropped

https://huggingface.co/ProCreations/grug-35b-mtp-gguf

PPS: Sadly also discovered my Tesla card has intermittent electrical fault. Took some troubleshooting to figure it out, but the long and short of it is you probably shouldn't buy second-hand server cards off Ebay. Oh well, $100. On the upside, tweaking throttlestop, re-pasting and re-seating greatly improved thermals.

[–] trem@lemmy.blahaj.zone 2 points 1 month ago (1 children)

Huh, does caveman just mean that it uses fewer fill words and less prosa, or does the output only look like that by chance?

I thought, I read at some point that the caveman models use some pseudo-language for their thinking tokens. If they waffle less, that would be much more interesting to me...

[–] SuspiciousCarrot78@aussie.zone 3 points 1 month ago

Both; it has internally reduced thinking tokens AND it's less verbose, so you get a boost on both prefill and print. I've benchmarked identical prompts and tool calls with stock Qwen 35B and Grug 35B; on average, what takes stock 1000 token/s, Grug does in about 200.