Hugging Face's logo

DavidAU
/
support

Any-to-Any
Transformers
English
sponsor
buy me a coffee
projects
help
support
contact
partners
ggufs
moes
mixture of experts
source
custom
models by sort
FAQs
help docs
SOTA models
about us
Turbo Brilliance Systems
NEO Imatrix
source code
Brainstorm Adapter
Cold Fusion Training
Model merging
model tuning
unsloth
mergekit
llamacpp
Model card Files
xet
 Community
1

SUPPORT / HELP / SPONSOR / PROJECTS / PREMIUM / SORT Models / FAQs

To support our team, sponsor, "Buy Me A Coffee", commission project(s), access premium item(s)/model(s)/module(s), model modifications, GET HELP, see a sorted list of models (380+) and more please see below.

We build precision models at scale for the community.

We are user-funded, and use local hardware to create, tune and deploy models worldwide.

We also create systems for fine tuning, and specialized model systems too as well as "augment" current model/AI technology.

That includes:

  • Model fine tuning across a diverse arch type of models and sizes, including MOEs/sparse moes. (over 1000)
  • Model merging (and also fine tuning) (over 2700)
  • Model expansion/contraction (27B-> 40B ; 27B to 21B, 13B... and so on => both "dense" and "moe models")
  • Cold Fusion Training Technology for model fine tuning.
  • Brainstorm Universal model adapter to improve intelligence and creativity.
  • Turbo Brilliance => A universal model generation, IQ, and token reduction / enhancement technology.
  • NEO Imatrix dataset => Strongest Imatrix dataset ever built.

Some models took hours to tune, whereas others took days to weeks including the countless hours of testing; especially for our SOTA models.

If you would like to sponsor our work the "Buy Me a Coffee" system was just added as per multiple requests and the fact is we can do a lot more, we just need the equipment to do it:

https://buymeacoffee.com/davidbelton

Thank you in advance.

ABOUT DavidAU:

  • As of this writing my repo (DavidAU) has over 4 million downloads per month. This does not account for "user setup repos" / mirrors of my teams models.
  • I have published over 2000 models over the past 3 years - Llama, Qwen, Mistral, LFM, OpenAI, GLM, Gemmas and many others. (see details below)
  • After surpassing 10,000 followers on Huggingface, it is now time to take things to the next level.
  • The journey is just getting underway...
  • PS: about "actual me"... see below.

Our latest project(s) (MAJOR) is here (names subject to change):

Contact / About Me - David Belton:

Main partner "in crime" is "nightmedia":

Here are some of the (380+ :: Number was 1500 at one point, but cut due to HGF storage limits) strongest models so far:

NOTE: I generate source and GGUF quants, for many of my models, the community has generated additional quant types/variations like MLX, NVFP4, and so on... you will find these quants in the "community" tab at the repo and/or under "Quantizations" on the main repo page.

BY SORT - GENERAL:

BY SORT - MOEs/Sparse MOES:

By SORT - SIZE (there will be other sizes/oddball sizes too):

BY SORT - MODEL:

Special note - "Out of print" models:

SOTA MODELS By My Team:

NOTE: Most models have benchmarks, and 3rd party benchmarks (at repo or in the community section)

Qwen3.6 27B: The strongest, overall qwen ever beating all other Qwens in total operational power with over 2400 likes // 4 million+ total downloads and has over 70 additional repos with additional quant types:

Qwen3.8 27B: The highest scoring Qwen in brute, raw intelligence, using Qwen 3.8's 3 new reasoning modes, plus token reduction (1/2 to 1/10) enhancements and over 1600 likes + and 100+ additional quant repos:

Qwen3.8 27B: Super smart and 1/2 to 1/20 the reasoning tokens AND 5 reasoning/5 instruct modes switchable on the fly (even in chat):

Qwen3.8 27B: 99% power of BF 16 at 4 and 8 bit. Power, Control and NO DE censoring for ultimate performance also with reasoning token reductions:

Qwen3.6 40B: The 40B Monster, specializing in creative and research with 730+ likes and over 2 million downloads:

Qwen3.5 9B: At just 9B parameters it beats most untuned 27B models in both intelligence (640 ARC-C) and performance, over 1000 likes, plus features 5 reasoning and 5 instruct modes (Qwen 3.8) too:

For model specific issues/help/FAQ see:

Support our team, sponsor, "Buy Me A Coffee":

Commission project(s):

  • coming soon.

Access premium item(s)/model(s)/module(s):

  • coming soon.

Model Modifications:

  • coming soon.

ALL OTHER INQUIRES:

  • Open a discussion in the community tab at this repo.
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Any-to-Any
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Collections including DavidAU/support