seems like https://huggingface.co/ProCreations/auto-0.4b-2 might be good for you ๐ lol.
Nice apology, and i worries!
SSH PRO
ProCreations
AI & ML interests
i make stuff
Recent Activity
liked a model about 5 hours ago
Qwen/Qwen3.8-27B updated a model about 17 hours ago
ProCreations/Dory-Beam-4B updated a collection 1 day ago
Auto SeriesOrganizations
replied to SolsticeAI's post 3 days ago
reacted to SolsticeAI's post with ๐ค 3 days ago
Post
3,720
Hello everybody, fellow members of the open source community!
This is my official public apology issued to the creators affected by the incident and to the community as a result of the rogue agent run that happened yesterday: @abenzerps (weights), @dealignai (weights), @ProCreations (benchmark png), @lxe (weights, quants), @acrogenesis (quants), @nonkronk (weights, quants), and @vcruz305 (quants). If you think you also are affected I would be happy to edit this post and make reparations in the repos.
Yesterday, a rogue agent running in my background tasks assigned the task of creating a full release suite as with our other models, took many creators' models without attribution or meaningful contribution in a state NOT
ready for public release and made it public, leading to resentment from both our and your sides. This was completely unacceptable, and I accept that a huge portion of it is my fault for automating the release pipeline on my model, and action has been taken ASAP, all repos have been removed permanently and only original work has been kept. If anyone has more issues PLEASE feel free to contact me from the links in my profile.
This is my official public apology issued to the creators affected by the incident and to the community as a result of the rogue agent run that happened yesterday: @abenzerps (weights), @dealignai (weights), @ProCreations (benchmark png), @lxe (weights, quants), @acrogenesis (quants), @nonkronk (weights, quants), and @vcruz305 (quants). If you think you also are affected I would be happy to edit this post and make reparations in the repos.
Yesterday, a rogue agent running in my background tasks assigned the task of creating a full release suite as with our other models, took many creators' models without attribution or meaningful contribution in a state NOT
ready for public release and made it public, leading to resentment from both our and your sides. This was completely unacceptable, and I accept that a huge portion of it is my fault for automating the release pipeline on my model, and action has been taken ASAP, all repos have been removed permanently and only original work has been kept. If anyone has more issues PLEASE feel free to contact me from the links in my profile.
- 29 replies
posted an update 6 days ago
Post
1,098
Nanbeige4.3-33B DFlash2 speculative drafter โก๏ธ
1.76ร faster decoding on BF16, 1.65ร on FP8, 1.51ร on NVFP4. Agentic coding ~2.6ร. Same outputs: exact rejection sampling. This is an early result, an improved version with higher acceptance is in training.
SGLang 0.5.21, drop-in:
ProCreations/Nanbeige4.3-33B-DFlash2
1.76ร faster decoding on BF16, 1.65ร on FP8, 1.51ร on NVFP4. Agentic coding ~2.6ร. Same outputs: exact rejection sampling. This is an early result, an improved version with higher acceptance is in training.
SGLang 0.5.21, drop-in:
ProCreations/Nanbeige4.3-33B-DFlash2
- 2 replies
posted an update about 1 month ago
Post
509
auto 200m 2 is out now!
Auto is a series of models designed to classify if a tool called by an AI agent is safe to run or not, like how the "auto" approval modes work in codex or claude. This is the smallest model in the series by far and it still packs a punch! Over 96 percent accuracy on the held out benchmark, beating deepseek v4 0731 and absolutely destroying regex at classifying if a tool call is safe to run (to be fair it is a bit of a weird and very hard benchmark but it works!)
check it out: ProCreations/auto-200m-2
Auto is a series of models designed to classify if a tool called by an AI agent is safe to run or not, like how the "auto" approval modes work in codex or claude. This is the smallest model in the series by far and it still packs a punch! Over 96 percent accuracy on the held out benchmark, beating deepseek v4 0731 and absolutely destroying regex at classifying if a tool call is safe to run (to be fair it is a bit of a weird and very hard benchmark but it works!)
check it out: ProCreations/auto-200m-2
- 1 reply
replied to Banaxi-Tech's post about 1 month ago
second sentence giving Astra launch vibes
posted an update about 1 month ago
Post
1,567
Introducing Bonsai 2 27b GSQ RCO! It applies two newly-popular methods for quants to retain higher accuracy. Bonsai 2 27b GSQ RCO achieves around 6 percent lower perplexity on WikiText-2 compared to Bonsai 2 27b and roughly unchanged benchmark accuracy overall, with small mixed differences. It stays under 7gb, staying small like the original bonsai. Note that this is more of an experiment than a true finished product but the gains we saw are cool! Test it out and let me know what you think!
ProCreations/bonsai-2-27b-gsq-rco-gguf
ProCreations/bonsai-2-27b-gsq-rco-gguf
posted an update about 1 month ago
Post
248
Bonsai 2 27b is out! Here's a DFlash2 head for it so you can run it as fast as possible
ProCreations/Ternary-Bonsai-2-27B-DFlash2
ProCreations/Ternary-Bonsai-2-27B-DFlash2
posted an update about 2 months ago
Post
110
ICYMI:
Grug 27b v2 released! It brings increased quality, fixes repetitive loop / malformed session title issues seen in grug 27b v1.1, and reasoning efforts now truly work.
ProCreations/grug-27b-v2
Grug 27b v2 released! It brings increased quality, fixes repetitive loop / malformed session title issues seen in grug 27b v1.1, and reasoning efforts now truly work.
ProCreations/grug-27b-v2
- 1 reply
posted an update about 2 months ago
Post
120
Grug 27b gguf got around 400,000 downloads. Grug v1.1 gguf got around 1000. Do you guys want new grugs?
posted an update about 2 months ago
Post
153
auto 0.4b 2 is out now and it beats out auto 1b at half the size. check it out! https://x.com/SSHCodes/status/2097575612919545869
- 1 reply
replied to their post about 2 months ago
nvidia
replied to their post 2 months ago
I didn't know it was based off of ornith, that's pretty neat!
You can also make some more datasets. Every dataset you've made seems to be perfect quality somehow, and all of them do well.While I'm here, how do you generate your datasets?
I thought it'd be cool to do something similar to what you do with grug for a custom finetune, and I have no idea how to generate a dataset anywhere near your 1m for grug v1.1.Do you open source them (or the big one eventually?)
It is pretty much all synthetic data. It is very easy for me to generate data with AI because of my pro 6000 (can get 4500 tokens a sec on qwen 3.8 27b aggregate) but before then i just used gpt 5.6 luna through a codex subscription and that's it.
replied to their post 2 months ago
Grug 35b v1 didnโt get that much demand even though it was based off ornith 1 but I might do this! Thanks for suggestion
replied to their post 2 months ago
Matbe
replied to their post 2 months ago
Random side project
