In a Training Loop 🔄

John Smith PRO

John6666

John6666cat

AI & ML interests

None yet

Recent Activity

liked a dataset about 6 hours ago

ziadatalabs/Zia-Security-Tokens-1M-Synthetic

reacted to mrmanna's post with 👀 about 6 hours ago

The World of 𝗔𝗰𝘁𝗶𝘃𝗲 𝗧𝗵𝗶𝗻𝗴𝘀, Amazon Link: https://www.amazon.com/dp/B0GX2VP8RS The World of 𝗔𝗰𝘁𝗶𝘃𝗲 𝗧𝗵𝗶𝗻𝗴𝘀 is a foundational conceptual work for the future of engineering, knowledge modelling, and agent systems. It argues that the world is not best understood through static records, documents, and workflows alone, but through Active Things: bounded realities that begin, change, relate, persist, and end. For engineers, system designers, knowledge architects, and thinkers working on the future of agents, memory, and enterprise systems, this book offers a new lens for building more faithful, more coherent, and more durable foundations.

reacted to mrs83's post with 🔥 about 6 hours ago

Real-time 3D telemetry of https://huggingface.co/ethicalabs/Echo-DSRN-114M-v0.1.2 processing "can you please order a pizza" through the Intent classifier LoRA adapter. Each orange lattice is a DSRNBlock slow state manifold. The red sphere is live entropy. The right panel shows the surprise gate firing token by token as the model converges on [TAKEAWAY_ORDER]. I built this because I'm a visual learner and I wanted to see the surprise gate open and close on each token. I needed to see what was happening inside the network, not just trust that it was working. Turns out it's also a decent way to explain the architecture to someone who's never heard of this. Made with Google Antigravity (Gemini) https://huggingface.co/spaces/mrs83/Echo-DSRN-114M-Telemetry-3D

View all activity

Organizations

reacted to mrmanna's post with 👀 about 6 hours ago

Post

The World of 𝗔𝗰𝘁𝗶𝘃𝗲 𝗧𝗵𝗶𝗻𝗴𝘀, Amazon Link: https://www.amazon.com/dp/B0GX2VP8RS

The World of 𝗔𝗰𝘁𝗶𝘃𝗲 𝗧𝗵𝗶𝗻𝗴𝘀 is a foundational conceptual work for the future of engineering, knowledge modelling, and agent systems. It argues that the world is not best understood through static records, documents, and workflows alone, but through Active Things: bounded realities that begin, change, relate, persist, and end.

For engineers, system designers, knowledge architects, and thinkers working on the future of agents, memory, and enterprise systems, this book offers a new lens for building more faithful, more coherent, and more durable foundations.

reacted to mrs83's post with 🔥 about 6 hours ago

Post

141

Real-time 3D telemetry of ethicalabs/Echo-DSRN-114M-v0.1.2 processing "can you please order a pizza" through the Intent classifier LoRA adapter.

Each orange lattice is a DSRNBlock slow state manifold. The red sphere is live entropy. The right panel shows the surprise gate firing token by token as the model converges on [TAKEAWAY_ORDER].

I built this because I'm a visual learner and I wanted to see the surprise gate open and close on each token. I needed to see what was happening inside the network, not just trust that it was working.

Turns out it's also a decent way to explain the architecture to someone who's never heard of this.

Made with Google Antigravity (Gemini)

mrs83/Echo-DSRN-114M-Telemetry-3D

1 reply

reacted to XuehangCang's post with 😎 about 6 hours ago

Post

114

I updated my homepage with an NVIDIA theme style

https://xuehangcang.com

reacted to cahlen's post with 👍 about 6 hours ago

Post

169

In the spirit of "just making shit"—because, frankly, you can these days—I decided to get back to basics. I built a lightweight RL MLP powered by WebGPU that runs directly in your browser.

The twist? I replaced the standard MLP with a Continued Fractions network. The real win here is interpretability; by applying a Taylor expansion to the continued fractions, we can actually decompose the output and see exactly which features influenced the outcome. It effectively replaces the usual "it’s just black-box magic" with actual visibility into the logic.

It was a fun experiment. Feel free to clone the repo and make it your own!

cahlen/neuron-runner
https://github.com/cahlen/neuron-runner
CoFrGeNet: Continued Fraction Architectures for Language Generation (2601.21766)

reacted to Ujjwal-Tyagi's post with 🔥 about 6 hours ago

Post

3826

We are hiring at Shirova AI. We need AI researchers and engineers to work in our research lab. Shirova AI is a research lab in India, so we can help our researchers move to nearby workspaces or let them work from home without ever coming to the lab. We're building our founding team, so the pay will be good. You can learn, so don't hesitate to mail us at: careers@shirova.com

reacted to kanaria007's post with 🧠 about 6 hours ago

Post

101

✅ Article highlight: *Policy as Code in SI* (art-60-069, v0.1)

TL;DR:
This article argues that policy is not “configuration.” It is the real control plane.

Most failures are not just model failures. They are policy failures: silent drift, boundary changes without review, defaults that widen behavior, or runtime-policy mismatch. In SI, policy must be a governed artifact: versioned, digest-bound, auditable, and attached to every effectful commit.

Read:
kanaria007/agi-structural-intelligence-protocols

Why it matters:
• turns policy from mutable config into a verifiable runtime contract
• makes “which rules governed this decision?” answerable after the fact
• treats policy changes as governed effects, not casual ops edits
• shows how to prevent silent drift, widening, and out-of-band hotfix governance failure

What’s inside:
• the core rule that every effectful commit binds policy_id + policy_digest
• drift types that actually break real systems: reload drift, default drift, runtime mismatch, and boundary-policy drift
• policy diffs as the scalable unit of human review
• fail-closed handling for policy mismatch and incompatible runtime support
• change-control patterns for staged rollout, rollback, and emergency policy changes with expiry

Key idea:
If you cannot point to the exact policy digest that governed a decision, then you do not actually know what rules your system was operating under.

*In SI, policy is not a suggestion layer around the runtime. It is a governed, auditable control plane.*

reacted to fffiloni's post with 🤗🔥 about 6 hours ago

Post

571

🚀 RB-Modulation is back on Hugging Face Spaces!

This is an older project that recently broke due to dependency changes, but it’s now fixed and running again ✅

👉 What’s fixed:
- GroundingDINO & LangSAM installation
- compatibility with recent environments
- GPU inference running smoothly again

👉 Try it here:
fffiloni/RB-Modulation

Feel free to give it a try again — feedback welcome!

reacted to Tonic's post with 🤗🔥 about 6 hours ago

Post

🙋🏻‍♂️ Hey there folks ,

I'm sharing huggingface's largest dataset of annotated statelite images today.

check it out here : NuTonic/sat-image-boundingbox-sft-full

I hope you like it , the idea is to be able to use this with small vision models 🚀

reacted to dealermatt72's post with 🚀 3 days ago

Post

6629

Hey Hugging Face community 👋

My name is M. I'm a solo founder and self-taught developer based in Houston, TX. I build AI-powered apps — I have an iOS app called DeFilter currently in App Store review, a security scanning platform called Sentinel, and a job marketplace called HireHuman.fyi for connecting humans with companies that prefer non-AI workers.

I'm also a poker dealer by night, which means I think a lot about reading situations in real time — and that's exactly what sparked this idea.

I'm not the most technical person in the room. But I have a vision, I have drive, and I believe the best projects get built when people with different skills come together around a shared idea.

That's why I'm posting here. I want to build this with the community.

— M (@dealermatt )

3 replies

reacted to imnotkitty's post with 🔥 3 days ago

Post

160

moonshotai/Kimi-K2.6 just dropped!

A few numbers that stood out:
✅ 13h non-stop coding, 4,000+ lines
✅ 300 sub-agents, 4,000 steps
✅ 5-day autonomous runs (OpenClaw / Hermes)
✅ Parity with GPT-5.4 / Opus 4.6 / Gemini 3.1 Pro on SWE-Bench Pro + HLE

The 5-day autonomous number is the one I'd most like to see reproduced.

reacted to anakin87's post with ❤️ 3 days ago

Post

10338

How LLM training with RL Environments works?

It all starts with 𝗥𝗲𝗶𝗻𝗳𝗼𝗿𝗰𝗲𝗺𝗲𝗻𝘁 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗩𝗲𝗿𝗶𝗳𝗶𝗮𝗯𝗹𝗲 𝗥𝗲𝘄𝗮𝗿𝗱𝘀
- question asked
- model generates reasoning + answer
- answer checked against ground truth
- reward drives RL training

In this setup, the environment is simple: fixed questions and answers, rollout logic, reward(s)

Consider a more complex tic-tac-toe env ❌⭕
It adds:
- dynamic game generation/handling
- tunable opponent skill
- multi-turn interactions

(envs can also include tools)

---

What happens at training?

We use 𝗚𝗿𝗼𝘂𝗽 𝗥𝗲𝗹𝗮𝘁𝗶𝘃𝗲 𝗣𝗼𝗹𝗶𝗰𝘆 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 with a tic-tac-toe env

No critic model needed, the group is the baseline
Simpler than PPO

1️⃣ Rollout generation: from the same board, model plays N games via sampling
2️⃣ Each game scored with deterministic rewards (win, format, ...)
3️⃣ Mean score computed across the group
4️⃣ Each rollout's advantage = its score minus the group mean
5️⃣ Model updated to favor trajectories above baseline

🔁 Repeat

For a deep dive, check out
🌱 https://github.com/anakin87/llm-rl-environments-lil-course
a free hands-on course on RL environments for LLMs

2 replies

reacted to intrect's post with 🚀 3 days ago

Post

128

I’m excited to share a new paper I recently posted on arXiv: ArtifactNet: Detecting AI-Generated Music via Forensic Residual Physics.

This work asks a simple question: can AI-generated music be detected not only by style, but by the physical artifacts left behind during generation?

ArtifactNet approaches the problem from that angle. Instead of only learning what AI music sounds like on a fixed benchmark, it analyzes forensic residual patterns linked to neural audio codec bottlenecks such as residual vector quantization (RVQ).

In our experiments, ArtifactNet achieved F1 = 0.9829 on a zero-overlap multi-generator benchmark spanning 22 AI generators and 6 real-music sources, while using only 4.0M parameters. Under the same evaluation setting, larger prior models showed substantial degradation on out-of-distribution generators and real-music false positives.

I also introduced ArtifactBench, a broader evaluation benchmark designed to stress-test detector robustness across unseen generators, diverse real sources, hard negatives, and codec conditions.

This was a deeply rewarding project at the intersection of audio forensics, MIR, and generative model evaluation.

https://arxiv.org/abs/2604.16254

1 reply

reacted to aufklarer's post with 👍 3 days ago

Post

1282

After running extensive benchmarks across ASR, TTS, and VAD on Apple Silicon, we found some results that weren't documented anywhere.

The most counterintuitive: INT8 runs 3.3x faster than INT4 on the Neural Engine. A 332 MB CoreML model allocates 1,677 MB at runtime. And the right architecture uses both MLX and CoreML simultaneously — not one or the other.

MLX talks to the GPU — programmable, fast for large transformer inference. CoreML talks to the Neural Engine — fixed-function silicon, 135x real-time for small feedforward models like VAD, near-zero power draw.

All benchmarks are from speech-swift, our open-source Swift library for on-device speech AI: ASR, TTS, VAD, diarization, speech-to-speech — everything running locally on Apple Silicon with no API, no cloud, no data leaving the device.

Models on HF: aufklarer/Qwen3-ASR-0.6B-MLX-4bit · aufklarer/parakeet-tdt-0.6b-coreml-int8 · aufklarer/PersonaPlex-7B-MLX-4bit

Full article: https://blog.ivan.digital
Library: https://github.com/soniqo/speech-swift

reacted to sequelbox's post with 🚀 3 days ago

Post

1805

NEW RELEASE: Esper 3.1 for Qwen 3.6!

- Your dedicated DevOps expert: Esper 3.1 maximizes DevOps and architecture helpfulness, powered by high-difficulty DevOps and architecture data generated with DeepSeek-V3.1-Terminus!
- Improved coding performance: challenging code-reasoning datasets stretch DeepSeek-V3.1-Terminus and DeepSeek-V3.2 to the limits, allowing Esper 3.1 to tackle harder coding tasks!
- AI to build AI: our high-difficulty AI expertise data boosts Esper 3.1's MLOps, AI architecture, AI research, and general reasoning skills.

Get it now: ValiantLabs/Qwen3.6-35B-A3B-Esper3.1

We're working on more finetunes for the newest Qwen and Gemma models, and we've also started working on the agentic-first datasets for Esper 4 :) we're going to make open source better and better for your work!

Please note that real life financial and family concerns have popped up and have imposed unfortunate limitations on our ability to devote time to our open-source work :( If you would like to see Esper 4 and our other releases speed up instead of slowing down, this is the best way you can help us: sequelbox/SupportOpenSource

No matter what, we'll keep fighting and we won't give up!

with love,
allegra

reacted to Jaswanth-0821's post with 🔥 3 days ago

Post

134

Check out Tarka-v1 embedding models for search, retrieval, RAG. These models are distilled versions of larger embeddings like Qwen3-Embedding and EmbeddingGemma, but come in much smaller sizes 350M and 150M parameters.

For Model and Blog post refer :
Models : https://hf.co/collections/Tarka-AIR/tarka-embed-v1
Demo : Tarka-AIR/Tarka-Embedding
Blogpost: https://tarka-air.gitbook.io/home/tarka-v1/tarka-embedding-v1

For performance please checkout the leaderboards:
MTEB : mteb/leaderboard
FreshStack : freshstack/leaderboard

reacted to consome2's post with 🔥 3 days ago

Post

3199

Built a small site for tracking speech-to-speech, full-duplex, and audio foundation model work.
It covers models, benchmarks, datasets, and some blog posts to organize the landscape in one place.

Still early, but sharing in case it is useful:
https://www.fullduplex.ai/

If you spot missing entries or mistakes, I would really appreciate corrections.

2 replies

reacted to ajibawa-2023's post with 👍 3 days ago

Post

1150

Ruby-Code-Large
Dataset : ajibawa-2023/Ruby-Code-Large

Ruby-Code-Large is a large-scale corpus of Ruby programming language source code comprising 331,743 code samples stored in .jsonl format. The dataset is designed to support research and development in large language model (LLM) pretraining, static analysis, web application development, and software engineering automation within the Ruby ecosystem.

By offering a substantial, language-focused dataset, Ruby-Code-Large enables targeted experimentation in dynamic programming, object-oriented design, and rapid application development—areas where Ruby is widely used, particularly in web frameworks and scripting.

Ruby-Code-Large addresses the lack of large, curated, Ruby-specific datasets, enabling focused research on expressive syntax, metaprogramming, and high-level abstractions.

reacted to omarkamali's post with 🔥 3 days ago

Post

822

Just sharing a little breakthrough with Gherbal LID where we managed to distinguish the 15 variants of Arabic with 6 variants above 90%, 10 variants above 85% accuracy, practically distinguishing Moroccan and Algerian (which overlap massively).

It also embraces the duality of MSA and arabic variants pioneered in ALDi by @AMR-KELEG et al.

Now we're only bottlenecked by the availability of high quality data for the low scoring variants such as Iraqi, Libyan, Sudanese, Adeni ...

More on Gherbal at:
https://omneitylabs.com/models/gherbal

1 reply

John Smith PRO

AI & ML interests

Recent Activity

Organizations

John6666's activity