In a Training Loop 🔄

John Smith PRO

John6666

John6666cat

AI & ML interests

None yet

Recent Activity

liked a model about 7 hours ago

vanta-research/PE-Type-3-Nova-4B

liked a model about 7 hours ago

mradermacher/PE-Type-3-Nova-4B-GGUF

liked a model about 7 hours ago

mradermacher/PE-Type-3-Nova-4B-i1-GGUF

View all activity

Organizations

reacted to perfecXion's post with 👍 about 11 hours ago

Post

461

# IntentGuard: Open-Source Vertical Intent Classifiers for LLM Guardrails

Three models published to the Hub:

- [perfecXion/intentguard-finance]( perfecXion/intentguard-finance)
- [perfecXion/intentguard-healthcare]( perfecXion/intentguard-healthcare)
- [perfecXion/intentguard-legal]( perfecXion/intentguard-legal)

DeBERTa-v3-xsmall fine-tuned for three-way classification: **allow**, **deny**, or **abstain**. ONNX + INT8 quantized, under 80MB, p99 <30ms on CPU. Margin-based thresholds (not argmax) — uncertain queries route to clarification instead of forcing a guess.

**Eval results (adversarial test sets, ~470-480 examples per vertical):**

| Vertical | Accuracy | Legit-Block Rate | Off-Topic-Pass Rate |
|----------|----------|------------------|---------------------|
| Finance | 99.6% | 0.00% | 0.00% |
| Healthcare | 98.9% | 0.00% | 0.98% |
| Legal | 97.9% | 0.00% | 0.50% |

docker run -p 8080:8080 ghcr.io/perfecxion/intentguard:finance-latest

curl -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role": "user", "content": "What are current mortgage rates?"}]}'

Apache 2.0. Full pipeline + Docker configs on [GitHub](https://github.com/perfecxion/intentguard).

Feedback welcome on domain coverage, adversarial robustness, and multilingual demand.

reacted to TravisMuhlestein's post with 👍 about 11 hours ago

Post

443

Moving AI from experiments to production systems (GoDaddy + AWS case study)

A recurring pattern across many organizations right now is that AI experimentation is easy — operationalizing it is much harder.

This case study from AWS describes how GoDaddy has been deploying AI systems in production environments using AWS infrastructure.

One example is Lighthouse, a generative AI system built using Amazon Bedrock that analyzes large volumes of customer support interactions to identify patterns, insights, and opportunities for improvement.

The interesting part isn’t just the model usage — it’s the system design around it:

- large-scale interaction data ingestion
- LLM-driven analysis pipelines
- recursive learning platforms where real-world signals improve systems over time
- infrastructure designed for continuous iteration

We’re starting to see a shift where organizations move from AI prototypes toward AI platforms and production systems.

Would be interested to hear how others in the community are thinking about:

- production AI architectures
- LLM evaluation pipelines
- Feedback loops in real-world systems
- infrastructure for scaling AI workloads

Case study:
https://aws.amazon.com/partners/success/godaddy-agenticai/

reacted to AbstractPhil's post with 👍 about 11 hours ago

Post

489

I've... done it. This, with experts, achieves near 100% R1 retrieval accuracy on an adjacent - unseen by the fusion transformer - dataset with around 40k steps from the seen dataset. This means the language of the models are at least tested fused within the constraints, not just projected or estimated.
AbstractPhil/geolip-procrustes

I encourage EVERYONE who is curious to check my work. Check it, double check it, and triple check it.

These were aligned using COCO and then validated with Flickr. Entirely different datasets. The experts arbitrated and the alignment yielded the correct answers. Preliminary tests show that with almost no alignment requirement, the models can reach 100% R1 retrieval accuracy.

Not to be confused with validation accuracy for a classification model or a text encoder's text response, this allows multispectral communication between entirely different models for direct downstream consumption with almost no training for the chosen models.

I have a working procrustes experiment that learns adjacent manifolds within a reasonable spectrum and the speed is... well, 1 epoch with COCO using Bert-Large and DinoV2 that allows the models to align nearly perfectly. For some scales in the experiment it shows that the 3 set epochs aren't quite enough to align R1 to highest, while many align nearly immediately.

These two were an obvious pair to pick, 60% similarity and >90% spectral similarity.

The trainer transfers layers, learns embeddings, and more - all by sticking strictly to geometric boundaries and procrustes informational accumulation within a modulation model's constraints.

I have many experiments to run.

reacted to umarbutler's post with 🔥 about 11 hours ago

Post

720

This awesome visualization by @abdurrahmanbutler tracks how reliant the High Court of Australia has been on UK precedents over time.

Back in the early 1900s, up to 70% of citations in High Court decisions were from the UK. Today, that number sits around 20%.

This change seems to have happened gradually as Australia gained more and more independence from the UK, culminating in the Australia Acts of 1986, where we see a nice bump in the proportion of Australian cases cited.

These insights would not be possible without our latest legal AI model, Kanon 2 Enricher, which we used to extract dates and citations from High Court decisions in isaacus/open-australian-legal-corpus and categorize citations by jurisdiction. You can learn about Kanon 2 Enricher here: https://isaacus.com/blog/kanon-2-enricher.

reacted to ronantakizawa's post with 👍 1 day ago

Post

1459

Introducing the github-codereview dataset: A compilation of 200k+ human-written code reviews from top OSS projects (React, Tensorflow, VSCode...).

I finetuned a Qwen2.5-Coder-32B-Instruct model with this dataset and saw significant improvements in generating better code fixes and review comments (4x improved BLEU-4, ROUGE-L, SBERT scores compared to base model).

#codereview #code #datasets

ronantakizawa/github-codereview

reacted to vincentg64's post with 👀 1 day ago

Post

220

I just started putting all the new stuff about xLLM v 2.0 on GitHub, starting with various agents. All home-made from scratch with radically different technology. You can check the new repository at https://mltblog.com/4riZjl9

reacted to alvdansen's post with 🚀 1 day ago

Post

776

Releasing Flimmer today — a video LoRA training toolkit for WAN 2.1 and 2.2 that covers the full pipeline from raw footage to trained checkpoint.
The standout feature is phased training: multi-stage runs where each phase has its own learning rate, epochs, and dataset, with the checkpoint carrying forward automatically. Built specifically with WAN 2.2's dual-expert MoE architecture in mind.

Data prep tools are standalone and output standard formats — they work with any trainer, not just Flimmer.

Early release, building in the open. LTX support coming next.

http://github.com/alvdansen/flimmer-trainer

reacted to alvarobartt's post with 🔥 1 day ago

Post

2595

Learn how to deploy Microsoft Research VibeVoice ASR on Microsoft Azure Foundry with Hugging Face to generate rich audio transcriptions with Who, When, and What! 💥

> 🕒 60-minute single-pass processing, no chunking or stitching
> 👤 Customized hotwords to guide recognition on domain-specific content
> 📝 Rich transcription: joint ASR + diarization + timestamping in one pass
> 🌍 50+ languages with automatic detection and code-switching support
> 🤗 Deployed on Microsoft Foundry via an OpenAI-compatible Chat Completions API

https://huggingface.co/docs/microsoft-azure/foundry/examples/deploy-vibevoice-asr

reacted to OzTianlu's post with 🤗 1 day ago

Post

1428

We deleted the Embedding Layer -- INTRO Our Collins-Embedding-3M
NoesisLab/Collins-Embedding-3M
Most "small" models are just giant vocab tables in a trench coat. Collins-3M changes that. By using 2-Universal Hashing and Chernoff-bound noise suppression, we’ve collapsed the embedding space into a fixed O(1) hash-map.
* STSB: 0.7114 (Beating many 100M+ models)
* Size: 3M (Edge-ready, IoT-ready)
* Tech: Randomized Sign-Hashing + RoPE positional injection.
Built by NoesisLab

reacted to danielhanchen's post with 🤗🔥 1 day ago

Post

1754

You can now fine-tune Qwen3.5 for free with our notebook! 🔥

You just need 5GB VRAM to train Qwen3.5-2B LoRA locally!

Unsloth trains Qwen3.5 1.5x faster with 50% less VRAM.
GitHub: https://github.com/unslothai/unsloth
Guide: https://unsloth.ai/docs/models/qwen3.5/fine-tune
Qwen3.5-4B Colab: https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/Qwen3_5_(4B)_Vision.ipynb

reacted to branikita's post with 🚀 1 day ago

Post

1334

We tested our 3D-printed parallel gripper for the SO-ARM100/101 robotic platform, successfully handling a 1.5 kg payload. The gripper features a 100.5mm full stroke and ±0.05mm repeatability — all for around $76 in parts and 30 minutes of assembly.

Full source code, STL files, and assembly guide are open-source and available on GitHub: https://github.com/roboninecom/SO-ARM100-101-Parallel-Gripper

reacted to BestWishYsh's post with 🚀 1 day ago

Post

2761

🚀 Introducing Helios: a 14B real-time long-video generation model!

It’s completely wild—faster than 1.3B models and achieves this without using self-forcing. Welcome to the new era of video generation! 😎👇

💻 Code: https://github.com/PKU-YuanGroup/Helios
🏠 Page: https://pku-yuangroup.github.io/Helios-Page
📄 Paper: Helios: Real Real-Time Long Video Generation Model (2603.04379)

🔹 True Single-GPU Extreme Speed ⚡️
No need to rely on traditional workarounds like KV-cache, quantization, sparse/linear attention, or TinyVAE. Helios hits an end-to-end 19.5 FPS on a single H100!

Training is also highly accessible: an 80GB VRAM can fit four 14B models.

🔹 Solving Long-Video "Drift" from the Core 🎥
Tired of visual drift and repetitive loops? We ditched traditional hacks (like error banks, self-forcing, or keyframe sampling).

Instead, our innovative training strategy simulates & eliminates drift directly, keeping minute-long videos incredibly coherent with stunning quality. ✨

🔹 3 Model Variants for Full Coverage 🛠️
With a unified architecture natively supporting T2V, I2V, and V2V, we are open-sourcing 3 flavors:

1️⃣ Base: Single-stage denoising for extreme high-fidelity.
2️⃣ Mid: Pyramid denoising + CFG-Zero for the perfect balance of quality & throughput.
3️⃣ Distilled: Adversarial Distillation (DMD) for ultra-fast, few-step generation.

🔹 Day-0 Ecosystem Ready 🌍
We wanted deployment to be a breeze from the second we launched. Helios drops with comprehensive Day-0 hardware and framework support:

✅ Huawei Ascend-NPU
✅ HuggingFace Diffusers
✅ vLLM-Omni
✅ SGLang-Diffusion

Try it out and let us know what you think!

3 replies

reacted to AbstractPhil's post with 👀 1 day ago

Post

218

The small projection-based approximator model for the geolip patchwork did not breach a certain level of accuracy as required by my specifications, so I've defaulted to harvesting direct geometric information from AI models until I get the comparative bounds required for a useful topology.

I must sincerely apologize for not solving this problem quickly.

This will take time. Without the approximator it's going to be considerably slower, but this model I begin training will be providing the approximations in a different way over time. As iterations progress, the system will conform to a huge array of geometric potentials and be capable at predicting those, but it will not be as powerful as the full patchmaker up front, and it will be slow training.

If I can get my hands on a cluster of A100's or H100's for a measure I'll make a post immediately, until then I must default to the slower process.

I really banked that the smaller version would have worked, but it simply couldn't hold complex topological shape without the correct boundaries being learnable AND endure entropic decay simultaneously. The only way to have a predominant shot at a full geometric shared language, is to make those boundaries learnable in the full spectrum of potentials, or at least more than I have placed on it.

I'll be refining my process in the coming days further, and I do apologize for pre-emptively announcing a potential that I have yet to fully explore.

There will be a full upgraded 38 shape geolip patchwork trained asap to fully encompass the Flux 1 AE spectrum, and another trained for SD15, SDXL, and Flux 2's VAE as well. These will accommodate DIRECT complex geometric patchwork learning, but not to the scale as promised yet. Autoregression is a complex mistress as many of you know, and I will be spending a great deal of time and compute analyzing all of the information required to build a uniformly useful and powerful autoregression patchwork to utilize as invariance to teaching.

2 replies

reacted to sequelbox's post with 👀👀 1 day ago

Post

1192

Multiple new releases for Qwen 3.5 27B!

Firstly, Guardpoint, our medical reasoning model; trained on medical knowledge, management, diagnosis, and tasks:
- Structured medical reasoning responses are efficient and informative, cutting token costs for faster inference!
- Wide-ranging knowledge base: trained on a wide variety of medical disciplines, patient types, and query structures!
- High quality medical responses emphasize performance, brevity, specificity, statistical rationality, and openness.

Get Guardpoint for Qwen 3.5 27B: ValiantLabs/Qwen3.5-27B-Guardpoint

Secondly, we've also brought DAG Reasoning to Qwen 3.5 27B:
- Create structured, analytical Directed Acyclic Graphs to provide insight into your queries and situations!
- Multi-step analysis identifies causal relationships, produces confidence measurements, and forms a single structured graph object.
- DAG Reasoning Format provides clear, readable JSON containing structured, useful information; easy to use for creating visualizations, doing analysis, or further conversation with your assistant.
- Trained in a variety of subjects for flexible analysis: programming, science, business, economics, finance, law, logistics, management, and more!

Get the newest DAG Reasoning release: sequelbox/Qwen3.5-27B-DAG-Reasoning

We also have Esper 3.1 available for Qwen 3.5 27B - focused on high-performance coding, DevOps, and architecture: ValiantLabs/Qwen3.5-27B-Esper3.1

We'll have a lot more to come for the high-performance Qwen 3.5! Most of it is waiting for Deepseek V4 to come out first :) We've got some fun ideas!

Help support our releases, donations used for our experimental models and datasets: sequelbox/SupportOpenSource

Fight for open source with us! We've got a lot to do.

for friendship,
allegra

reacted to JonnaMat's post with 🚀 1 day ago

Post

1001

Qwen3.5 on-device benchmarks on the Nvidia Jetson lineup are now live 🚀

We've added the latest Qwen3.5 models (0
8B - 9B) to our on-device inference benchmarks (Nvidia Jetson Orin Nano Super, AGX Orin, AGX Thor).

👉 Explore TPS, TTFT, E2E latency, and TPOT. Measured on real hardware: embedl/Edge-Inference-Benchmarks

🌟 Stay tuned for additional benchmarks and Embedl-optimized models: Enabling models run faster and on less expensive hardware.

If you're working on edge LLM deployment, we'd love to discuss your use case.

1 reply

reacted to ajibawa-2023's post with 🔥 1 day ago

Post

3550

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

Cpp-Code-Large is a large-scale corpus of C++ source code comprising more than 5 million lines of C++ code. The dataset is designed to support research in large language model (LLM) pretraining, code intelligence, software engineering automation, and static program analysis for the C++ ecosystem.

By providing a high-volume, language-specific corpus, Cpp-Code-Large enables systematic experimentation in C++-focused model training, domain adaptation, and downstream code understanding tasks.

Cpp-Code-Large addresses the need for a dedicated C++-only dataset at substantial scale, enabling focused research across systems programming, performance-critical applications, embedded systems, game engines, and large-scale native software projects.

3 replies

reacted to NJX-njx's post with 🔥 1 day ago

Post

879

I feel that I have become more and more obsessed with studying some "primitive" CLI operations recently.

Compared to the so-called MCP and Skill, enabling AI to understand and use CLI is actually more feasible, explainable, and powerful in terms of code.

I recently deployed a website for my OpenSoul on Vercel. In the past, I might have needed to spend a lot of cognitive effort or time to understand how to operate on the Vercel page, and I would have had to spend a great deal of time reading documents (smarter people might directly feed the documents to AI and let AI summarize feasible and reliable steps).

But in fact, after ChatGPT told me that Vercel actually has a CLI, I directly asked my Copilot in VS Code to download this command line, clearly stated my needs, and it quickly solved everything else. The only thing I actually needed to do was log in to Vercel and create a key.

This suddenly reminds me of a blog post I read earlier that interviewed the father of Claude Code. The reason why Claude Code did not develop front-end pages and the like is precisely because he believes that we should focus most of our energy on the most meaningful interaction logic.

So, in an era where AI capabilities are becoming increasingly strong, perhaps what we really need is to pick up those tools that we used with the sole goal of achieving functionality when computing power was tight. What do you think?

1 reply

reacted to etemiz's post with 👍 1 day ago

Post

236

Published a new checkpoint for Ostrich 32B
etemiz/Ostrich-32B-Qwen3-260303-GGUF

Started fine tuning Qwen 3.5 27B. Soon high density intelligence meets human alignment!

John Smith PRO

AI & ML interests

Recent Activity

Organizations

John6666's activity