2 4 13

Dcas89 PRO

Dcas89

Dcas89

AI & ML interests

None yet

Recent Activity

reacted to ST-x-Tony's post with ❤️ about 13 hours ago

Hello everyone, We are excited to share that SKT-NRS is now live on Hugging Face. We’ve developed a Neural Reasoning System (NRS) designed to enhance the capabilities of foundation models — giving them stronger reasoning, improved performance, and more reliable outputs across a wide range of tasks. Our goal is to bring meaningful quality improvements to both new and existing models. You’ll start seeing boosted versions of various models released here soon, each refined with our NRS approach. **What to Expect* ❤️‍🩹 Regular releases of Neural Reasoning-enhanced models Clear focus on better reasoning and overall model quality Ongoing improvements based on community feedback If you’d like to stay updated, feel free to follow this space — we’ll be posting the first boosted models very soon. **Community Requests** Have a specific model you’d like us to work on? Looking for improvements on an existing model, or have any other requests? We’re happy to hear from you. Please share your suggestions here: ## Community Requests → https://huggingface.co/spaces/SKT-NRS/README/discussions/1 **Thank you for your support! We look forward to building better models together.**

reacted to owensong's post with 🔥 6 days ago

I just released Inflect-Nano-v1, an ultra-small 4.63 parameter text-to-speech model. The main idea is simple: instead of only making the acoustic model tiny and relying on a larger external vocoder, Inflect-Nano-v1 keeps the complete text-to-waveform stack under 5M parameters. Quick facts: - 4.63M total inference parameters - 3.46M acoustic model - 1.17M vocoder - 24 kHz audio - English-only - Single male voice - Runs locally with a simple PyTorch inference script Why I made it: Most modern TTS models are much larger, and even many “small TTS” projects depend on a separate vocoder. I wanted to see how far a complete tiny TTS stack could be pushed while still producing usable speech. It is not SOTA, and I am not trying to claim it competes with large TTS systems. The interesting part is the size-to-functionality ratio. What works: It can generate arbitrary English speech locally, and the model is small enough to be interesting for: - local voice assistants - embedded/edge experiments - browser or WASM-style TTS exploration - efficient inference research - tiny-model baselines Limitations: The quality is still limited. It can sound robotic, stumble on difficult unseen text, and the vocoder is still a clear bottleneck. Long or unusual prompts are less reliable. So I would frame this as a research/demo release, not a production TTS engine. I’d love feedback from people interested in: - tiny speech models - vocoders - local TTS - efficient inference - embedded speech synthesis - improving small-model generalization If people find it useful, I’m interested in putting more training budget into a stronger v2. Model page: https://huggingface.co/owensong/Inflect-Nano-v1

reacted to danielhanchen's post with 🔥 about 1 month ago

We’re excited to announce that Unsloth has joined the PyTorch Ecosystem! 🔥🦥 Unsloth is an open-source project that makes training & running models more accurate and faster with less compute. Our mission is to make local AI accessible to everyone. Thanks to all of you for making this possible! 💕 Blog: https://unsloth.ai/blog/pytorch GitHub: https://github.com/unslothai/unsloth

View all activity

Organizations

None yet

reacted to ST-x-Tony's post with ❤️ about 13 hours ago

Post

5892

Hello everyone,

We are excited to share that SKT-NRS is now live on Hugging Face.
We’ve developed a Neural Reasoning System (NRS) designed to enhance the capabilities of foundation models — giving them stronger reasoning, improved performance, and more reliable outputs across a wide range of tasks.

Our goal is to bring meaningful quality improvements to both new and existing models. You’ll start seeing boosted versions of various models released here soon, each refined with our NRS approach.

**What to Expect* ❤️‍🩹

Regular releases of Neural Reasoning-enhanced models
Clear focus on better reasoning and overall model quality
Ongoing improvements based on community feedback

If you’d like to stay updated, feel free to follow this space — we’ll be posting the first boosted models very soon.

**Community Requests**

Have a specific model you’d like us to work on? Looking for improvements on an existing model, or have any other requests?
We’re happy to hear from you. Please share your suggestions here:

## Community Requests → SKT-NRS/README#1

**Thank you for your support! We look forward to building better models together.**

5 replies

reacted to owensong's post with 🔥 6 days ago

Post

6401

I just released Inflect-Nano-v1, an ultra-small 4.63 parameter text-to-speech model.

The main idea is simple: instead of only making the acoustic model tiny and relying on a larger external vocoder, Inflect-Nano-v1 keeps the complete text-to-waveform stack under 5M parameters.

Quick facts:
- 4.63M total inference parameters
- 3.46M acoustic model
- 1.17M vocoder
- 24 kHz audio
- English-only
- Single male voice
- Runs locally with a simple PyTorch inference script

Why I made it:
Most modern TTS models are much larger, and even many “small TTS” projects depend on a separate vocoder. I wanted to see how far a complete tiny TTS stack could be pushed while still producing usable speech.

It is not SOTA, and I am not trying to claim it competes with large TTS systems. The interesting part is the size-to-functionality ratio.

What works:
It can generate arbitrary English speech locally, and the model is small enough to be interesting for:

- local voice assistants
- embedded/edge experiments
- browser or WASM-style TTS exploration
- efficient inference research
- tiny-model baselines

Limitations:
The quality is still limited. It can sound robotic, stumble on difficult unseen text, and the vocoder is still a clear bottleneck. Long or unusual prompts are less reliable.

So I would frame this as a research/demo release, not a production TTS engine.

I’d love feedback from people interested in:
- tiny speech models
- vocoders
- local TTS
- efficient inference
- embedded speech synthesis
- improving small-model generalization

If people find it useful, I’m interested in putting more training budget into a stronger v2.

Model page:
owensong/Inflect-Nano-v1

reacted to danielhanchen's post with 🔥 about 1 month ago

Post

5873

We’re excited to announce that Unsloth has joined the PyTorch Ecosystem! 🔥🦥

Unsloth is an open-source project that makes training & running models more accurate and faster with less compute. Our mission is to make local AI accessible to everyone. Thanks to all of you for making this possible! 💕

Blog: https://unsloth.ai/blog/pytorch
GitHub: https://github.com/unslothai/unsloth

2 replies

reacted to Parveshiiii's post with 🔥 3 months ago

Post

2961

Just did something I’ve been meaning to try for ages.

In only 3 hours, on 10 billion+ tokens, I trained a custom BPE + tiktoken-style tokenizer using my new library microtok — and it hits the same token efficiency as Qwen3.

Tokenizers have always felt like black magic to me. We drop them into every LLM project, but actually training one from scratch? That always seemed way too complicated.

Turns out it doesn’t have to be.

microtok makes the whole process stupidly simple — literally just 3 lines of code. No heavy setup, no GPU required. I built it on top of the Hugging Face tokenizers library so it stays clean, fast, and actually understandable.

If you’ve ever wanted to look under the hood and build your own optimized vocabulary instead of just copying someone else’s, this is the entry point you’ve been waiting for.

I wrote up the full story, threw in a ready-to-run Colab template, and dropped the trained tokenizer on Hugging Face.

Blog → https://parveshiiii.github.io/blogs/microtok/
Trained tokenizer → https://huggingface.co/Parveshiiii/microtok
GitHub repo → https://github.com/Parveshiiii/microtok

reacted to SeaWolf-AI's post with 🔥 4 months ago

Post

5096

ALL Bench — Global AI Model Unified Leaderboard

FINAL-Bench/all-bench-leaderboard

If you've ever tried to compare GPT-5.2 and Claude Opus 4.6 side by side, you've probably hit the same wall: the official Hugging Face leaderboard only tracks open-source models, so the most widely used AI systems simply aren't there. ALL Bench fixes that by bringing closed-source models, open-weight models, and — uniquely — all four teams under South Korea's national sovereign AI program into a single leaderboard. Thirty-one frontier models, one consistent scoring scale.
Scoring works differently here too. Most leaderboards skip benchmarks a model hasn't submitted, which lets models game their ranking by withholding results. ALL Bench treats every missing entry as zero and divides by ten, so there's no advantage in hiding your weak spots.
The ten core benchmarks span reasoning (GPQA Diamond, AIME 2025, HLE, ARC-AGI-2), coding (SWE-bench Verified, LiveCodeBench), and instruction-following (IFEval, BFCL). The standout is FINAL Bench — the world's only benchmark measuring whether a model can catch and correct its own mistakes. It reached rank five in global dataset popularity on Hugging Face in February 2026 and has been covered by Seoul Shinmun, Asia Economy, IT Chosun, and Behind.
Nine interactive charts let you explore everything from composite score rankings and a full heatmap to an open-vs-closed scatter plot. Operational metrics like context window, output speed, and pricing are included alongside benchmark scores.
All data is sourced from Artificial Analysis Intelligence Index v4.0, arXiv technical reports, Chatbot Arena ELO ratings, and the Korean Ministry of Science and ICT's official evaluation results. Updates monthly.

reacted to SeaWolf-AI's post with 🔥 4 months ago

Post

4327

FINAL Bench Released: The Real Bottleneck to AGI Is Self-Correction

We release FINAL Bench, the first benchmark for measuring functional metacognition in LLMs — the ability to detect and correct one's own reasoning errors. Every existing benchmark measures final-answer accuracy. None measures whether AI knows it is wrong.

Dataset: [FINAL-Bench/Metacognitive]( FINAL-Bench/Metacognitive) | 100 Tasks | 15 Domains | 8 TICOS Types | Apache 2.0

Leaderboard: FINAL-Bench/Leaderboard

Article: https://huggingface.co/blog/FINAL-Bench/metacognitive

Core Innovation

Our 5-axis rubric separates what no prior benchmark could: MA (Metacognitive Accuracy) — the ability to say "I might be wrong", and ER (Error Recovery) — the ability to actually fix it. This maps directly to the monitoring-control model of Nelson & Narens (1990) in cognitive psychology.

Three Findings Across 9 SOTA Models

We evaluated GPT-5.2, Claude Opus 4.6, Gemini 3 Pro, DeepSeek-V3.2, Kimi K2.5, and others across 100 expert-level tasks:

1. ER Dominance. 94.8% of MetaCog gain comes from Error Recovery alone. The bottleneck to AGI is not knowledge or reasoning — it is self-correction.

2. Declarative-Procedural Gap. All 9 models can verbalize uncertainty (MA = 0.694) but cannot act on it (ER = 0.302). They sound humble but fail to self-correct — the most dangerous AI safety profile.

3. Difficulty Effect. Harder tasks benefit dramatically more from metacognition (Pearson r = -0.777, p < 0.001).

from datasets import load_dataset
dataset = load_dataset("FINAL-Bench/Metacognitive", split="train")

Paper: FINAL Bench: Measuring Functional Metacognitive Reasoning in LLMs

FINAL Bench is the first tool to tell apart what AI truly knows from what it merely pretends to know.

6 replies

reacted to hassenhamdi's post with 🔥 5 months ago

Post

2075

Google published the paper. I shipped the code. 🚀

DeepMind just released PACEvolve (Progress-Aware Consistent Evolution), a massive overhaul of the AlphaEvolve framework. It solves the critical issues of "Context Pollution" and "Mode Collapse" that have historically crippled evolutionary coding agents.

But there was no public implementation. So I built one.

Introducing OpenPACEvolve: A fully open-source, production-grade implementation of the PACEvolve framework.

🛠 I engineered this framework solo, but I wasn't working alone. I orchestrated a custom coding agents powered by Claude Opus 4.5 as Engineer and Gemini Pro 3 Preview ensuring fiedelity and quallty.

By leveraging these SOTA models, I was able to translate complex theoretical research into functional, modular Python architecture in record time. This is what the future of AI engineering looks like: Human architectural oversight + AI velocity.

🧠 What OpenPACEvolve Solves: Unlike standard agents that get "stuck" in loops, this framework implements the paper's full recipe for long-horizon stability: ✅ Hierarchical Context Management (HCM): Bi-level pruning to keep the agent's memory clean. ✅ Momentum-Based Backtracking (MBB): Uses "power-law backtracking" to detect stagnation and force pivots. ✅ Self-Adaptive Crossover: Intelligent code-sharing between parallel "islands."

👨‍💻 This project is more than a repo; it's a demonstration of rapid research-to-production cycles using next-gen AI workflows.

📎 Link of the paper : https://arxiv.org/abs/2601.10657

The code is live. The agents are ready. Check out the repository below. 👇
https://github.com/hassenhamdi/OpenPACEvolve
Star the repo 🌟.

reacted to IlyasMoutawwakil's post with 🔥 5 months ago

Post

2477

After 2 months of refinement, I'm happy to announce that a lot of Transformers' modeling code is now significantly more torch-compile & export-friendly 🔥

Why it had to be done 👇
PyTorch's Dynamo compiler is increasingly becoming the default interoperability layer for ML systems. Anything that relies on torch.export or torch.compile, from model optimization to cross-framework integrations, benefits directly when models can be captured as a single dynamo-traced graph !

Transformers models are now easier to:
⚙️ Compile end-to-end with torch.compile backends
📦 Export reliably via torch.export and torch.onnx.export
🚀 Deploy to ONNX / ONNX Runtime, Intel Corporation's OpenVINO, NVIDIA AutoDeploy (TRT-LLM), AMD's Quark, Meta's Executorch and more hardware-specific runtimes.

This work aims at unblocking entire TorchDynamo-based toolchains that rely on exporting Transformers across runtimes and accelerators.

We are doubling down on Transformers commitment to be a first-class citizen of the PyTorch ecosystem, more exportable, more optimizable, and easier to deploy everywhere.

There are definitely some edge-cases that we still haven't addressed so don't hesitate to try compiling / exporting your favorite transformers and to open issues / PRs.

PR in the comments ! More updates coming coming soon !

1 reply

reacted to sergiopaniego's post with 🤗 7 months ago

Post

2306

ICYMI, transformers v5 is out!

Grab a coffee ☕ and go read the announcement blog https://huggingface.co/blog/transformers-v5

reacted to Jofthomas's post with 🔥 7 months ago

Post

4516

The new Mistral 3 models are here !

Today, we announce Mistral 3, the next generation of Mistral models. Mistral 3 includes three state-of-the-art small, dense models (14B, 8B, and 3B) and Mistral Large 3 – our most capable model to date – a sparse mixture-of-experts trained with 41B active and 675B total parameters.

All models are released under the Apache 2.0 license.

Ministrals :
https://huggingface.co/collections/mistralai/ministral-3

Mistral Large 3:
https://huggingface.co/collections/mistralai/mistral-large-3

2 replies

liked a dataset 8 months ago

minwoosun/CholecSeg8k

Updated Jan 25, 2024 • 215 • 6

reacted to Kseniase's post with 🔥 8 months ago

Post

11260

11 Fascinating new Policy Optimization techniques

Policy optimization (PO) algorithms are central to training AI models with preference-based feedback. In recent weeks, numerous new PO methods have emerged that build on or replace the popular PPO and GRPO, solving their issues. Here are 11 of them:

1. BAlanced Policy Optimization (BAPO) → BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping (2510.18927)
Dynamically adjusting the clipping bounds in PPO-style updates to balance positive and negative gradients and prevent entropy collapse

2. Training-Free GRPO → Training-Free Group Relative Policy Optimization (2510.08191)
Instead of using numeric rewards, it compares rollouts semantically to distill useful knowledge as a token prior, which is then applied during inference to guide the model’s behavior

3. Asymmetric Importance Sampling Policy Optimization (ASPO) → ASPO: Asymmetric Importance Sampling Policy Optimization (2510.06062)
Fixes imbalanced token weighting in LLM training. It flips the importance sampling ratios for positive tokens to correct over- and under-updates, and adds a soft dual-clipping step to keep gradients stable

4. In-Context Steered Policy Optimization (ICPO) → https://arxiv.org/abs/2510.26519
Uses a model’s own in-context learning ability to guide training with existing data. It combines Mixed-Policy GRPO with Implicit Expert Forcing to expand exploration and adds Expert Region Reject Sampling and Annealed Expert-Bonus Reward Shaping to ensure stability and balanced expert influence

5. Graph-Enhanced Policy Optimization (GEPO) → https://arxiv.org/abs/2510.26270
Builds a graph of an agent’s experiences to understand how different states connect, guide exploration and assign rewards more effectively

6. Information Gain-based Policy Optimization (IGPO) → Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn LLM Agents (2510.14967)
Uses the model’s own belief updates to create dense, informative feedback for smoother multi-turn learning

Read further below ⬇️
If you like this, also subscribe to the Turing post: https://www.turingpost.com/subscribe

2 replies

reacted to prithivMLmods's post with 👍 9 months ago

Post

4567

Try the Hugging Face Space demo for Logics-MLLM/Logics-Parsing, the latest multimodal VLM from the Logics Team at Alibaba Group. It enables end-to-end document parsing with precise content extraction in markdown format, and it also generates a clean HTML representation of the document while preserving its logical structure. 🤗🔥

Additionally, I’ve integrated one of my recent works — prithivMLmods/Gliese-OCR-7B-Post1.0 — which also excels at document comprehension.

⭐ Space / App : prithivMLmods/VLM-Parsing
📄 Technical Report by the Logics Team, Alibaba Group : Logics-Parsing Technical Report (2509.19760)
🖖 MM: VLM-Parsing: prithivMLmods/mm-vlm-parsing-68e33e52bfb9ae60b50602dc
⚡ Collections : prithivMLmods/multimodal-implementations-67c9982ea04b39f0608badb0

Other Pages:

➔ Multimodal VLMs - July'25 : prithivMLmods/multimodal-vlms-until-july25-688312e6b840e1e156f13027
➔ Multimodal VLMs - Aug'25 : prithivMLmods/multimodal-vlms-aug25-68a56aac39fe8084f3c168bd
➔ VL caption — < Sep 15 ’25 : prithivMLmods/vl-caption-sep-15-25-68c7f6d737985c63c13e2391

.
.
.
To know more about it, visit the app page or the respective model page!!

reacted to MonsterMMORPG's post with 🔥 9 months ago

Post

3353

Ovi - Generate Videos With Audio Like VEO 3 or SORA 2 - Run Locally - Open Source for Free

Download and install : https://www.patreon.com/posts/140393220

Quick demo tutorial : https://youtu.be/uE0QabiHmRw

Ovi: Twin Backbone Cross-Modal Fusion for Audio-Video Generation

Project page : https://aaxwaz.github.io/Ovi/

SECourses Ovi Pro Premium App Features

Full scale ultra advanced app for Ovi - an open source project that can generate videos from both text prompts and image + text prompts with real audio.

Project page is here : https://aaxwaz.github.io/Ovi/

I have developed an ultra advanced Gradio app and much better pipeline that fully supports block swapping

Now we can generate full quality videos with as low as 8.2 GB VRAM

Hopefully I will work on dynamic on load FP8_Scaled tomorrow to improve VRAM even further

So more VRAM optimizations will come hopefully tomorrow

Our implemented block swapping is the very best one out there - I took the approach from famous Kohya Musubi tuner

The 1-click installer will install into Python 3.10.11 venv and will auto download models as well so it is literally 1-click

My installer auto installs with Torch 2.8, CUDA 12.9, Flash Attention 2.8.3 and it supports literally all GPUs like RTX 3000 series, 4000 series, 5000 series, H100, B200, etc

All generations will be saved inside outputs folder and we support so many features like batch folder processing, number of generations, full preset save and load

This is a rush release (in less than a day) so there can be errors please let me know and I will hopefully improve the app

Look the examples to understand how to prompt the model that is extremely important

RTX 5090 can run it without any block swap with just cpu-offloading - really fast

2 replies

reacted to sergiopaniego's post with 🔥 9 months ago

Post

1414

This summer TRL leveled up for multimodal alignment 🌞

✅ New VLM alignment methods (MPO, GRPO, GSPO)
✅ Extended RLOO & Online DPO for VLMs
✅ Native SFT support
✅ Ready-to-use training scripts

🔗 https://huggingface.co/blog/trl-vlm-alignment

reacted to vikhyatk's post with 🔥 9 months ago

Post

6295

Just released a preview of Moondream 3! moondream/moondream3-preview

This is a 9B parameter, 2B active MoE VLM with state of the art visual reasoning capabilities.

More details in the release blog post: https://moondream.ai/blog/moondream-3-preview

3 replies

reacted to sergiopaniego's post with 🤗 11 months ago

Post

2938

So you can now SFT a model with hf jobs + TRL in ONE command lol 🏎️💨

Without worrying about infrastructure since it runs entirely on HF!

docs: https://huggingface.co/docs/huggingface_hub/main/en/guides/jobs
blog: https://huggingface.co/blog/hf-cli

reacted to prithivMLmods's post with 👍 11 months ago

Post

4416

On the verge of releasing Poseidon-Reasoning-5M, a dataset built to excel in general thought processes, mathematics, and science across a diverse mixture of domains, I’m also dropping the Gargantua-R1-Compact dataset, a collection of over six million high-quality reasoning QA pair traces. 🤗🚀

✦ Gargantua-R1-Compact : prithivMLmods/Gargantua-R1-Compact

from datasets import load_dataset

dataset = load_dataset("prithivMLmods/Gargantua-R1-Compact", split="train")

Additionally, I’m adding the mini version of Gargantua — the Gargantua-R1-Wee : prithivMLmods/Gargantua-R1-Wee

from datasets import load_dataset

dataset = load_dataset("prithivMLmods/Gargantua-R1-Wee", split="train")

The composition spans 73.93% core mathematical reasoning involving problems, proofs, and computational challenges, 12.11% across diverse scientific domains such as physics, chemistry, biology, and interdisciplinary topics, 11.35% in competitive coding covering algorithms and data structures, 1.37% in academic science focusing on research-level methodology, 0.95% in creative and analytical reasoning through logic puzzles and problem-solving tasks, 0.25% in specialized technical areas like MLOps, LLMs, diffusion models, and CUDA, and 0.06% involving data from graphs and charts converted into structured JSON formats. Designed with both rich contextual depth and formal structural clarity, Gargantua-R1-Compact is an optimal resource for advancing research in symbolic reasoning, interpretability, and high-precision question answering in mathematical domains.

✦ Collection : prithivMLmods/gargantua-r1-mod-6896bfd7834e82b89ad2b38b

To know more about it, visit the dataset card of the respective dataset. !!

reacted to codelion's post with 🔥 11 months ago

Post

4721

Released 17 production-ready adaptive text classifiers that learn from just 100 examples per class and continuously improve without retraining.

These models achieve 93% average accuracy across enterprise use cases like email routing, fraud detection, document classification, and support ticket categorization. Built on ModernBERT with prototype memory and elastic weight consolidation.

Key benefits: 90% cost reduction vs API solutions, 90-120ms local inference, dynamic class addition, and zero vendor lock-in.

All models available under adaptive-classifier organization. Install with pip install adaptive-classifier.

Full technical details: https://huggingface.co/blog/codelion/enterprise-ready-classifiers
Code: https://github.com/codelion/adaptive-classifier

2 replies

reacted to sergiopaniego's post with 🤗 11 months ago

Post

2763

We just released TRL v0.20 with major multimodal upgrades!

👁️ VLM support for GRPO (highly requested by the community!)
🎞️ New GSPO trainer (from @Qwen , released last week, VLM-ready)
🐙 New MPO trainer (multimodal by design, as in the paper)

📝 Full release notes here: https://github.com/huggingface/trl/releases/tag/v0.20.0

Dcas89 PRO

AI & ML interests

Recent Activity

Organizations

Dcas89's activity