👋 Open to Work

CK

Chris4K

27 18 1154

zX14_7
ckal

AI & ML interests

Phonetics, Alpha-Beta Pruning, Bayes, Go the game, NLP, Bots

Recent Activity

reacted to theirpost with 🔥 about 6 hours ago

Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four? Partly. I trained a GF(2⁸) integer model (8.4M params, CPU-only) on 320k games labelled by a negamax search teacher. What it learned: • agrees with the teacher on 59.6% of training / 41.0% of board-disjoint holdout positions • beats a random player 85.2% of the time What it didn't: • loses to a 1-ply "win now / block now" rule (wins 37.2%) • loses to 2-ply lookahead (wins 24.3%) Agreeing with a teacher is not the same as playing well, and this release measures both. Caveat up front: the float32 read-out is 99.6% of the parameters. The trunk is float-free; the full model isn't yet. The repo also documents six bugs that each produced a plausible but wrong number. In one, 41% of the supervision was silently all-zero. 🎮 Play: https://huggingface.co/spaces/Chris4K/connect-four-gf256 (Space) 🧠 Model: https://huggingface.co/Chris4K/connect-four-gf256 📦 Dataset: https://huggingface.co/Chris4K/connect-four-gf256 320k games + per-ply targets, MIT 💻 Code + paper: https://github.com/Ckal/connect-four-gf256 Feedback welcome, especially: which baseline would convince you?

reacted to theirpost with 😎 1 day ago

posted an update 1 day ago

View all activity

Organizations

reacted to their post with 🔥 about 6 hours ago

Post

1,654

Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four?

Partly. I trained a GF(2⁸) integer model (8.4M params, CPU-only) on 320k games labelled by a negamax search teacher.

What it learned:
• agrees with the teacher on 59.6% of training / 41.0% of board-disjoint holdout positions
• beats a random player 85.2% of the time

What it didn't:
• loses to a 1-ply "win now / block now" rule (wins 37.2%)
• loses to 2-ply lookahead (wins 24.3%)

Agreeing with a teacher is not the same as playing well, and this release measures both.

Caveat up front: the float32 read-out is 99.6% of the parameters. The trunk is float-free; the full model isn't yet.

The repo also documents six bugs that each produced a plausible but wrong number. In one, 41% of the supervision was silently all-zero.

🎮 Play: Chris4K/connect-four-gf256 (Space)
🧠 Model: Chris4K/connect-four-gf256
📦 Dataset: Chris4K/connect-four-gf256 320k games + per-ply targets, MIT
💻 Code + paper: https://github.com/Ckal/connect-four-gf256

Feedback welcome, especially: which baseline would convince you?

1 reply

reacted to their post with 😎 1 day ago

Post

1,654

Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four?

Partly. I trained a GF(2⁸) integer model (8.4M params, CPU-only) on 320k games labelled by a negamax search teacher.

What it learned:
• agrees with the teacher on 59.6% of training / 41.0% of board-disjoint holdout positions
• beats a random player 85.2% of the time

What it didn't:
• loses to a 1-ply "win now / block now" rule (wins 37.2%)
• loses to 2-ply lookahead (wins 24.3%)

Agreeing with a teacher is not the same as playing well, and this release measures both.

Caveat up front: the float32 read-out is 99.6% of the parameters. The trunk is float-free; the full model isn't yet.

The repo also documents six bugs that each produced a plausible but wrong number. In one, 41% of the supervision was silently all-zero.

🎮 Play: Chris4K/connect-four-gf256 (Space)
🧠 Model: Chris4K/connect-four-gf256
📦 Dataset: Chris4K/connect-four-gf256 320k games + per-ply targets, MIT
💻 Code + paper: https://github.com/Ckal/connect-four-gf256

Feedback welcome, especially: which baseline would convince you?

1 reply

posted an update 1 day ago

Post

1,654

Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four?

Partly. I trained a GF(2⁸) integer model (8.4M params, CPU-only) on 320k games labelled by a negamax search teacher.

What it learned:
• agrees with the teacher on 59.6% of training / 41.0% of board-disjoint holdout positions
• beats a random player 85.2% of the time

What it didn't:
• loses to a 1-ply "win now / block now" rule (wins 37.2%)
• loses to 2-ply lookahead (wins 24.3%)

Agreeing with a teacher is not the same as playing well, and this release measures both.

Caveat up front: the float32 read-out is 99.6% of the parameters. The trunk is float-free; the full model isn't yet.

The repo also documents six bugs that each produced a plausible but wrong number. In one, 41% of the supervision was silently all-zero.

🎮 Play: Chris4K/connect-four-gf256 (Space)
🧠 Model: Chris4K/connect-four-gf256
📦 Dataset: Chris4K/connect-four-gf256 320k games + per-ply targets, MIT
💻 Code + paper: https://github.com/Ckal/connect-four-gf256

Feedback welcome, especially: which baseline would convince you?

1 reply

reacted to azettl's post with 👍 11 months ago

Post

735

After building Consilium, my multi-AI expert consensus platform for the Gradio Agents and MCP Hackathon, Deborah Dahl introduced me to the Open Floor Protocol.

This protocol provides a standardized JSON message format for communication between conversational agents and human users across different platforms.

𝗞𝗲𝘆 𝗶𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝗼𝗻 𝗽𝗮𝘁𝘁𝗲𝗿𝗻𝘀:
✅ 𝗗𝗲𝗹𝗲𝗴𝗮𝘁𝗶𝗼𝗻 - transferring control between agents
✅ 𝗖𝗵𝗮𝗻𝗻𝗲𝗹𝗶𝗻𝗴 - passing messages without modification
✅ 𝗠𝗲𝗱𝗶𝗮𝘁𝗶𝗼𝗻 - coordinating behind the scenes
✅ 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 - multiple agents collaborating

I am already working on a version of Consilium where you can add any Open Floor-compliant agents 🚀.

➡️ 𝗖𝗵𝗲𝗰𝗸 𝗼𝘂𝘁 𝘁𝗵𝗲 𝗢𝗙𝗣 𝗱𝗲𝗺𝗼 𝗳𝗿𝗼𝗺 𝗘𝗺𝗺𝗲𝘁𝘁: https://ejtalk.com/Pages/ann2.html
Also see the specification on GitHub https://github.com/open-voice-interoperability/openfloor-docs.

reacted to lysandre's post with ❤️ over 1 year ago

Post

8,932

SmolVLM-2 and SigLIP-2 are now part of transformers in dedicated releases!

They're added on top of the v4.49.0 release, and can be installed from the following tags: v4.49.0-SmolVLM-2 and v4.49.0-SigLIP-2.

This marks a new beginning for the release process of transformers. For the past five years, we've been doing monthly releases featuring many models (v4.49.0, the latest release, features 9 new architectures).

Starting with SmolVLM-2 & SigLIP2, we'll now additionally release tags supporting new models on a stable branch. These models are therefore directly available for use by installing from the tag itself. These tags will continue to be updated with fixes applied to these models.

Going forward, continue expecting software releases following semantic versioning: v4.50.0 will have ~10 new architectures compared to v4.49.0, as well as a myriad of new features, improvements and bug fixes. Accompanying these software releases, we'll release tags offering brand new models as fast as possible, to make them accessible to all immediately.

1 reply

reacted to singhsidhukuldeep's post with 👍👀 over 1 year ago

Post

2,052

Exciting breakthrough in AI: AirRAG - A Novel Approach to Retrieval Augmented Generation!

Researchers from Alibaba Cloud have developed a groundbreaking framework that significantly improves how AI systems reason and retrieve information. AirRAG introduces five fundamental reasoning actions that work together to create more accurate and comprehensive responses.

>> Key Technical Innovations:
- Implements Monte Carlo Tree Search (MCTS) for exploring diverse reasoning paths
- Utilizes five core actions: System Analysis, Direct Answer, Retrieval-Answer, Query Transformation, and Summary-Answer
- Features self-consistency verification and process-supervised reward modeling
- Achieves superior performance across complex QA datasets like HotpotQA, MuSiQue, and 2WikiMultiHopQA

>> Under the Hood:
The system expands solution spaces through tree-based search, allowing for multiple reasoning paths to be explored simultaneously. The framework implements computationally optimal strategies, applying more resources to key actions while maintaining efficiency.

>> Results Speak Volumes:
- Outperforms existing RAG methods by over 10% on average
- Shows remarkable scalability with increasing inference computation
- Demonstrates exceptional flexibility in integrating with other advanced technologies

This research represents a significant step forward in making AI systems more capable of complex reasoning tasks. The team's innovative approach combines human-like reasoning with advanced computational techniques, setting new benchmarks in the field.

reacted to andrewrreed's post with 👍 almost 2 years ago

Post

1,118

Trace LLM calls with Arize AI's Phoenix observability dashboards on Hugging Face Spaces! 🚀

✨ I just added a new recipe to the Open-Source AI Cookbook that shows you how to:
1️⃣ Deploy Phoenix on HF Spaces with persistent storage in a few clicks
2️⃣ Configure LLM tracing with the 𝗦𝗲𝗿𝘃𝗲𝗿𝗹𝗲𝘀𝘀 𝗜𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗔𝗣𝗜
3️⃣ Observe multi-agent application runs with the CrewAI integration

𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗶𝘀 𝗰𝗿𝘂𝗰𝗶𝗮𝗹 for building robust LLM apps.

Phoenix makes it easy to visualize trace data, evaluate performance, and track down issues. Give it a try!

🔗 Cookbook recipe: https://huggingface.co/learn/cookbook/en/phoenix_observability_on_hf_spaces
🔗 Phoenix docs: https://docs.arize.com/phoenix

reacted to averoo's post with 👀 almost 2 years ago

Post

1,060

Just updated the cozy HF Daily Papers review page.

➕ Affiliations extraction (filters are coming soon)
➕ Redesign
➕ Top by month page
🔄 Syncing every 1 hour

👉 https://hfday.ru

Your feedback is appreciated.

2 replies

reacted to ahmed-masry's post with 👀 about 2 years ago

Post

1,588

🚀 Introducing ColFlor: An Efficient, OCR-Free Vision-Language Document Retrieval Model 🌟

Earlier this year, ColPali revolutionized document retrieval by eliminating the need for error-prone OCR pipelines. Instead, it directly processes the document images. However, with its 3 billion parameters, ColPali is computationally heavy for large-scale applications.

That’s where ColFlor comes in—a smaller, faster alternative! 🎉 At 17x smaller than ColPali, ColFlor offers a more efficient, OCR-free document retrieval solution, making it ideal for users with limited computing resources (GPU Poor). 💡

Key Highlights:
🧠 174M parameters (vs. 3B for ColPali)
⚡ 9.8x faster query encoding, 5.25x faster image encoding
📉 Only 1.8% performance drop on text-rich English documents

Check out the full blog post for more insights on modeling, training, and evaluations across various document retrieval tasks! 🚀
Also, feel free to try our demo on huggingface 🤗

🔗 Resources:
📄 Blog post: https://huggingface.co/blog/ahmed-masry/colflor
🧠 Model: ahmed-masry/ColFlor
💻 Demo: ahmed-masry/ColFlor-Demo
🏋️‍♂️ Training code: https://github.com/AhmedMasryKU/colflor
📊 Evaluation code: https://github.com/AhmedMasryKU/vidore-benchmark-colflor

reacted to celinah's post with ❤️ about 2 years ago

Post

1,293

📣 𝚑𝚞𝚐𝚐𝚒𝚗𝚐𝚏𝚊𝚌𝚎_𝚑𝚞𝚋 v0.26.0 is out with some new features and improvements!

✨ 𝗧𝗼𝗽 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀:
- 🔐 Multiple access tokens support: Easily manage multiple access tokens with new CLI commands. Perfect for handling multiple tokens with specific permissions in production or when collaborating with external teams.
- 🖼️ Conversational VLMs inference is now supported with InferenceClient's chat completion!
- 📄 Daily Papers API: Seamlessly search and retrieve detailed paper information from the Hub!

We’ve also introduced multiple bug fixes and quality-of-life improvements - thanks to the awesome contributions from our community! 🤗

Check out the release notes here: Wauplin/huggingface_hub#9

and you can try it out now 👇

pip install huggingface_hub==0.26.0

reacted to vikhyatk's post with ❤️ over 2 years ago

Post

Just released moondream2 - a small 1.8B parameter vision language model. Now fully open source (Apache 2.0) so you can use it without restrictions on commercial use!

vikhyatk/moondream2

8 replies

CK

AI & ML interests

Recent Activity

Organizations

Chris4K's activity