43 stories
LFM2.5-24B-A2B is a 24B-parameter reasoning model with just 2B active parameters, built for fast, private agents on-device. It thinks before it answers, calls tools, and works through multi-step tasks inside popular agent harnesses.
You can teach an open-weight LLM to text like a specific persona: their slang, their reply length, their humor, even how they write in a roleplay scene, creative writing exercise, or chatbot customer interactions.
Explore 10 Hugging Face models used across text, vision, audio, and retrieval in 2026.
From demystifying the online softmax trick to hitting a 41.6x speedup on an NVIDIA RTX 4050 GPU — a beginner’s hands-on journey into hardware-aware deep learning.
Article URL: https://nishtahir.com/build-your-own-decision-model/ Comments URL: https://news.ycombinator.com/item?id=50037949 Points: 266 # Comments: 61
A one-line formula for how often an LLM's top token flips, and why identical completions fall apart after about a hundred tokens The post Why Temperature 0 Isn't Deterministic appeared first on Towards Data Science.
If you’re more interested in asking why a model works than just using it off the shelf, a career as an AI research scientist might be the right fit for you.
This is the final post in a four-part series about post-training. If you missed them, check out part 1, part 2, and part 3. Time to get your hands dirty! I’ll take you through implementing the key pieces of the classic ChatGPT pipeline: SFT, then reward model training, then PPO.
Learn how vLLM achieves higher throughput than Hugging Face Transformers by using PagedAttention to eliminate memory waste, boost inference.
How OpenAI stripped down the original encoder-decoder architecture — and why removing half of it made language models better, not worse.Continue reading on Towards AI »
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training. - Release Release v5.19.0 · huggingface/transformers
The director and producer of the 2007 original will reunite to reignite the Hasbro franchise.
Embeddings, Masked Language Model Tasks, Attention Mechanisms, and Feed-Forward Networks Explained Through Equations, Matrix…
This is the full series of Deepseek-V3 technical report, where i explain all the technical details in simpler words with code…Continue reading on Towards AI »
How to run Naive-N0.5-Flash locally: hardware needs, FP8 quantization, context window, and setup steps for this 309B MoE coding model.
Naive-N0.5-Flash packs 309B params, 15.5B active, native 1M context via sparse attention, and MIT-licensed weights aimed at coding and AI R&D.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training. - Release Release 5.18.0 · huggingface/transformers
This is a summary of, and insights into, what I found digging into the recently-released Superwhisper S1 family of voice-to-text models.
Showing 18 of 43 stories
Load more storiesGet the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
