Your curated collection of saved posts and media
Revisiting @sarkodieโs stellar debut album, โSarkologyโ ๐ฅ https://t.co/JRBw6m5Oyo
โIf you are driven by fear, anger or prideโ Nature will force you to compete. If you are guided by courage, awareness, tranquility and peaceโฆ nature will serve you.โ ~ A. Ray, https://t.co/5H27c6LKSr
ใใใผใฑใผใใซๅญใใใใใใผใฑใผใใซๅญ OL Ver.ใใๅๅๅถไฝ๏ผๅ ซๅทป #wf2016s #ใใคใใฃใ https://t.co/yQIEmir5Ph
for youโ ๐ฅฐ Cr. Hello_asian #เนเธเนเธฒเนเธเนเธกเธเนเธญเธ #Fluke_Natouch https://t.co/nY8YvUA2UJ
เธเนเธฒเธงเนเธเนเธเธกเธฒเธเธฑเธเธเธ ๐ฆ #เนเธเนเธฒเนเธเนเธกเธเนเธญเธ #Fluke_Natouch https://t.co/LaBbA8TqDT
เธขเธญเธกเนเธเนเธเธเธฒเธชเนเธเธญ ๐ณ #เนเธเนเธฒเนเธเนเธกเธเนเธญเธ #Fluke_Natouch https://t.co/iDeKPoK1Kw
Iโve been thinking about the โfine-tuningโ question in cosmology. Many physical constants (gravity, the cosmological constant, particle masses) seem to sit in extremely narrow ranges that allow stars, galaxies, and life to exist. Even small changes could make a lifeless universe. Some people interpret this as a hint of intentional design. Others argue it could be explained by the multiverse (we simply live in the universe that works), or by deeper laws of physics we havenโt discovered yet. Iโm not claiming a final answer โ I just find the question fascinating: What do you think fine-tuning actually tells us?
ใใฉในใ2ๆฅใ ใใคใใฃใใใใใใใผใฑใผใใซๅญ ใใธใๅจOL ver.ใ๏ผใใใใๅๆณจ็ท ใใงใ๏ผ(`๏ฝฅฯ๏ฝฅยด)ใ https://t.co/oXuavWaOj7 https://t.co/Iit0vZfvUp
ใ 12ๆ19ๆฅ(ๆฐด) ๅๆณจ็ท ๅ ใ momiๆฐใฎๆใ็ฝใใ็พๅฐๅฅณใไธๆตท ่ตใใใญใฑใใใใผใคใๅฟ ๅฎใซ็ซไฝๅ๏ผ ใใใฃใฎไบคๆใซใใใๆฐด็ใฎ้ฃใ่พผใฟ่กจ็พใจ่งฃๆพ็ถๆ ใฎไธกๆนใๅ็พๅฏ่ฝใซใใใใพใใใ https://t.co/nRqEPywzKj https://t.co/ulxa5KmFVw
ใๅณถ็ฐใใใซใใใใใคใฉในใใใใผใฉใ ๅๅๅถไฝ๏ผใขใใฉ #wf2019w #ใใคใใฃใ https://t.co/EsfqPyHxxP
โThose who know how to maintain silence, knows how to maintain everything.โ ~>N.Namdeo https://t.co/YM1tmM35dm
๐จ NEW DIGITAL COVER ALERT๐จ The NATIVE Presents: Sounds From ๐ฃ๐ฑ๐ฒ๐ผ Side featuring: Street Pop 3.0 ๐ณ๐ฌ Amapiano ๐ฟ๐ฆ Asakaa Drill ๐ฌ๐ญ FULL STORY: https://t.co/Ka8lhCfueu https://t.co/9ETnVgdVJd
GREAT PHOTO OF >ADAM BEACH< HAVE A BLESSED WEEKEND BROTHER https://t.co/aSGWjCReBM
Found old @modal swag. Smells good! https://t.co/Ld3VV0RzRI
Looking for a festive Yule log to brighten up your terminal? Youโll love @leereillyโs GitHub CLI extension that gives you a cozy, animated Git log. ๐ฅ ๐ชต https://t.co/2oMEsTkEMP https://t.co/bxhmd254GE
Weโre releasing Bloom, an open-source tool for generating behavioral misalignment evals for frontier AI models. Bloom lets researchers specify a behavior and then quantify its frequency and severity across automatically generated scenarios. Learn more: https://t.co/TwKstpLSy3
โLogan was terrified for Jakeโ โ and honestlyโฆ you can feel the tension in his face. What do you think โ genuine fear, or just a bad freeze-frame/angle? ๐ฅ๐ @LoganPaul @jakepaul https://t.co/Hku9WLy1GC
Nana conference Attacca conference https://t.co/pcGv8RkuQP
RAG systems struggle with multi-hop reasoning. In most cases, the problem isn't the LLMs. It's the retrieval system. Standard RAG treats each piece of evidence as equally reliable, ignoring how documents connect to each other. Why is this a problem? When questions require reasoning across multiple sources, single-shot retrieval often misses "bridge" documents whose entities aren't mentioned in the original query. Iterative retrieval helps, but it introduces new issues: LLM-guided graph traversal can hallucinate or become stuck on partial reasoning from previous steps. This new research introduces SA-RAG, a framework that applies spreading activation, a mechanism from cognitive psychology, to knowledge-graph-based retrieval. How does it work? Instead of relying on the LLM to decide which documents to fetch next, activation propagates automatically through a knowledge graph. Starting from entities matched to the query, activation spreads outward through weighted connections, with strength diminishing over distance. Documents linked to highly activated entities get retrieved. The system builds a hybrid structure during indexing. An LLM extracts entities and relationships from text chunks, creating a knowledge graph where documents connect to entities through "describes" links. At query time, seed entities are identified by embedding similarity, then activation flows through the graph in a breadth-first manner. On MuSiQue, SA-RAG alone achieves 67% answer correctness with phi4, outperforming naive RAG at 45% and CoT-based iterative retrieval at 55%. When combined with chain-of-thought iterative retrieval, it reaches 74% on MuSiQue and 87% on 2WikiMultiHopQA. This system demonstrates a 25% to 39% absolute improvement over naive RAG across benchmarks. Notably, these results come from small, open-weight models like phi4 and gemma3, which require no fine-tuning. Spreading activation captures associative relevance rather than surface-level similarity. The method works as a plug-and-play module, boosting any training-free RAG pipeline without architectural changes. Paper: https://t.co/jLZLkacDAX Learn to build effective RAG and AI agents in our academy: https://t.co/zQXQt0PMbG
Check out the other skill examples in the repo. https://t.co/Oj3Oh5kJ9v
@gr00vyfairy You can also use it to create what I think is the best OG image ever https://t.co/0D8Hr798NW
Action produces information. https://t.co/MMP7wfuWCw
Action produces information. https://t.co/MMP7wfuWCw
Skills is now officially supported in Codex. There is a neat built-in skill for planning. This is the best way to pull in the right context at the right time. Also, a great way to build highly specialized skills for your coding agents. https://t.co/fviSC12aci
Me and the homies. https://t.co/pU9i20ako1
What are you buying? https://t.co/UsipE27Stv
This is wild. A real-time webcam demo using SmolVLM from @huggingface and llama.cpp! ๐คฏ Running fully local on a MacBook M3. https://t.co/BQ1HyP7RoC
I really didn't expect another major open-weight LLM release this December, but here we go: NVIDIA released their new Nemotron 3 series this week. It comes in 3 sizes: 1. Nano (30B-A3B), 2. Super (100B), 3. and Ultra (500B). Architecture-wise, the models are a Mixture-of-Experts (MoE) Mamba-Transformer hybrid architecture. As of this morning (Dec 19), only the Nano model has been released as an open-weight model, so this post will focus on that one (shown in my drawing below). Nemotron 3 Nano (30B-A3B) is a 52-layer hybrid Mamba-Transformer model that interleaves Mamba-2 sequence-modeling blocks with sparse Mixture-of-Experts (MoE) feed-forward layers, and uses self-attention only in a small subset of layers. Thereโs a lot going on in the figure above, but in short, the architecture is organized into 13 macro blocks with repeated Mamba-2 โ MoE sub-blocks, plus a few Grouped-Query Attention layers. In total, if we multiply the macro- and sub-blocks, there are 52 layers in this architecture. Regarding the MoE modules, each MoE layer contains 128 experts but activates only 1 shared and 6 routed experts per token. The Mamba-2 layers would take a whole article itself to explain (perhaps a topic for another time). But for now, conceptually, you can think of them as similar to the Gated DeltaNet approach that Qwen3-Next and Kimi-Linear use, which I covered in my Beyond Standard LLMs article. The similarity between Gated DeltaNet and Mamba-2 layers is that both replace standard attention with a gated-state-space update. The idea behind this state-space-style module is that it maintains a running hidden state and mixes new inputs via learned gates. In contrast to attention, it scales linearly instead of quadratically with the input sequence length. Whatโs actually quite exciting about this architecture is its really good performance compared to pure transformer architectures of similar size (like Qwen3-30B-A3B-Thinking-2507 and GPT-OSS-20B-A4B), while achieving much higher tokens-per-second throughput. Overall, this is an interesting direction, even more extreme than Qwen3-Next and Kimi-Linear in its use of only a few attention layers. However, one of the strengths of the transformer architecture is its performance at a (really) large scale. I am curious to see how the larger Nemotron 3 Super and especially Ultra will compare to the likes of DeepSeek V3.2.
Just updated the Big LLM Architecture Comparison article... ...it grew quite a bit since the initial version in July 2025, more than doubled! https://t.co/oEt8XzNxik https://t.co/RZuwp6ZUaF
https://t.co/gBgot9C1ZV
https://t.co/1VkhHOw1X6
https://t.co/gBgot9C1ZV
Believe so brightly that everyone sees the beauty in believing. ~ Native American ๐ชถโจ https://t.co/PLFFDE0g86
Native Beauty ๐นโค๏ธโ๐ฅโค๏ธโ๐ฅ๐น๐น If you're a Native beauty fan of mine can I get a bigโฆ.YESS !!! I love you Allโค๏ธ https://t.co/gG4VDMMcWq
