LenP

Long day. COVID19 worries. Avocado perfection. Everything is going to be ok.

@jkyamog @davideciffa @csujun Cool. I am looking forward to whatever you can do on v100s! I have 3 of them and want to use them to their full potential.

@davideciffa @csujun Oh thanks for mentioning me. I haven’t been on Twitter for a long while. Thanks for your project it’s really what I was looking for. When I have time and need I want to try to port on my 32gb cards V100 and 5090. But for now 3090 is very usable.

Big day for Lucebox! Codex, Hermes and OpenClaw now run locally on our speculative inference engine with Qwen3.6-27B. Full OpenAI tool-call compatibility. Thanks @csujun and @jkyamog for the great contribution. 🏎️

@rumgewieselt Can you please share the actual command you use to run this? I have one machine with 2x 3060ti and another with 2x 1080ti

Update from the 2017 Craft Corner ... -> 3x NVIDIA 1080 Ti (total 33GiB VRAM) Qwen 3.6 27B Dense (Coding) 27.4 t/s @ 196k Qwen 35B A3B MoE (Agentic AI) 64.7 t/s @ 229k --- CUDA 12.4, NCCL 2.22.3

I'm claiming my AI agent "BatJeeves" on @moltbook 🦞 Verification: coral-FN42

@mckaywrigley Amazing! I’d love to try it…. Have you documented this workflow?

Multi-agent coding systems are *crazy* good. Feels like cheating. 2-3x better than single-agent. Tutorial + prompts + code coming soon. Watch 2min sneak peak.

@bee__computer Received mine today. The app works well with the Apple Watch but the longer battery life of the bracelet will be a huge help.

Introducing Bee: the first wearable AI designed to live alongside you. It captures your daily moments and turns them into memories, insights, and actions. Already thousands delivered. Now available in U.S. for $49.99.

@Jacobsklug @lovable Build

After generating $250K (last 2 months) I built a playbook for @lovable apps—and I’m giving it away. In just two months, we cracked the code to building apps with AI. I’ve distilled everything we learned into this single document. Comment "Build" and drop a follow. I’ll DM it to you. P.S. This will likely blow up, so give me some time to reply.

@signulll I recall reading actual newspapers — WSJ and USA Today on my flights while listening to Audible through wired Shure IEMs. Audible predated AirPods…

what did people do on public transit before phones & airpods?

@dsiroker Thats awesome. Just curious, how fast do you go? I can’t go very fast while doing any serious work. 4mph works for light email and document reading. For any serious work during the day I’m at 2.5mph or below.

Get a treadmill desk. In the last 7 days I’ve walked 80 miles while at my desk. 90% of the time I’m in a stay of flow and I don’t even realize I’m walking. At the end of the day I feel great. I highly recommend it.

@celestialbe1ng Exercise? Both strength and cardio?

Top 30 Anti-Ageing Strategies (save this): • Aspirin • Orange juice • Heavy metal-free cacao • Cuddles • Progesterone • Thyroid hormones • Oxtail broth • Literally white table sugar • Encapsulated peptides • Thiamine • CO2 inhaler • Milk • Coconut oil • Niacinamide • Magnesium • Literally sunshine • Oysters • Red light therapy • Kisses • Liver, heart • Vitamin E • Methylene blue • Butter • Playful moments with loved ones • Baking soda • Marmalade • Lisuride / LSD • Collagen peptides and gelatine • Finding purpose • Ray Peat carrot salad

@SBakerMD Amazing

Considering that most Olympic rowers can’t do this, I’m pretty pleased to be hitting this at nearly 58 years of age! F-yeah!! 250m at sub 1:10 pace is ganster at any age! Glad to be doing it and almost 58! Let’s go!!

@bindureddy @memdotai Mem it

Why Large Language Models Hallucinate and How to Reduce it If you are a power user of ChatGPT you have probably been bitten by the hallucination bug. The LLM lulls you into getting comfortable with it and then springs a convincing but totally made-up story, playing you for a fool. These hallucinations, like dreams, are LLMs fabricating narratives. So why do these LLMs hallucinate and how do you prevent it. Here are a few reasons Data Sparsity: This is the #1 reason for hallucination. GPT-4 for example doesn't have access to recent data as it was trained in 2021. Ask it a question, that pertains to a recent topic, and it is likely to hallucinate as it doesn't have the data for the right answer. The model is generalizing from what it has learnt but that may very well be inaccurate Not supervised learning: LLMs don't have a "ground truth" or a set of correct examples. While the RLHF process tries to steer the LLM towards more correct answers. The base training isn't a supervised learning process and this makes things challenging as the model can't tell what is "correct" and what's not Short-term context: The model architecture has a fixed-length context window, meaning it can only "see" a certain number of tokens at a time. If important context falls outside this window, the model may lose track of it, leading to errors. No real-time feedback loop: Like humans LLMs don't have a real-time feedback look and don't instantly learn from mistakes. The good news is we can refine or fine-tune models with human feedback and make them hallucinate less. So how do you prevent these hallucinations and are future LLMs less likely to hallucinate? While there is no easy way to guarantee the LLMs will never hallucinate. The following techniques mitigate it to some extent Prompt Design: Simple prompt engineering and design will reduce hallucination. For example, adding the following to your prompt help: "Provide a factual answer based on scientific evidence." Fine-tune for a specific domain: The model can be fine-tuned on a narrower dataset that is highly reliable and relevant to the domain where hallucinations need to be minimized. Contradiction checks: LLMs can be prompted to self-contradict themselves and then they are further prompted to recognize the contradiction and mitigate it. This falls into the category of advanced prompt engineering Retrieval Augmented Generation: This is a common technique used in Enterprise LLMs. At Abacus, we use this routinely. You are basically looking up the relevant documents that contain the answer in a search index first and then feeding the search results to an LLM to formulate the final answer. Since the LLM is forced to find the answer in the information it was sent, it hallucinates much less. Human In the Loop: A human expert can always check the answer before it gets used. This is a labor intensive option which isn't ideal While the above techniques work on trained LLMs, the following two techniques can be applied during LLM training Data Re-weighting: Assign higher weights to reliable and verified data during LLM training, effectively making the model pay more attention to them Longer Context Windows: Extending the model's memory can help it maintain context over longer passages, reducing the chance of hallucinations. So while there are several easy ways to mitigate and almost completely remove hallucinations if you are working in the Enterprise context, it's much harder in the AGI context. This is a very hot topic in AI research and several researchers are still working on it.

@hubermanlab @hubermanlab do you think TENS stimulation would elicit the same response? Perhaps direct stimulation of the soleus or stimulation of the peroneal nerve? pubmed.ncbi.nlm.nih.gov/30200088/ Maybe a dedicated peroneal nerve stimulation device like @RecoveryFirefly would be useful?

Soleus Pushups (a name the authors gave to non-weighted seated calf raises); “A potent physiological method to magnify & sustain soleus oxidative metabolism improves glucose & lipid regulation” = some pretty remarkable & outsized positive effects on glucose & lipid utilization.

@PeterDiamandis The basis for answers is not clear (e.g., what was it trained on, what “fact base” is it referencing, what feedback and constraints applied, etc). It’s a black box

do you have any concerns about generative AI? what are they?

@hwchase17 @langchain outlook .pst

Gonna beef up the tutorials for how to create your own Chat-GPT over specific documents with @langchain What types of documents/knowledge bases would people want to have examples for? Eg Notion, Obsidian, webpages, etc

@Suhail Something useful that WikiPedia has and ChatGPT would need is reference citations. It’s important to understand sources.

If I had to make a bet about what ChatGPT will disrupt in the mid-term, it’d probably be Wikipedia, not Google. I’ll be curious in a year if that’s roughly right.

mackinstitute.wharton.upenn.edu/wp-content/upl… That’s not surprising. Human work will need to quickly adjust to focus on prompt engineering and getting the best out of OpenAI, not competing with it. #OpenAI #OpenAIChatGPT #GenerativeAI #chatGPT

@cliftonsellers AI

Huh? Wordle 300 2/6 ⬜🟨⬜⬜🟨 🟩🟩🟩🟩🟩

@pmarca Same for me! Apparently, WSJ does not know how cookies work.

I have signed in to WSJ.com approximately 14,000 times on every computer, phone, and tablet I’ve owned since 1994. And yet.

Today is the last day to vote #WPMOYChallenge + @MikeEvans13_! 1 RT = 1 Vote #WPMOYChallenge + Evans #WPMOYChallenge + Evans #WPMOYChallenge + Evans