Here's what we got: • Grok 3: Digs into academic papers, tech blogs, and corporate research reports. • Perplexity: Covers mainstream breakthroughs but may lack depth in cutting-edge theoretical advancements.
GENERATIVE AI
-
5 Essential ChatGPT Prompts for HR Professionals
By
–
5 Essential ChatGPT Prompts Every HR Professional Should Know As #artificialintelligence reshapes the #workplace, #HR professionals and recruiters are discovering powerful ways to leverage #ChatGPT for everything from crafting compelling #job descriptions to conducting thorough
-
Weird bug: AI ignores pre-Python guess despite no yapping
By
–
My first thought was this depends entirely on “no yapping” but I was able to reproduce (in 1 of 5 attempts) with simpler instructions and no request for brevity. Weird bug — you’d think flatly ignoring the pre-Python guess in context would be well tuned by now.
-
Distributed Training Playbook: LLM Infrastructure Fundamentals and Benchmarks
By
–
Excellent playbook on the intricacies of distributed training. Covers basics of LLM training infra from the very ground-up, practicals/code, and efficiency benchmarks grounded on 4100 distributed experiments. Topics covered include:
– Data parallelism
– Tensor parallelism
– More -
Goodside’s post on LLM design imitation and sycophancy
By
–
A lot is imitation of design choices IMO. OpenAI’s ChatML was once new, now it’s like oxygen. At launch, Claude was much more “self-aware” than ChatGPT, which didn’t know its own name, but now they all act like that. Also see Anthro’s work on base LLM sycophancy etc.
-
Psychology Research Flaws and Token Prediction Analogy
By
–
this is a great example of how obviously silly psychology research is a lot of the time should you treat unwanted thoughts as irrelevant or dig in to them? depends highly on context a similar analogous question is: “what’s the next token in a sequence?” it’d be obviously
-
Grok 3’s fresh writing novelty effect versus GPT-4
By
–
I notice this too — Grok 3's writing is very fresh, and I'll be using it alongside Claude for inspiration. But I also think writing quality is mostly novelty effect. I was shaken to my core the first time I saw GPT-4 write a love letter. It goes away.
-
Unhinged LLM fallback for content moderation annotation
By
–
Working on self-contained ones as separate posts but a general one: I work with content moderation datasets. I often send big CSV of offensive filth to an LLM to annotate and get back refusals or quasi-refusals. Any competent "unhinged" fallback LLM saves me work.
-
New AI model matches 4o and Claude in writing without refusals
By
–
Nothing groundbreaking on capabilities so far, it's more that it's at least in the same league as 4o or Claude at writing without subject-related refusals. An Onion-style Buzzfeed listicle about weapons you can make in a toilet stall may not be funny, but it's not a crime:
