Pedro Lima
425 posts
@pvl
@cognitiva, machine learning, generative AI. Kaggle master.
Simply adding Gaussian noise to LLMs (one step—no iterations, no learning rate, no gradients) and ensembling them can achieve performance comparable to or even better than standard GRPO/PPO on math reasoning, coding, writing, and chemistry tasks. We call this algorithm RandOpt. To verify that this is not limited to specific models, we tested it on Qwen, Llama, OLMo3, and VLMs. What's behind this? We find that in the Gaussian search neighborhood around pretrained LLMs, diverse task experts are densely distributed — a regime we term Neural Thickets. Paper: arxiv.org/pdf/2603.12228 Code: github.com/sunrainyg/Rand… Website: thickets.mit.edu
Honestly: across the board, Portugal has gotten significantly WORSE since we started making investments in the country. If the trend continues, we’ll stop investing. And if you’re considering it as a tech firm, you’d be crazy to without some hard reassurances from the government.
Seems that Qwen2-VL is not good enough to appear on Llava o1 paper.
Coding interviews should be using ai. If you disagree you literally don’t understand what ai is for Sure you can use it to fill gaps & find easy answers More importantly you can use it to accelerate and do much bigger things Failing to select for this skill will ruin tech co’s
I asked, “What company is worth less than $10 billion today but you think could be worth $500+ billion in a few decades?” I received 710 answers Here are 20 top stocks that have 50x+ potential:
