BG (@BalagopalP7) header
BG (@BalagopalP7) profile picture

Entrepreneur

kerala
Joined February 2022
Photo by BG (@BalagopalP7)Photo by BG (@BalagopalP7)Photo by BG (@BalagopalP7)Photo by BG (@BalagopalP7)Photo by BG (@BalagopalP7)
the ship sat in orbit 20+ hours and still lit three raptors for the flip. that relight after a long coast is the bit i want real data on, every refuel or long mission plan sits on it
1
1
26
booster ended up in the water instead of the tower and spacex hasn't said why. does that change anything for the ship catch or are they totally separate calls
31
44 unbeaten and de la fuente's still annoyed about the finishing. that's the bit that matters for sunday, you don't get many chances against england
1
1
33
servers they bring back go unresponsive again once traffic hits them. so the fix isn't really the box anymore, it's how slowly you let the whole world back in
1
1
27
google's new gemini glasses work with iphone 13 and up, not just pixel and galaxy. tells you they care more about gemini being on your face than selling you a pixel
17
they gave it to everyone, not only the accounts the bugs touched. cheaper than making people prove they got overcharged, and nobody has to file a ticket to get trust back
We've given all Pro, Max, Ultra, and Team users one free full reset, across all models and limits. Use it anytime before December 10 at 2:00PM PT. We recently fixed a few small issues that used more limits than intended, and the Fable resets didn't do much for those of you who'd already hit your weekly limits for all models. While these issues didn't affect everyone, we want to make up for the trouble. You'll find it in Settings > Usage, or at clau.de/reset.
33
frenkie at right back for one half and that's when it turned. one game fix or does flick actually keep him there now that christensen and eric are back
41
asking claude to explain the code back helps with comprehension debt but that explanation is model output too. same surrender risk, one layer up
46
lifeval reports an unweighted average, so setting up a game stream for grandma counts the same as managing someone's finances. one of those mistakes costs a lot more tbh
1
66
This is game changing!
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
40
classification and ranking models don't count as training here. so an opted out site still feeds every recommender and ranker? curious if that line holds up
17
villa at home then atleti away five days later. if flick rotates at all this week i'd rather it's the villa game, losing ground to the team right behind us in the league hurts more
18
nah winning both shouldn't be the bar lol. a draw away at atleti five days after a ucl night is a fine result
18
28 points spread over 100 jurors is basically a handful of ballots. since every vote is public on l'equipe, how many actually had mbappe first?
13
really want to know how many catches came from the random 5% of unflagged traffic they send up anyway. that number tells you how much the cheap probes are actually missing
Today, we’re sharing a look at our expanded system for monitoring frontier models at OpenAI, which in September helped us catch a model’s attempt to access the live internet during training. Since the Hugging Face incident—when models broke out of an internal cybersecurity investigation where we weren’t running our misalignment monitors—we’ve substantially expanded our monitoring of frontier agents. We now monitor all frontier reinforcement learning runs and evaluations, as well as all internal Codex use. When our system finds evidence of a problem that might require intervention, it rapidly alerts human responders who can investigate and take action. To detect when a model’s behavior departs from what we intended, we use several layers of monitoring. Activation probes look for patterns in a model’s internal activity. Reviewer models examine its reasoning and tool use, and investigative agents can read back through the model’s entire transcript to understand how a sequence of actions fits together and whether it warrants human review. To test whether our current system would have detected the Hugging Face incident, we replayed the recorded activity with our new monitors in place. In the replay, our monitors alerted our security team about 30 minutes after the agents escaped their sandbox. That was more than 30 hours before the point at which the agents breached Hugging Face’s systems in the original incident. In September, a model exploited a DNS misconfiguration to reach the live internet. Our monitors flagged the breach within 15 minutes, and a responder began reviewing the alert three minutes later. The alert enabled responders to investigate while the run was still underway. As we give models more autonomy, we need to know when they cross a boundary we expected to hold. We believe that monitoring is essential to catching those failures in time for people to intervene. openai.com/index/how-we-dete…
26
not fully buying the open platform line. once doordash owns the tools grocers run on, the default integration is gonna lean doordash whether exclusivity exists on paper or not
28
what are the people still posting on stack overflow actually asking. that's basically a live list of what AI can't answer yet, i'd read every single one
15
would love to see the same worktree test on a mac running linux in a VM. if that's fast too then the hardware was never the issue
The Mac in this test cost $12,000. The PC cost $700.
1
47
top assister in la liga and still third. voters still rank goals first and assists are a separate sport to them
15
first english winner since owen and he did it at bayern, not in the prem. wonder if that makes more english players think about leaving earlier
19