Majromax

8 ms·

91.
by Majromax 8mo ago
Pi supports restricting the set of tools given to an agent. For example, one of the examples in pi --help is: # Read-only mode (no file modifications possible) pi --tools read,grep,find,ls -p "Review the code in src/&qu
92.
by Majromax 8mo ago
What is your threat model? If you're worried about a hostile agent, then indeed sandboxing is not enough. In the worst case, an actively malicious agent could even try to escape the sandbox with whatever limited subset of commands it&
93.
by Majromax 8mo ago
> I'm just guessing, but seems the people who write these agent CLIs haven't found a good heuristic for allowing/disallowing/asking the user about permissions for commands, so instead of trying to sit down and actuall
94.
by Majromax 9mo ago
> In other words, such a structure would not dissuade bad actors with large financial incentives to push something through a process that grants validity to a hypothesis. That's not the right threat model. The existing peer review
95.
by Majromax 9mo ago
Chicken and egg problem: since conferences have limited capacity, you need to pre-filter submissions to see who gets a presentation spot.
96.
by Majromax 9mo ago
Welcome to Hacker News 26! I know you're trying to read a comment now, but would you like to take a tour of new features? [ ] — Yes, I'd welcome a half-hour distraction from the thing I need to use the computer for right now! [ ]
97.
by Majromax 9mo ago
I wonder if a breakpoint was out-of-order execution. Many computations would use some values from memory plus other that could be computed, and out-of-order execution would allow the latter to proceed while waiting on memory for the former
98.
by Majromax 9mo ago
You can infer the discount from the pricing of the batch API, which is presumably arranged for minimum inference costs. Anthropic offers a 50% discount there, which is consistent with other model providers.
99.
by Majromax 9mo ago
From this article: > For the purposes of this post, I’ll use the figures from the 100,000 “maximum”–Claude Sonnet and Opus 4.5 both have context windows of 200,000 tokens, and I run up against them regularly–to generate pessimistic estim
100.
by Majromax 9mo ago
> As long as it's unaccounted for by users it's at best anexternality. Why is it an externality? Anthropic (or other model provider) pays the electricity cost, then it's passed along in the subscription or API bill. The
101.
by Majromax 9mo ago
True, but from the pdf it seems like the fee charged of market makers is 1.75¢ × P × (1-P) per contract. Near P=0 that's approximately 1.75% of the notional amount invested, but near P=1 that's approximately 1.75% of the potentia
102.
by Majromax 9mo ago
The analysis is interesting, but I think it ignores a few factors: 1. The article mentions the bid/ask spread for contracts, but I believe that Kalshi also has its own fee structure. Small edges (an expected loss of 0.57¢ on a 1¢ cont
103.
by Majromax 9mo ago
I did some math for this particular case by asking Google’s Gemini Pro 3 (via AI studio) to evaluate the press release. Nvidia has since edited the release to remove the “tons of copper” claim, but it evaluated the other numbers at a repor
104.
by Majromax 9mo ago
The gist showsmthat the first line ofmthe system prompt must be "You are Claude Code, Anthropic's official CLI for Claude." That’s a reasonable attempt to enforce the ToS. For OpenCode, they also take the next step of addit
105.
by Majromax 9mo ago
That seems to have nice properties, but 2019 was a while ago. Is the problem of top-k sampling still relevant with much better frontier models?
106.
by Majromax 9mo ago
It could be a trademark violation, even in the US, under the argument that DoorDash was “passing itself off” as the infringed-upon company. However, DoorDash would then argue that it was being honest – it was genuinely delivering authentic
107.
by Majromax 9mo ago
Why? Grain is a commodity, but I buy flour at the store rather than grow my own. The “commmodity” argument suggets that new companies should stay away from model training unless they have a cost edge.
108.
by Majromax 9mo ago
In that case though, why the lock-in? If the combination really does have better performance than competitors’ offerings, then Anthropic should encourage an open ecosystem, confident in winning the comparison.
109.
by Majromax 9mo ago
> Anthropic is protecting its huge asset: the Claude Code value chain Why is that their “huge asset?” The genus of this complaint is that Opencode et al replace everything but the LLM, so it seems like the latter is the true “huge asset
110.
by Majromax 9mo ago
If you remove the conditional and keep the same math, you divide by zero and get nans. In the limit as temperature goes to zero, you do in fact get maximum likelihood sampling.
111.
by Majromax 9mo ago
It sounds like Starlink uses GPS to localize the receiver, rather than for any active step in the communication link. Since most receivers are static, I wonder if an effective workaround to this is for the receiver to just remember its las
112.
by Majromax 9mo ago
GPS signals are extremely weak, and they're necessarily received from omnidirectional antennas that can't provide much antenna gain. In some sense it's a miracle of signal processing that GPS can ever be received.
113.
by Majromax 9mo ago
> If all the models are interchangeable at the API layer, wouldn't they be incentivized to add value at the next level up Just the other day, a 2016 article was reposted here [ https://news.ycombinator.com/item?id=465
114.
by Majromax 9mo ago
> If you don't use or understand how LLMs fit in your workflows, you are not the target customer. I feel like this is a major area of divergence. The "vibes" are bifurcating between "coding agents are great!" an
115.
by Majromax 9mo ago
It's not just troll levels, see https://enclose.horse/play/n9YDZm
116.
by Majromax 9mo ago
Share repurchases are also a way of returning cash-flow to owners. A stock buyback is equivalent to a dividend, just with automatic reinvestment in the default 'do-nothing' case. Including buybacks, few large and mature companies
117.
by Majromax 9mo ago
> Where does the line fall between provider responsibility when providing a tool that can produce protected work, and personal responsibility for causing it to generate that work? There's one more line at issue here, and that's
118.
by Majromax 10mo ago
> Not having thought about it too hard, if the three norths all align at -2 degrees west longitude, wouldn't they also be aligned at 178 degrees East longitude, somewhere near the east coast of New Zealand? There is no required equi
119.
by Majromax 10mo ago
> A lot of them will fail, but if a few are extremely successful, they could make up for the failures. It kind of felt like the studios were doing this for a while (Grow Home, Little Nightmares) but my sense is that it has kind of stoppe
120.
by Majromax 10mo ago
> I'm not sure how they could constantly try double spends without risking losing the money each time the attack doesn't happen. If you're not trying to profit from the double spend itself but rather from a collapse follow

More ›