nijave

6 ms·

151.
by nijave 4mo ago
Any chance you'll retest with latest iterations of models? Opus 4.8, GLM-5.2, Kimi K2.7, etc
152.
by nijave 4mo ago
At the metrics level, you can compare old vs new release. Have been bitten before by resource requirements dramatically change (regardless of whether it's a bug or functionality change)
153.
by nijave 4mo ago
That is fine and would have even been a better reply. I connected to a VPN in Belgium, opened a private browsing window, and Yandex, Google, and DuckDuckGo all find "latitude 41 in Columbus" as the first search result. It figured
154.
by nijave 4mo ago
I tried 3 more times. Two were nearly identical and 1 recognized Latitude 41 as a restaurant but had a similar useless reply
155.
by nijave 4mo ago
> Me: What was the sushi place near latitude 41 in Columbus? Did it go out of business I don't see it on Google maps anymore > Sonnet 5 (medium): None of these past chats mention a sushi place — I don't have anything on reco
156.
by nijave 4mo ago
Pretty sure they invented Firecracker for Lambda. Iirc they were previously using a hot pool of EC2 instances behind the scenes with each customer getting their own instances and lambdas sharing capacity on an instance. Firecracker made it
157.
by nijave 4mo ago
I don't know much about academia but it seems journals should be cooperatively edited and managed by universities rather than fully privitized. Ideally with an off ramp into public archives after a period of time.
158.
by nijave 4mo ago
Privatizing gains, socializing loses Bots and automation aren't inherently bad, but often times the motivation is pure cost savings for the company and users pay the price. A more rational design could have a periodic human auditing pr
159.
by nijave 4mo ago
Ahh yes, the AI influencers... >the people who made fortunes during a gold rush weren't the miners, but the ones selling the shovels
160.
by nijave 4mo ago
Maybe it depends on product offering. WAF + CDN was competitive with Akamai (through a VAR) and AWS and came with free Zero Trust seats. Iirc Argo and some other routing products are quite expensive.
161.
by nijave 4mo ago
The anti AWS rhetoric is exhausting. The fact AWS crappy hosted ES gained any market share is more a testament to how bad ES sales practices were than anything else. ES really missed the mark by not having a simple, self-serve sales model a
162.
by nijave 4mo ago
Yeah curious why they didn't use Presto/Trino, DuckDB, or Clickhouse SQL directly with UDFs and views to augment Zuora exposes a Trino-based data warehouse which is quite nice and powerful Besides the parser side, existing dev too
163.
by nijave 4mo ago
DuckDB Also Windows used to have a free VHD with a trial license you could download (and convert to different format with qemu-img)
164.
by nijave 4mo ago
It is a tool for some and a crutch for others
165.
by nijave 4mo ago
> Strands I like it quite a bit. Imo it's sort of like the "Flask" of frameworks. It's pretty easy to get started with and has a pretty pluggable ecosystem where you can choose models, providers, tools without much lo
166.
by nijave 4mo ago
It is common for support to closely work with product and be able to respond with product roadmap and feature guidance. At least they can say "we sympathize but it's not a priority" or "it's a medium priority but we
167.
by nijave 4mo ago
In fairness to Claude, I've nuked my homedir (had 2 tmux panes open, 1 in home and 1 in /tmp/... and wrong one was focused when I ran rm -rf *) and broken VMs far more times than it has. I now embrace IaC and backups
168.
by nijave 4mo ago
I read the Github issue. They also initially refused a refund until the whole thing blew up Here's another one https://github.com/anthropics/claude-code/issues/52809 It's definitely not an isolated
169.
by nijave 4mo ago
I imagine a smaller single node model would have a significantly better experience at significantly lower cost. When I was poking around with infra estimates it seemed the main issue/cost was once you crossed from single-node to multi-
170.
by nijave 4mo ago
I gave Sonnet 4.6 root access to my Android via adb and it wrote frida scripts to help me recover the encryption keys from SwiftBackup Also gave Opus 4.6 access to a Kubernetes container and it was able to use pyrasite (a Python replacement
171.
by nijave 4mo ago
Which model? I see a suspiciously similar post on amd.com running 2 bit Kimi quant on a four node cluster over 5Gbps Ethernet Assuming math works here although I think there's some caveats depending on the model architecture, 1T 4 bit
172.
by nijave 4mo ago
I think we'll see - better hardware - more efficient model runtime algorithms/code - smarter/more efficient models (same capability with less parameters) So ideally these will all come together and help.
173.
by nijave 4mo ago
Their "safety" system that arbitrarily flags things. Tons of examples of innocuous strings setting it off and sometimes with financial impact like the OpenClaw/hermes thing (just having the word "hermes" would insta
174.
by nijave 4mo ago
That's fair for new hardware. You probably want to prompt "homelab" or "used hardware" to compare what's in this thread.
175.
by nijave 4mo ago
I got a 2U rackmount with 192Gi DDR4 for $1.1k USD in 2023. Around 1.5 yrs ago, server RAM could be had pretty cheap--especially slower LRDIMMs (I wanna say 512Gi DDR4 was <$500 USD). I checked a couple old ServeTheHome threads and seein
176.
by nijave 4mo ago
Well, it's about GPU VRAM if you want something competitive with cloud-hosted offerings at the performance levels showing in benchmarks. This is a heavy quant with quality degradation and significantly lower performance. Cloud offering
177.
by nijave 4mo ago
I agree single digit tk/sec is painfully slow, but I also doubt anyone with these local/homelab setups are using them for work. Likely fire off and check back later. That said, I've had terrible results one-shotting so you&#x
178.
by nijave 4mo ago
Yes, especially if I know the person/have heard them talk before.
179.
by nijave 4mo ago
z.ai coding plan is a fairly decent deal at ~$16/mon USD considering it's supposed to have a fair bit more usage than the comparable $20/mon Claude plan. On the other hand, z.ai seems a bit on the slower side for raw model to
180.
by nijave 4mo ago
Not this one. It's a SOTA-class model >800Gi VRAM required at fp8

More ›