darkbatman

also its pretty big model inference costs are high even with margins running a 2.8T model costs a lot. if they release oss may be it goes down to $10-12 per million tokens.

This is crazy and would be frustrating, I probably would just be using another model as authority and keep fable as reviewer only in this case.

I built https://meow.gs as a side project. It gives you a cloud shell via SSH, authenticated with Face ID / Touch ID instead of SSH keys.

ssh meow.gs from any terminal iPad, iPhone, laptop. Easy to work while remote and have ai coding tools on private vps and access from anywhere.

Though claude/codex now has remote access, this was created 1 month back but still helpful if you need own private tools on vps and accessibility without managing ssh keys.

codex/gpt is a stubborn model, doubt it would accept claude reviews or counter it. have seen cases where claude is more willing to comply if shared feedback though its just sycophancy too.

By looking at the paper, memory needed per layer seems to be higher than transformer architecture. Pretty sure that would be blowing up the vram of gpu at scale.

Its so useful to use Cerebras api for other tasks too not just coding with qwen coder but even simpler things like lets say analysing with gpt-120 oss or llama.

Just plug it in with normal chat interface like Jan or Cherry studio and its incredibly fast.

We kinda use https://github.com/googleapis/genai-toolbox but for databases looking forward if klavis provide more or general solution.

Ideally when we are writing agents we need mcp to support auth, custom headers because by design when deploying for saas we need to pass around client params to be able to isolate client connections.

We do token optimisation and other smart stuff to save token money. Looking forward to try this as well if this solves similar problems as well

true, I was using Cline/Roocode from almost an year and it always made sure to read things from memory-bank which i really liked. Claude has gone downhill from August mid for me and often it doesn't follow instructions from claude.md or forget things mid-way.

Mostly agree with article, though what happens in few years when juniors will eventually become senior.

Personally seeing trend juniors are relying so much on AI that they can't even explain what they wrote even in interview or coding assignments or even PR. Its like blackbox to them.

I believe then we would see the higher impact or may be by then its solved problem already.

With already https://agentcommunicationprotocol.dev (ACP) same name seems confusing now. even though differences are there in both.

Would be nice to see some benchmarks.

Also from my experience you need more power to get some significant result. Mostly fine tuning would work if base model is very close to what you are trying to achieve and you won't be much happy with the results though.

Also context length becomes an issue trying to fit in with gpu with lesser ram.

Are there any load test results available, we would like to use this at zenskar but at high scale really need it to work.

System merges and final are definitely unpredictable so nice project.

Is it similar to rethinkdb, I remember using RethinkDB for similar use case for live queries running directly on database layer back in 2016.

Not the temporal you are thinking of. its Date replacement in JS.

We do quite similar but through Debezium/Kafka CDC pipeline to clickhouse. This way primary database is protected. Directly querying from files from postgres might get slower eventually IMO compare.

Hi Do you support visa transfer or visa requests?

I tried google.com. It says invalid url. looks half baked. can you not force to write protocol.

Thanks these config are helpful

we are actually trying something similar but possible kinesis + clickhouse or kafka + clickhouse. Currently kinesis seems easier to deal with but not a good intergration or sink connector available to process records at scale for kinesis to put into clickhouse. Were you ever felt into similar problems where you had to process records at huge scale to be able to insert into clickhouse without much delay.

One more thing is kinesis can have duplicates while kafka is exactly once delivery.

how are you ingesting from kinesis to clickhouse. are you using some custome sink connector or processes on ec2 or lambda?

Does someone know similar guide for javascript/node.js or its a good thing to start one?

Would you be interested in hiring someone from out of UK?

If someone need my suggestion, here is what you do - try doing different sort of questions (multiple topics too) on leetcode from different topics may be just 20-30 in language you want to learn.

Its different way but you will learn a lot of new libraries, ways to mutate objects, lists, all sort of data structures and new things really really fast.

This is so true, most top companies expect folks to write flawless bug free code of problem in 20 minutes or less now. Some expect two problems in 45 minutes now (if you are unlucky really one medium and one hard). If you write a sub-optimal solution you are rejected for sure. If you haven't grinded leetcode/hackerrank and haven't seen it before there is no way you are going to solve a hard problem in 20 minutes - lets say 10 minutes to think and 10 minutes to write the code for a hard leetcode question.

Thanks for looking :)

It looks like this is being reported here as well: https://www.cloudflarestatus.com/

Login seems to be in redirect loop: https://imgur.com/a/dWTpG99

It's crazy how in India its approximately $850 which is huge.

This looks more like a problem that could have easily been solved with traditional algorithms, ranking keywords, etc.

I am doing deeplearning.ai. The course seems to be really nice.