tarruda

32 ms·

481.
by tarruda 4y ago
It is basically free if you consider companies will be automating a lot of the dev grunt work for now on. Learning to code using tools like Copilot will be a requirement, but instead of 10 developers you hire 1.
482.
by tarruda 4y ago
When you use copilot integrated in the editor, does microsoft collect all source code data on your project or only context used to perform the completions?
483.
by tarruda 4y ago
> I think it's now clear that higher level capabilities are emerging in these models and we need to take this seriously as a social/economic/political challenge. It is a hard truth to face. I admit I always feel a little b
484.
by tarruda 4y ago
Apparently can fine tune and run Alpaca-lora 7B using a single RTX 4090: https://github.com/tloen/alpaca-lora A few days ago before it was taken offline I did some tests I did with Alpaca-stanford. If I can run a GPT-3
485.
by tarruda 4y ago
Deno reminds me of how Python 3 decided to break compatibility in order to get rid of legacy code, and after finally replacing Python 2, has accumulated its own share of legacy.
486.
by tarruda 4y ago
I did not dislike your comment. The reason I considered easy was because I have very little knowledge in this area, in fact this is the first time I ever ran a machine learning model on my computer. I could not do it with the unmodified pyt
487.
by tarruda 4y ago
> if these models are trained on essentially raw WWW sourced data, I would personally want raw output. Llama is a very high-quality foundation LLM, you can already run it very easily using llama.cpp and will get the raw output you need.
488.
by tarruda 4y ago
I have very limited understanding of how these systems are created, but as far as I know they are not programmed to simulate anything. It is just a system that outputs the most likely next word based on a really large dataset of training sa
489.
by tarruda 4y ago
Maybe if you train/tune GPT-4 with enough samples of similar interactions, it will learn to do it too.
490.
by tarruda 4y ago
That was what I thought until a few months ago when ChatGPT was released. I never cared much about LLMs because it always felt like a brute force method to solving problems. What I'm seeing now is that some kind of intelligence seem to
491.
by tarruda 4y ago
To me it is really mind blowing that these properties (coding, emulate chatbot) emerge from just from feeding these neural nets with text data. Thanks for the info.
492.
by tarruda 4y ago
Thanks, I've saw this leaked chat prompt and that gave me some ideas. So under the hoods, ChatGPT is just a model like Llama where they prepend every user input with a context that makes it behave like a chatbot?
493.
by tarruda 4y ago
Can you show some example prompts you use for coding? Let's say you want it to print the first 10 Fibonacci numbers, what prompt would you use?
494.
by tarruda 4y ago
What kind of hardware is necessary to run non-quantized LLama-30B?
495.
by tarruda 4y ago
Still impressed with the output of a 4gb model, thanks for this.
496.
by tarruda 4y ago
My first attempt: $ ./chat main: seed = 1678992123 llama_model_load: loading model from 'ggml-alpaca-7b-q4.bin' - please wait ... llama_model_load: ggml ctx size = 4529.34 MB llama_model_load: memory_siz
497.
by tarruda 4y ago
The answer is yes: It ran on a x86 virtual machine with 4 CPU threads and 12gb of RAM
498.
by tarruda 4y ago
Looking forward to try it, but I don't have a macbook. I wonder if it runs on i7-11800h (8 core 16 thread CPU) with 64gb RAM
499.
by tarruda 4y ago
> In fact, I expect to be shortage of programmers even if none of those would happen (and they will) Did you see GPT-4 demo yesterday? Seems to me that after a LLM has been trained to the level of GPT-3.5, we programmers not only are not
500.
by tarruda 5y ago
Any chance you're related to Griffin Johnsen the Armchair Historian? Really nice YouTube channel, and I really admire that guy's productivity.
501.
by tarruda 5y ago
I wonder how long until DIY weaponized drones are available to common criminals. It doesn't seem like we are prepared to deal with something like this.
502.
by tarruda 5y ago
> It is my main gripe with JavaScript (and with typescript): a lack of a standard library that everyone uses and trusts. Something that should be predictable and boring, is absolute chaos in the js world This is something that Deno is at
503.
by tarruda 5y ago
It seems the person responsible for this edit mistook Deno's version with a decimal number.
504.
by tarruda 5y ago
> I just know very well that Ryan is redoing exactly the same mistakes as Node with the same obsession he had on "EPOLL"[0] back then that will end up in a new fiasco. "epoll" is a Linux API for listening on multiple
505.
by tarruda 5y ago
I don't understand what you mean by "default". Deno simply provides a builtin compiler which allows it to run Typescript transparently. Clearly Deno also supports JS out of box.
506.
by tarruda 5y ago
> Here I don't see how Deno is solving this , all the APIS seems again so barebone.. instead of having 1000+ dependency from NPM you'll have 1000+ dependency from remote URL With Deno you can already do a lot with only what is
507.
by tarruda 5y ago
Have you considered the possibility that your high salary is not entirely due to a shortage of people with the same skill level, but rather because SF/NY companies have too much investor money to spend? I'm not saying that good so
508.
by tarruda 5y ago
http://web.archive.org/web/20210511131438/https://twitter.co...
509.
by tarruda 5y ago
It seems a common trend that software engineers think their high salaries are something they can take for granted forever. I prefer to think that we are riding in a bubble that can pop any minute now. This line of thinking allows me to be m
510.
by tarruda 5y ago
> My eyebrows would be a little raised. Not that I would call them out in a tweet, but I would at least be curious where that guy is spending his time. It is one thing to raise the issue in private with "Project A", another thi

More ›