HNHacker News
TopNewBestAskShowJobs

axpy906

344 karma · joined May 15, 2021

Opinions are my own. On my phone so forgive typos.
submissionscomments
axpy906··on Amazon to managers: You can now sack employees who won't come 3 times a week
Mostly pip at Amazon.
axpy906··on Ask HN: Best Tech Courses/Certifications Under $400 for a Back End Engineer?
AWS Certified Data Engineer: https://aws.amazon.com/certification/certified-data-engineer...
axpy906··on Ask HN: How do your ML teams version datasets and models?
You included data lineage tracking in the first part which probably needs to be piped from an orchestrator.

At this point it’s a build or buy type deal for models using a model registry service.

Data versioning still feels unsolved to me.

axpy906··on Ask HN: How do your ML teams version datasets and models?
I really like your read me.
axpy906··on Ask HN: How do your ML teams version datasets and models?
Is the bucket versioned?
axpy906··on The Data Science Manifesto (2020)
Add 2020 to title
axpy906··on Lead Toxicity from Ayurvedic Medicines
Ok so it doesn’t mention what she was taking. It could be turmeric, ashwghanda or something else. Said 4
axpy906··on Fine-tune your own Llama 2 to replace GPT-3.5/4
Unfortunately, Lama2 is not a fully open source license.
axpy906··on The Power of Food for People with Dementia
> Max Lugavere enters the chat
axpy906··on Asking 60 LLMs a set of 20 questions
Step in the right direction. Would be nice to have some objective metric added to score against.
axpy906··on Advanced NLP with SpaCy
Large Language Model in my opinion means that it has billions of parameters and possibly cannot fit on a single machine.

Not sure what Spacy is doing under the hood these days but always thought of “neural net” as word2vec type model which won’t hit the above definition.

axpy906··on Refact Code LLM: 1.6B LLM for code that reaches 32% HumanEval
When I wrote “backend” was a poor choice of a word. “Meta-model” is probably a better choice of wording.

I hope it did not detract too much from the point of focusing on subtasks and modalities for FOSS as GPT 4 was built on a $163 million budget.

Finally, good point. We’ve got no idea of what OpenAI’s MoE approach is and how it works. I went back to Metas 2022 NLLB-200 system paper and they didn’t even publish the exact details of the router (gate).

axpy906··on Show HN: finetune LLMs via the Finetuning Hub
Thanks for putting this project together. How does your project differ from similar ones? (looked at the main repo)
axpy906··on Refact Code LLM: 1.6B LLM for code that reaches 32% HumanEval
The key in that is models. Per the GPT4 leaked details, it’s not a a single model but 16 MOE mixture of experts. There’s probably quite a lot of complexity on the backend in sourcing the right model for the right query. In short, it’s probably better to focus on single models for specific tasks in the OS community as evidenced by Code Llama. Having a system like GPT4 is still difficult to replicate. Getting it to run on a consumer hardware for specific tasks like code gen at almost GPT4 level is doable.
axpy906··on Remote Work Is Here to Stay, and Employers Aren’t Happy About It
30 increase and 40 decrease which is a conflicting finding.
axpy906··on Remote Work Is Here to Stay, and Employers Aren’t Happy About It
> 35% of all jobs that could be done remotely were being done by people who worked remote full-time—five times as much as before the pandemic

I had to read this two times as it’s poorly worded.

An any rate, it should be 80% and not 35%. You don’t need to go into an office to make software.

There’s something just sinister how employers are handling this as it’s got nothing to do with actual output.

axpy906··on Anti-AI Hype LLM Reading List
Wasn’t this posted before?
axpy906··on US inflation means families are spending $709 more per month than two years ago
Assuming $709 an average number over all cost of living locations. We’re spending more than that.
axpy906··on Azure ChatGPT: Private and secure ChatGPT for internal enterprise use
Use SageMaker: https://www.philschmid.de/sagemaker-llama-llm
axpy906··on My $0->$100M->$0 in 5 years story
> Imagine you hire someone to scale an area and their solution is to divide that area into 5 sub areas and hire someone for each of them to figure it out, etc. that’s their solution.

FANNG hires?

axpy906··on A practical way to detect cargo-cult thinking
This article is missing a lot of the meaning of cargo cult. It’s got religious tones as well as irrational beliefs. Good YouTube video on it it: https://youtu.be/x5jbHDMeDFE
axpy906··on Attention Is Off By One
I walked into this one thinking it might be about a/b testing. Pleasantly surprised.
axpy906··on Consumer software is expected to be next fast-growing segment (1994)
I think AI finally is at point where it is. It’s peak hype right now but there is definitely a difference between knowledge workers that use it and one’s that don’t. It’s only going to grow wider in the coming years.
axpy906··on Every time you click this link, it will send you to a random Web 1.0 website
https://webhamster.com/

I am just going to leave my hit here.

axpy906··on Every time you click this link, it will send you to a random Web 1.0 website
Yes, if you look at the bottom it’s got updated citations from the 2010’s.
axpy906··on Stop eliminating good candidates by asking them the wrong questions
Interview Question: “Show me how to reverse a linked list.” Position : Backend AI Engineer
axpy906··on Show HN: Visual intuitive explanations of LLM concepts (LLM University)
You sir get an up vote for simply being Jay on HN. Thank you for all you do.
axpy906··on “Don Knuth Plays with ChatGPT” but with ChatGPT-4
Nailed every one. Some by saying not possible to answer but still.
axpy906··on Writing summaries is more important than reading more books
Thank for this comment; it’s gold.
axpy906··on Hitler’s Antarctic base: the myth and the reality (2007) [pdf]
So this information about Highjump was never declassified?
← PreviousPage 6 of 10Next →