About a year ago, I watched this 3Blue1Brown LLM tutorial (https://www.youtube.com/watch?v=eMlx5fFNoYc&t=367s) on how a model’s self-attention mechanism is used to predict the next token in a sequence, and I was surprised by how little we know about what actually happens when processing the sentence “A fluffy blue creature roamed the verdant forest.”
A year later, the field of mechanistic interpretability has seen significant advancements, and we're now able to "decompose" models into interpretable circuits that help explain how LLMs produce predictions. Using the second iteration of an LLM "debugger" I've been working on, I compare the hypothetical representations used in the tutorial to the actual representations I see when extracting a circuit that describes the processing of this specific sentence. If you're into model interpretability, please take a look! https://peterlai.github.io/gpt-circuits/
Emburse (YC W16) | Software Engineer | Full Time | ONSITE preferred, REMOTE | https://www.emburse.com/
Emburse is a corporate credit card platform that generates virtual and physical credit cards with specific budget restrictions. We manage both traditional corporate card expenses like T&E as well as backend vendor payments.
Emburse is built using Python/Django and is deployed on Heroku for development and AWS for production. We're looking for someone to join us as one of the first few employees and who would be comfortable embracing sizable infrastructure responsibilities which include:
Emburse is a corporate credit card platform that generates virtual and physical credit cards with specific budget restrictions. We manage both traditional corporate card expenses like T&E as well as backend vendor payments.
Emburse is built using Python/Django and is deployed on Heroku for development and AWS for production. We're looking for someone to join us as one of the first few employees and who would be comfortable embracing sizable infrastructure responsibilities which include:
Emburse is a corporate credit card platform that generates virtual and physical credit cards with specific budget restrictions. We manage both traditional corporate card expenses like T&E as well as backend vendor payments.
Emburse is built using Python/Django and is deployed on Heroku for development and AWS for production. We're looking for someone to join us as one of the first few employees and who would be comfortable embracing sizable infrastructure responsibilities which include:
For some companies such as those in manufacturing, enhanced oversight and better spend control are necessary card program requirements. I agree that business credit would add significant value. Thanks for the suggestion.
For now, we're starting with a debit card product because they're easier to launch. Eventually, as the product gains more traction, it will be easier to launch a credit product with the advantages you're looking for.
Which that said, there's plenty of opportunity as a debit card product. We believe that no financial product right now really prioritizes spend control and bookkeeping accuracy in the way we do within our product.
Founder here - looking for feedback on the product. You can think of Emburse as being like AmEx with Expensify-like features built-in. If you're in the process of transitioning from having all spend on a shared corporate card to having cards for each department or employee, definitely take a look.
Hi there, co-founder of Crocodoc here. Crocodoc will continue being the best platform for document viewing and collaboration. Box is investing heavily in our platform and will make sure we stay awesome. Any updates to the API going forward should be totally seamless, but if you run into any problems, just give me a ping (I'm Peter). I'll make sure the transition happens smoothly.
Hey guys - Peter here from Crocodoc. In the official release, pages on mobile devices will be lazy-loaded to prevent the pages from using too much memory like it does on this demo here.
San Francisco, CA | Full time | Director of Product
## Description
Crocodoc’s HTML5 document collaboration platform reaches tens of millions of users, and our business is profitable and growing fast. Our customers include Dropbox, LinkedIn, and Yammer, and we're funded by investors including YC, SV Angel, 500 Startups, and top angels from Silicon Valley.
Right now we're inundated with sales inquiries from SMB customers who want to embed documents within their web and mobile applications. We’re looking for a multifaceted candidate to take charge of our SMB product and blow it out of the water. This is a key leadership role with enormous growth potential.
## Responsibilities
- Talk to LOTS of prospective customers (both product managers and developers)
- Develop our SMB product offering
- Optimize our pricing plans and product tiers
- Develop a scalable sales process and design a self-service experience
- Develop user acquisition strategy (everything from inbound marketing to direct sales)
- Qualify leads, own our sales pipeline, and close lots of deals
TL;DR: you must be capable of growing our SMB product like crazy while wearing many hats in the process.
## Qualifications
- World-class ability to communicate with clients and develop customer relationships
- Experience with SaaS and/or B2B technology solutions
- Basic technical experience (you’ve written a web app) required
- Startup experience (you can set own agenda and thrive in a fast-paced environment) a plus
- Product experience (you understand principles of UI and UX) a plus
- Product marketing experience (you’ve been responsible for user acquisition) a plus
This is a tremendous opportunity to own an entire product from soup to nuts. If you think you'd be a great fit, please contact us at [email protected] or visit https://crocodoc.com/jobs/.
The focus of this experiment isn't so much proposing new UX that we think should be adopted. More so, this experiment is to showcase what is now possible using the newest HTML5 features - to provide inspiration for those trying to implement novel HTML5 animations. A page flip allowed us to show off some of my favorite new HTML5 effects.
Crocodoc - San Francisco, CA | Web Developer, Biz Dev, Growth Hacker
Crocodoc converts documents to HTML5 for some of the largest web companies you use everyday: Dropbox, LinkedIn, Yammer. We have a small team, and yet business is growing fast (5x this year). We're looking for amazing early startup employees to help us grow the company even faster.
I'm flattered the author mentions Crocodoc. Crocodoc is hiring by the way if anyone wants to hack on stuff like this full time: https://crocodoc.com/jobs/
I'm a dev at Crocodoc. For document assets, we're doing tremendously obscure things to our CSS and Javascript in order to optimize our document loading experience. The unconventional use of Javascript to embed resources relates our needing to work around a few browser quirks in order to load resources faster when fetching assets from our server.
But yes, we really could do a better job of merging the other browser resources mentioned.
A year later, the field of mechanistic interpretability has seen significant advancements, and we're now able to "decompose" models into interpretable circuits that help explain how LLMs produce predictions. Using the second iteration of an LLM "debugger" I've been working on, I compare the hypothetical representations used in the tutorial to the actual representations I see when extracting a circuit that describes the processing of this specific sentence. If you're into model interpretability, please take a look! https://peterlai.github.io/gpt-circuits/