zurfer

10 ms·

151.
by zurfer 2y ago
Sophisticated investors might think they can sell quickly enough if the tide turns.
152.
by zurfer 2y ago
"I apologise for writing such a long letter, but I didn't have time to write a short one."The quote, " is generally credited to Blaise Pascal, a French mathematician and philosopher. In his work "Lettres Provincial
153.
by zurfer 2y ago
to be fair, I was waiting for this release the whole day
154.
by zurfer 2y ago
that's cool thank you. any chance you will update the github repo [1] with the latest info? or any hint, how to run this thing myself. :) [1] https://github.com/prakhar897/indiehacker-scraped
155.
by zurfer 2y ago
Some complementary analysis of the same dataset where I look into how the top categories trend over time, what an AI product was before 2022 and who the most successful makers are: https://test.getdot.ai/share/78626655-
156.
by zurfer 2y ago
I wonder how they got the data. it says: "I found a publicly available list of all ProductHunt users, launches, upvotes, and comments." but perplexity or google search didn't help me
157.
by zurfer 2y ago
Is it fair to say that 2 of the 3 leading models are from Chinese labs? It's really incredible how fast China has caught up.
158.
by zurfer 2y ago
I love that they included some unsuccessful attempts. MCTS doesn't seem to have worked for them. Also wild that few shot prompting leads to worse results in reasoning models. OpenAI hinted at that as well, but it's always just a s
159.
by zurfer 2y ago
Has anyone build a linter that optimizes code for an LLM? The idea would be to make it more token efficient and (lower accidental perplexity), e.g. by renaming variable names, fixing typos and shortening comments. It should probably run aft
160.
by zurfer 2y ago
Hello fellow builder. I think the BI space is definitely ripe for change and I do wish more products had better dashboards for users! That being said, embedded analytics is a tough space, because established BI tools (PowerBI, etc.) can be
161.
by zurfer 2y ago
The authors are from: a Swiss Reinsurance Company, Ltd, Switzerland b Zardini Lab (research affiliate), Massachusetts Institute of Technology, USA c Autonomous Systems Laboratory (research affiliate), Stanford University, USA d Waymo LLC, U
162.
by zurfer 2y ago
It seems like they compared apples to apples. Which make that an exciting result! “The benchmark was calibrated using both mileage (driving exposure) and residence zip-code (geographic region). Specifically, only the claims associated with
163.
OpenAI o1 API and new tools for developers(openai.com)
9 points by zurfer 2y ago|1 comments
164.
by zurfer 2y ago
for anybody who wants to schedule or automatically run jupyter notebooks I recommend also looking into papermill: https://papermill.readthedocs.io/en/latest/
165.
by zurfer 2y ago
Hmm, I was expecting more of a discussion on HN for this release. I have so many question and no answer in the comments because nobody commented :P What is this poorly documented reasoning_effort parameter and how can I use it? Why is re-en
166.
by zurfer 2y ago
Model releases without comprehensive coverage of benchmarks make me deeply skeptical. The worst was the gpt4o update in November. Basically a 2 liner on what it is better at and in reality it regressed in multiple benchmarks. Here we just g
167.
by zurfer 2y ago
You give 3 data analysts the same question on a typical enterprise database. How many correct answers do you get? How many different answers? Based on anecdotal experience (we should make a study) LLMs don't make more errors than human
168.
by zurfer 2y ago
I do wonder if the command line will get a new generation of users with more tools like this that make systems more accessible. Our bet with a similar tool (getdot.ai) is that the shift goes into existing communication tools (Slack, etc.),
169.
by zurfer 2y ago
"Prompting is not enough, test-time fine-tuning is needed Clearly this competition has shown that LLMs need test-time fine-tuning to do new tasks. Few-shot prompting is not enough for the model to learn novel tasks." pretty intere
170.
3rd Place Solution for the Arc Prize 2024 Competition: Omni-Arc Approach(kaggle.com)
1 points by zurfer 2y ago|1 comments
171.
by zurfer 2y ago
I laughed and upvoted, but if anything I bet they put their best people on it to replicate this offering. What I take away from this is: we are just getting started. I remember in 2023 begging OpenAI to give us more than 7 tokens/secon
172.
by zurfer 2y ago
Python 3.13 is not removing the GIL. You just have an option to run without it.
173.
by zurfer 2y ago
It's not the same thing. ChatGPT search is an end user facing offering that is not available via API. Google's grounding feature is mostly for API users and seems like a parameter that can be used to first search before providing
174.
by zurfer 2y ago
nice thank you. I just added that as a custom search to alfred (on mac). Works well!
175.
by zurfer 2y ago
If I remember correctly, the claim was that M3 is 1.6x faster than M1. M4 is now 1.8x faster than M1. It sounds more exciting than M4 is 12.5% faster than M3.
176.
by zurfer 2y ago
If I don't know exactly what I am looking for perplexity.ai is best for me. If I know (e.g. company logo png) nothing beats google for me so far.
177.
by zurfer 2y ago
I gave your prompt to o1-preview and with one correction it did something that seems good to me (I am not a rust programmer, so please double check). :) first attempt: https://onecompiler.com/rust/42w2duuqh final resul
178.
Ask HN: Are models changing for pinned releases?
2 points by zurfer 2y ago|1 comments
179.
by zurfer 2y ago
hmm there is a follow up to show the difference in percent instead of absolute values, which is similar to the type of interaction you can have in chat and there is a sort of history on the left side, so things are chat like to some degree.
180.
by zurfer 2y ago
Here is the link to one of the prompts. It seems like all the LLM tasks are in the agents directory: https://github.com/microsoft/data-formulator/blob/main/py-sr... Some of these "agents" are u

More ›