ramraj07
Joined 10,576 karma
PhD in biomedical engineering, data scientist day before yesterday, software engineer yesterday, AI engineer today. http://ramrajv.com
- Cancer isn't caused by proteins in the way you might think. Its definitely not infectious at the protein level. You could ask if this disruption spreads out cancer cells themselves and that would be fair to ask. But then the cancer cells were already in your body and were likely trying to migrate to other sites anyway.
- Isn't it better to put it in an agent loop, with the structured output json just specified as a tool? The function call can then just return a summary of the parsed input. We can add in the system prompt a validation step to ask the llm to verify it has provided inputs correctly. This will allow the llm itself to self reflect and correct if needed.
- Until this administration there was no mandate to move manufacturing home, and importantly why would any company forgo significant profit to match an ideological framework, unless the ideology is what they sell or market?
- People from different countries especially where English is not their first language often have more esoteric words in their vocabulary.
- Those benchmarks mean nothing. Anthropic still makes the models that gets real work done in enterprise. We want to move but are unable to.
If anyone disagrees,I would like to see their long running deep research agents built on gemini or openai.
- Can folks who have compared Amp with other agents share their experience? Some of my colleagues swear this is the best agent out there.
- While we can agree that adding AI just to tick a box will win no awards, it will be a laughable proposition to suggest that Apple doesn't need to do anything on AI.
If anything its laughable and points to the unoriginality of product creators that we haven't fundamentally transformed how we interact with technology given how much AI offers as functionality. Anyone (I'll bet 20% on Ive) who figures this out will eat Apple's dinner.
- "trained on our public and internal docs" trained how? Did you mean fine-tuned haiku? Did you actually fine tune correctly? Its not even a recommended architecture.
Or did you just misuse basic terminology about LLMs and are now saying it misbehaved, likely because your org did something very bad with?
- Its a 2 day project at best to create your own bespoke llm as judge e2e eval framework. Thats what we did. Works fine. Not great. Still need someone to write the evals though.
- Vehement disagree. We implemented our own context editing features 4 months back. Claude released a very similar featureset we had all along last month. We were still glad we did it because (A) it took me half a day to do that work (B) our solution is still more powerful for our use case (C) our solution works on other models as well.
It all comes down to trying to predict what will be your vendors' roadmap (or if youre savvy, get a peek into it) and whether the feature you want to create is fundamental to your applications behavior (I doubt encryption is unless youre a storage company).
- Anthropic doesnt even allow temperature changes when you turn thinking on.
- Sure, but its an instruction that applies and the model will consider fairly relevant in every single token. As an extremely example imagine instructing the llm to not use the letter E or to output only in French. Not as extreme but it probably does affect.
- By pure nature of how companies work, a space company with this mandate and so much funding, unless its being used for money laundering, will have a modicum of progress. BO has barely had that. No space company with so much money and so much runway has achieved so little.
- Yes.. its called snowflake? Theyre exactly that and why they work so well. I know youre asking for an OSS but what snowflake offers is a fleet of servers that can build your cluster in a second as opposed to minutes that you need if you want to spin it up yourself..
- Everybody you replied to you made a completely different hypothesis but the waymo head itself mentioned why they waited on highways: on regular roads, if the computer fails to maneuver, you have an extremely simple, generally safe temporary solution: you just stop the car. Stopping a car is always kinda acceptable in regular roads. Its not an acceptable solution to undefined problems in the highway. This becomes important because in a Tesla theres still a requirement for a driver to be there to take care of worst case scenarios but in a waymo thats not true.
- Hey! you called me out: https://www.hackerneue.com/item?id=35178196
- Seconding GHD. They have added features very slowly, very thoughtfully; HN tends towards experts (or at least people who think they are). I am aware that I'm NOT good with git. I will never do anything that has "hard" or "rebase" in it without spending 20 minutes making sure its what I want to do. Unfortunately I have seen way too many semi junior engineers who think they're git lords who force push bad histories and ruin our git repo. I tend to suggest strongly that people should use github desktop if they are in my team though very few people take up that suggestion :)
- I will trust Google to abide by the rules more than any other big tech firm. Like with all my money ill make that bet. Not because I think they're good guys but from everything I have learned they have a culture that abides by rules like these. If they say they wont train on api use (they do say it) I feel assured they wont.
- Here's an on topic question: all the frontier model companies "promise" that they wont store and train on your api use if you pay for it. Who do you trust? I for sure will absolutely assume grok will just use the data I submit to train in perpetuity. Thats a scary thing for me and if anyone else does anything thats real work this should be great cause for worry if they wish to use grok.
I know compilers can do some crazy optimizations but wouldn't have guessed it'll transform something from O(n) to O(1). Having said that, I dont still feel this has too much relevance to my actual job for the most part. Such performance knowledge seems to be very abstracted away from actual programming by database systems, or managed offerings like spark and snowflake, that unless you intend to work on these systems this knowledge isn't that useful (being aware they happen can be though, for sure).