Live data from Hacker News

GPT-4.5 or GPT-5 being tested on LMSYS?

rentry.co

101–110 of 380 posts

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#101
I just tested against a prompt about using PVDF vs Piezo pickups on a hammered dulcimer, running the same conversation in parallel on 3.5. 3.5 gave more-or-less the output quality you'd expect. This blew it out of the water. I particularly liked this snippet (snipped and formatted for comment size) where it managed to pull out typical values and and actually showed the math.

> Can you explain more about the response and impedance of each? I am an engineer, so you can feel free to be quite technical.

---

>Certainly! Given your engineering background, let's delve deeper into the response characteristics and impedance aspects of both traditional ceramic piezo pickups and PVDF piezo pickups. Understanding these can significantly impact your choice, especially in terms of how they integrate with the rest of your audio chain (like preamps and amplifiers).

>

> Electrical Characteristics

> 1.

> Piezoelectric Coefficient

> Ceramic Piezo Pickups: The piezoelectric coefficient ( d_{33} ) of ceramics like Lead Zirconate Titanate (PZT) is typically very high. This coefficient measures the electric displacement produced per unit stress. For PZTs, ( d_{33} ) can range from 100 to 600 pC/N (picoCoulombs per Newton), depending on the formulation and orientation of the crystal structure.

> PVDF Pickups: PVDF generally has a lower piezoelectric coefficient, around 20 to 30 pC/N. Despite this lower coefficient, the material's flexibility and the ability to form larger, more adaptive surfaces can compensate in practical applications, offering sufficient voltage output under less stringent mechanical deformations.

> 2.

> Voltage Output

> The output voltage ( V ) from a piezo pickup can be estimated by the formula:

> [ V = d_{33} \cdot F \cdot A / t ]

> where:

> ( F ) is the force applied,

> ( A ) is the area,

> ( t ) is the thickness of the piezo material.

> Ceramic Piezo: Due to their high ( d_{33} ) value, even small deformations can produce relatively high voltages, which makes them sensitive to shock and vibrations, potentially capturing unwanted mechanical noises.

> PVDF Pickups: The lower ( d_{33} ) value means lower voltage outputs for the same mechanical stress compared to ceramics, but their larger possible surface area and thinner construction can enhance output levels and fidelity across a broader frequency range.

>

> [SNIP]

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#102

Earlier quoted context omitted.

sorry can you explain what the correct solution is. Do you need to know how much the density of alcohol is rel. water? Also water and alcohol volumes don't add as far as I remember, so you need to account that that too?

Yes, you need to know the densities, but the models outputted the correct densities even as part of the wrong answer. Not sure what you mean by volumes "not adding". One way to calculate it is like: density_alcohol = 0.789g/ml density_water = 1g/ml weight_total = 16g (density = weight / volume) volume_alcohol = 0.4 * volume_total volume_water = 0.6 * volume_total weight_total = weight_alcohol + weight_water weight_al…

https://www.flinnsci.com/api/library/Download/bc2d91b9d50f4e... I don't really know what ABV even means when the total volume of a mixture is not the sum of the alcohol volume and the water volume. I think practically it means your calculation is slight overestimation?

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#104

Earlier quoted context omitted.

Yes, you need to know the densities, but the models outputted the correct densities even as part of the wrong answer. Not sure what you mean by volumes "not adding". One way to calculate it is like: density_alcohol = 0.789g/ml density_water = 1g/ml weight_total = 16g (density = weight / volume) volume_alcohol = 0.4 * volume_total volume_water = 0.6 * volume_total weight_total = weight_alcohol + weight_water weight_al…

https://www.flinnsci.com/api/library/Download/bc2d91b9d50f4e... I don't really know what ABV even means when the total volume of a mixture is not the sum of the alcohol volume and the water volume. I think practically it means your calculation is slight overestimation?

Oh, I had no idea! I didn't account for that (or a bunch of other things like temperature). So yeah, might be a bit off, but not as much as the wrong initial answers from the LLM - which also failed to mention this.

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#106
post #19

Prompt: code up an analog clock in html/js/css. make sure the clock is ticking exactly on the second change. second hand red. other hands black. all 12 hours marked with numbers. ChatGPT-4 Results: https://jsbin.com/giyurulajo/edit?html,css,js,output GPT2-Chatbot Results: https://jsbin.com/dacenalala/2/edit?html,css,js,output Claude3 Opus Results: https://jsbin.com/yifarinobo/edit?html,css,js,output None is correct.…

When I first read this prompt I thought "Okay but what time do you want the clock to be set to?"

naturally the current time

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#107

Man, its knowledge is insane. I run a dying forum. I first prompted with "Who is at ?" and it gave me a very endearing, weirdly knowledgeable bio of myself and my contributions to the forum including various innovations I made in the space back in the day. It summarized my role on my own forum better than I could have ever written it. And then I asked "who are other notable users at " and it gave me a list of some mo…

It actually knew info about an account of mine on Newgrounds I abandoned around 2011, and I wasn't famous on there or anything.

Re: GPT-4.5 or GPT-5 being tested on LMSYS?

#109
I'm impressed. I gave the same prompt to opus, gpt-4, and this model. I'm very impressed with the quality. I feel like it addresses my ask better than the other 2 models.

GPT2-Chatbot: https://pastebin.com/vpYvTf3T

Claude: https://pastebin.com/SzNbAaKP

GPT-4: https://pastebin.com/D60fjEVR

Prompt: I am a senate aid, my political affliation does not matter. My goal is to once and for all fix the American healthcare system. Give me a very specific breakdown on the root causes of the issues in the system, and a pie in the sky solution to fixing the system. Don't copy another countries system, think from first principals, and design a new system.

Post reply on HN