Live data from Hacker News

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

rentry.co

121–130 of 380 posts

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

#121
I asked it to "Write me a short story", and it came up with this, which I thought was definitely different. Longer, mispellings (grandiothers), but much more interesting than I would typically see out of a LLM. Not sure what to think...

--output below--

In the heart of the bustling city of Thessaloniki, where the aroma of freshly brewed Greek coffee mingled with the salty breeze of the Thermaic Gulf, an old watchmaker named Elias operated a tiny shop that had seen better days. The shop, cluttered with clocks of every kind, from grandiothers to delicate wristwatches, was a capsule of time itself, untouched by the rush of the modern world.

Elias, a man of few words, wore the years on his face as gracefully as the patina on his timepieces. His hands, though steady when working on the intricate gears and springs, trembled slightly as he sipped his coffee in the quiet moments between the ticking.

One rainy morning, as the city awoke to the tapping of rain against the cobblestones, a young woman named Zoe hurried into Elias's shop, escaping a sudden downpour. Her appearance was like a breeze through stale air, her bright eyes scanning the ancient clocks with a curious sparkle.

“I need a gift,” Zoe announced, her voice cutting through the silence like the chime of a bell. “Something unique.”

Elias nodded, his gaze lingering slightly longer than usual. He sensed her distress, a hidden layer beneath her composed exterior. “Who is it for?” he asked, his voice as rough as the gears he polished.

“My father,” she replied, her fingers tracing the outline of a pocket watch displayed under the glass. “He’s retiring, and I want to give him something special to mark the end of his time working.”

“A good choice,” Elias said, picking up the pocket watch. “This watch is over a hundred years old. It stopped working a long time ago, much like its owner, perhaps waiting for the right moment to come alive again.”

Zoe smiled, intrigued. “Can you fix it?”

With a nod, Elias set to work. As his tools moved with precision, Zoe watched, seeing the care he took with each tiny screw and wheel. It was not just a watch he was mending, but weaving hope into an artifact left behind by time.

“Tell me about him, your father,” Elias said, as he worked.

“He’s always been punctual, a man of routine and schedules. Lately, I see him lost, unsure of what lies ahead when there’s no office to go to, no meetings to attend.”

“A new chapter,” Elias muttered, more to himself than to Zoe. “Time teaches us to move forward, even when our gears resist.”

When the watch was finally ticking again, its hands sweeping smoothly as if no time had passed since it last worked, Zoe’s eyes lit up. She listened to it, the sound a sweet melody to her anxious heart.

“How much do I owe you?” she asked, reaching for her purse.

“Nothing,” Elias said, closing her fingers around the watch. “Consider it a gift, for both you and your father. A reminder that time, even when it seems to stop, moves on, and so must we.”

Zoe left the shop with the watch securely in her hand, the weight of the gift heavier than its size suggested. When she gave her father the watch, his eyes reflected the same storms and sunsets that Elias’s did. The retirement party passed, speeches were made, and laughter filled the air, but the true gift was in the quiet moment Zoe and her father shared, listening to the steady tick of the old pocket watch, a heartbeat promising new beginnings.

Back in his shop, Elias sat alone, the tick-tock of countless clocks filling the room like a symphony. He didn’t need to see Zoe’s father’s face to know the impact of his gift. In fixing the watch, he had set something much greater into motion, a timeless reminder that life, like time, always finds a way to move forward. The little shop wasn’t just a place of business, but a haven where moments were kept alive through the magic of gears and springs, and where every repair mended more than just objects, but hearts too.

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

#125
post #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 sys…

They all did pretty well tbh. GPT2 didnt talk about moving away from fee for service which I think the evidence shows is the best idea, the other 2 did. GPT2 did have some other good ideas that the others didnt touch on though.

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

#126
post #40

Very impressive Prompt: > there are 3 black blocks on top of an block that we don't know the color of and beneath them there is a blue block. We remove all blocks and shuffle the blocks with one additional green block, then put them back on top of each other. the yellow block is on top of blue block. What color is the block we don't know the color of? only answer in one word. the color of block we didn't know the col…

llama-3-70B-instruct and GPT-4 both got it right for me

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

#127
post #119

Earlier quoted context omitted.

I'm asking it about how to make turbine blades for a high bypass turbofan engine and it's giving very good answers, including math and some very esoteric material science knowledge. Way past the point where the knowledge can be easily checked for hallucinations without digging into literature including journal papers and using the math to build some simulations. I don't even have to prompt it much, I just keep saying…

You mean it's giving very good sounding answers.

That's what I've observed, I gave it a task for a PoC on something I've been thinking about for a while and it's answer while syntactically correct is entirely useless (in the literal sense) due to it ignoring parts of the task.

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

#130
Plot twist!

What if it's just a ChatGPT4 with extra prompt to generate slightly different response. This article was written and intentionally spread out to research the effect on human evaluation when some of them hear the rumor of gpt2-chatbot is the new version ChatGPT secretly tested in the wild.

Post reply on HN