Загрузка видео...

Не удалось загрузить видео

На главную

I put an entire codebase in a text file and gave it to Gemini 1.5 Pro. I gave it a task that required updating ~10 files across the entire codebase - so decently complex. It got about 80% of the work done for me on a 1st attempt. Watch...

mckaywrigley's profile picture

Mckay Wrigley

228,348 subscribers

85,724 просмотров • 2 лет назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 9

Фото профиля Mckay Wrigley
Mckay Wrigley2 лет назад

This is a very realistic example. Highlights how nice the context window is for code-related work while being honest about LLMs needing better reasoning capabilities to get to 100%. Super happy with the result! Lots of gains to be had from prompting/retrieval on top of this.

Фото профиля FUTURESaiLOR ⚓ 🌊
FUTURESaiLOR ⚓ 🌊2 лет назад

This looks interesting:

Фото профиля Mckay Wrigley
Mckay Wrigley2 лет назад

If they actually have a reasoning breakthrough then wow

Фото профиля Tom Bielecki
Tom Bielecki2 лет назад

Gemini context + Cursor RAG + Agentic testing loops

Фото профиля Giordano Sugar
Giordano Sugar2 лет назад

Thank you so much for this in depth example and your full, honest explanation. Really gives me a full sense of what to expect.

Фото профиля Tribhuvan Joshi
Tribhuvan Joshi2 лет назад

This is quite amazing. Eagerly waiting for its public release.

Фото профиля efwerr
efwerr2 лет назад

thanks for the honest evaluation

Фото профиля Andrew!
Andrew!2 лет назад

Did you just copy and paste your codebase into the txt file? Or a script or something

Фото профиля Mckay Wrigley
Mckay Wrigley2 лет назад

@AndrewW42079596 py script

Похожие видео

The future of fixing bugs? Just record them. I filmed 3 separate bugs in an app and gave the videos to Gemini 1.5 Pro with my entire codebase. It correctly identified & fixed each one. AI is improving insanely fast.
3:14

The future of fixing bugs? Just record them. I filmed 3 separate bugs in an app and gave the videos to Gemini 1.5 Pro with my entire codebase. It correctly identified & fixed each one. AI is improving insanely fast.

Mckay Wrigley

1,138,035 просмотров • 2 лет назад

I found an excellent use case for Remotion & claude code - get it to create a video explaining how the codebase works. In this case I got it to explain how codebase works
1:15

I found an excellent use case for Remotion & claude code - get it to create a video explaining how the codebase works. In this case I got it to explain how codebase works

Peter Gostev

15,379 просмотров • 6 месяцев назад

🔥 💻 🎥 How to provide o1 Pro with your FULL CODEBASE through the ChatGPT Cursor connection yesterday I recorded a video connecting o1 Pro to Cursor through the ChatGPT Desktop app more than one person commented on how limiting it is for o1 Pro to only have access to a single open file in Cursor here's a walkthrough of how you can provide o1 Pro with your FULL CODEBASE as context instead of just a single file: 1. Write a Python function that concatenates full codebase into a single snapshot in a .txt file (link in comment) 2. Open that "codebase_snapshot.txt" in a separate Cursor pane 3. Go to the ChatGPT app and ask it what context it has access to through Cursor - should say it has two files with one of them being the codebase snapshot and boom there ya go. o1 Pro has access to everything in your codebase full walkthrough here 👇
6:14

🔥 💻 🎥 How to provide o1 Pro with your FULL CODEBASE through the ChatGPT Cursor connection yesterday I recorded a video connecting o1 Pro to Cursor through the ChatGPT Desktop app more than one person commented on how limiting it is for o1 Pro to only have access to a single open file in Cursor here's a walkthrough of how you can provide o1 Pro with your FULL CODEBASE as context instead of just a single file: 1. Write a Python function that concatenates full codebase into a single snapshot in a .txt file (link in comment) 2. Open that "codebase_snapshot.txt" in a separate Cursor pane 3. Go to the ChatGPT app and ask it what context it has access to through Cursor - should say it has two files with one of them being the codebase snapshot and boom there ya go. o1 Pro has access to everything in your codebase full walkthrough here 👇

Dan McAteer

99,719 просмотров • 1 год назад

🤯 this genius stores his entire codebase syntax in a graph database and queries it so provide context to an llm
2:15

🤯 this genius stores his entire codebase syntax in a graph database and queries it so provide context to an llm

Dan McAteer

1,991,798 просмотров • 1 год назад

Hermes agent is insane I gave Hermes access to Browser Use so it has access to my social media accounts. It retains context about my codebase, tone, and workflows. Now i just ask it to post on X for me...
0:23

Hermes agent is insane I gave Hermes access to Browser Use so it has access to my social media accounts. It retains context about my codebase, tone, and workflows. Now i just ask it to post on X for me...

shawn

121,027 просмотров • 4 месяцев назад

o1-pro is probably the best model i've used for coding, hands down i gave it a pretty complicated codebase and asked it to refactor while referencing docs the difference between claude/gemini/o1 and o1 pro is night and day. first time in a while i've been this impressed. full comparison in the video + code.
7:33

o1-pro is probably the best model i've used for coding, hands down i gave it a pretty complicated codebase and asked it to refactor while referencing docs the difference between claude/gemini/o1 and o1 pro is night and day. first time in a while i've been this impressed. full comparison in the video + code.

Sully

255,372 просмотров • 1 год назад

Hermes learned my entire repo in minutes. I’m not exaggerating. I pointed the new /learn skill at my codebase… and watched it turn the repo into a reusable skill. Files. Patterns. Architecture. Commands. Workflows. It didn’t just “read” the repo. It created a playbook for how to work with it again later. This blew my mind. Memory remembers facts. Skills remember how to do the work. That’s the difference.
0:40

Hermes learned my entire repo in minutes. I’m not exaggerating. I pointed the new /learn skill at my codebase… and watched it turn the repo into a reusable skill. Files. Patterns. Architecture. Commands. Workflows. It didn’t just “read” the repo. It created a playbook for how to work with it again later. This blew my mind. Memory remembers facts. Skills remember how to do the work. That’s the difference.

Luke The Dev

48,867 просмотров • 1 месяц назад

Claude Code with Sonnet 4.5 is actually incredible I gave it a prompt for a super complex app, and it one shot the entire thing In this video I walk you through how to use Claude Code to build a prompt library app you can start using immediately (no coding experience required):
19:42

Claude Code with Sonnet 4.5 is actually incredible I gave it a prompt for a super complex app, and it one shot the entire thing In this video I walk you through how to use Claude Code to build a prompt library app you can start using immediately (no coding experience required):

Alex Finn

40,955 просмотров • 10 месяцев назад

🐱ah right, actually my cover of “ordinary people” i was there for & participated in the editing of the entire video & gave a lot of my opinions on, so it's a project that i was right there for, from the start to the end. i filmed everything needed very well, so it was so nice. please watch it a lot!
0:35

🐱ah right, actually my cover of “ordinary people” i was there for & participated in the editing of the entire video & gave a lot of my opinions on, so it's a project that i was right there for, from the start to the end. i filmed everything needed very well, so it was so nice. please watch it a lot!

🍰

36,085 просмотров • 1 год назад

Loop Engineering is getting hype now. But not many talks about how to actually do it So I open-sourced the template my team uses to build agent loops: - a shared artifact / knowledge layer - logging, verification - and a codebase harness so work compounds across runs Plus a 20-min deep dive on how to think about it and set it up for real: Copy the template. Adapt it to your own loops.
18:58

Loop Engineering is getting hype now. But not many talks about how to actually do it So I open-sourced the template my team uses to build agent loops: - a shared artifact / knowledge layer - logging, verification - and a codebase harness so work compounds across runs Plus a 20-min deep dive on how to think about it and set it up for real: Copy the template. Adapt it to your own loops.

Jason Zhou

391,345 просмотров • 1 месяц назад

I kind of like Gemini 3.0 Flash in Antigravity. It is fast, and the Browser Subagent works great with it. A few days ago, I asked it to rebuild my website entirely based on the existing codebase. It redesigned everything and tested it too. I did not have much time for iteration, but it was good enough, and I deployed it.
0:38

I kind of like Gemini 3.0 Flash in Antigravity. It is fast, and the Browser Subagent works great with it. A few days ago, I asked it to rebuild my website entirely based on the existing codebase. It redesigned everything and tested it too. I did not have much time for iteration, but it was good enough, and I deployed it.

AshutoshShrivastava

41,433 просмотров • 7 месяцев назад

Wow. This takes "chatPDF" to an entirely new level. Don't think this is possible with regular rag either I gave 4 highly tabular pdfs to gemini 1.5 (tesla financials) it understood the entire context, summed all the tabular values and gave me a nearly perfect response.
2:31

Wow. This takes "chatPDF" to an entirely new level. Don't think this is possible with regular rag either I gave 4 highly tabular pdfs to gemini 1.5 (tesla financials) it understood the entire context, summed all the tabular values and gave me a nearly perfect response.

Sully

85,590 просмотров • 2 лет назад

I used the new Gemini CLI and gave it only the YouTube link of a tech tutorial, and it set up the entire project for me, perfectly. This is a massive win for productivity and builders everywhere. Check it out here:
2:17

I used the new Gemini CLI and gave it only the YouTube link of a tech tutorial, and it set up the entire project for me, perfectly. This is a massive win for productivity and builders everywhere. Check it out here:

Smitha Kolan

107,998 просмотров • 1 год назад

In an effort for me to say less and work more, our team put together a video about how I ended up working on @DoubleZero in the first place. I’ve been reducing latency my entire life so when presented with original ideas behind it, I couldn’t say no.
6:04

In an effort for me to say less and work more, our team put together a video about how I ended up working on @DoubleZero in the first place. I’ve been reducing latency my entire life so when presented with original ideas behind it, I couldn’t say no.

Andrew McConnell

18,315 просмотров • 1 месяц назад

Next up, letter W: “Where I’m From” #JayZ30 #JayZ Deric “D-Dot” Angelettie (a.k.a. Madd Rapper): “Ron Lawrence created the track from a live record. D-Dot came in and put the... that’s when the magic happens, when I’m there [in the studio]” Ron “Amen-Ra” Lawrence: “When Jay heard it, it was just in a raw form. It was really nothing to it; it was just bare. ...By Jay putting his lyrics on that track, it really gave me a vision on where I knew that I could take it because he painted that picture about living in Marcy Projects. He gave it that whole sinister feel. Most people want to have a finished track so they can rhyme to [it]. It’s very challenging, and it’s a lot of fun if you can have a rapper do a song on top of a song that's not finished ... it just gave me so much insight on what I should do with it, and it worked.”
5:33

Next up, letter W: “Where I’m From” #JayZ30 #JayZ Deric “D-Dot” Angelettie (a.k.a. Madd Rapper): “Ron Lawrence created the track from a live record. D-Dot came in and put the... that’s when the magic happens, when I’m there [in the studio]” Ron “Amen-Ra” Lawrence: “When Jay heard it, it was just in a raw form. It was really nothing to it; it was just bare. ...By Jay putting his lyrics on that track, it really gave me a vision on where I knew that I could take it because he painted that picture about living in Marcy Projects. He gave it that whole sinister feel. Most people want to have a finished track so they can rhyme to [it]. It’s very challenging, and it’s a lot of fun if you can have a rapper do a song on top of a song that's not finished ... it just gave me so much insight on what I should do with it, and it worked.”

Luis

13,694 просмотров • 1 месяц назад

I fed an *entire* biology textbook into Gemini 1.5 Pro. 491,002 tokens. I asked it 3 extremely specific questions, and it got each answer 100% correct. 1M token context windows are a gamechanger.
5:13

I fed an *entire* biology textbook into Gemini 1.5 Pro. 491,002 tokens. I asked it 3 extremely specific questions, and it got each answer 100% correct. 1M token context windows are a gamechanger.

Mckay Wrigley

627,976 просмотров • 2 лет назад

GPT Engineer github: Specify what you want it to build, the AI asks for clarification, and then builds it. GPT Engineer is made to be easy to adapt, extend, and make your agent learn how you want your code to look. It generates an entire codebase based on a prompt.
1:01

GPT Engineer github: Specify what you want it to build, the AI asks for clarification, and then builds it. GPT Engineer is made to be easy to adapt, extend, and make your agent learn how you want your code to look. It generates an entire codebase based on a prompt.

AK

349,149 просмотров • 3 лет назад

i watched gemma 4 12b build something genuinely impressive today, and then loop itself to death right in front of me. the full run is in the video, sped up but completely uncut, watch it to the end and you will catch the exact moment it stops building and starts looping right in the middle of the work. the task was clean, build a single file gravity simulator, n-body physics, orbits, collisions, running locally on one 3090 through an agent. and for ten minutes it was a joy to watch. it reached for a symplectic integrator on its own, the correct one, the kind that keeps orbits stable instead of spiralling out. real gravity with softening, proper orbital velocities, momentum conserved on collision. the physics was right. the thing actually worked. then on the very last step, writing a few tests to prove its own code, it fell into a loop. not a crash, a loop. it started repeating itself and would not stop. ten more minutes, thirty four thousand tokens into a single answer, the same fragments over and over, until i killed it myself. so it's not that gemma can't code. it did the hard part beautifully. it cannot finish. it cannot hold a long task together without unravelling, and finishing is the entire job in agentic work. here's the part that stings. i run this exact task, same harness, same card, on the chinese open models, qwen especially, and i never see this. they build it, they test it, they stop. every single time. google has the raw capability, you can see it sitting right there in the code, and then the model loops itself to death on a task a 27b from alibaba finishes clean. open weights, apache 2.0, so much to love on paper. i just need it to know when to stop talking.
4:15

i watched gemma 4 12b build something genuinely impressive today, and then loop itself to death right in front of me. the full run is in the video, sped up but completely uncut, watch it to the end and you will catch the exact moment it stops building and starts looping right in the middle of the work. the task was clean, build a single file gravity simulator, n-body physics, orbits, collisions, running locally on one 3090 through an agent. and for ten minutes it was a joy to watch. it reached for a symplectic integrator on its own, the correct one, the kind that keeps orbits stable instead of spiralling out. real gravity with softening, proper orbital velocities, momentum conserved on collision. the physics was right. the thing actually worked. then on the very last step, writing a few tests to prove its own code, it fell into a loop. not a crash, a loop. it started repeating itself and would not stop. ten more minutes, thirty four thousand tokens into a single answer, the same fragments over and over, until i killed it myself. so it's not that gemma can't code. it did the hard part beautifully. it cannot finish. it cannot hold a long task together without unravelling, and finishing is the entire job in agentic work. here's the part that stings. i run this exact task, same harness, same card, on the chinese open models, qwen especially, and i never see this. they build it, they test it, they stop. every single time. google has the raw capability, you can see it sitting right there in the code, and then the model loops itself to death on a task a 27b from alibaba finishes clean. open weights, apache 2.0, so much to love on paper. i just need it to know when to stop talking.

Sudo su

39,574 просмотров • 2 месяцев назад

Kevin O’Leary says he wept when Rolex gave him a watch even its own staff had never seen. “Grown men weep just looking at pictures of this mystical beast. Does it exist? I got on my knees and begged for it.” “The morning before the Oscars, the phone rings. They gave me an address and said, ‘Be there.’ And there it was.” “The entire Rolex staff had never seen it. All the dealers from around LA came to see it.” “For an hour we looked at it. We didn’t touch it. We just looked at it and said, ‘It’s real.’” “I did, and I wore it. I was weeping. It was mine. I own it. Do you know how many people have that piece? Maybe nobody.”
1:12

Kevin O’Leary says he wept when Rolex gave him a watch even its own staff had never seen. “Grown men weep just looking at pictures of this mystical beast. Does it exist? I got on my knees and begged for it.” “The morning before the Oscars, the phone rings. They gave me an address and said, ‘Be there.’ And there it was.” “The entire Rolex staff had never seen it. All the dealers from around LA came to see it.” “For an hour we looked at it. We didn’t touch it. We just looked at it and said, ‘It’s real.’” “I did, and I wore it. I was weeping. It was mine. I own it. Do you know how many people have that piece? Maybe nobody.”

Umar 🐂🀄️

6,270,944 просмотров • 9 дней назад