Sometimes I have it write functions that are very boilerplate to save time, but I mostly like to use it as a tool to think through problems, among other tools like writing in a notebook or drawing diagrams. I enjoy programming too much that I’d want an AI to do it all for me (it also helps that I don’t do it as a job though).
Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
201–210 of 336 posts
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#202I had huge success letting Gemini 2.5 oneshot whole codebases in a single text file format and then split it up with a script. It's putting in work for like 5 minutes and spits out a working codebase, I also asked it to show of a little bit and it almost one shotted a java cloud service to generate pdf invoices from API calls, (made some minor mistakes but after feeding them back it fixed them) I basically use two sc…
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#203Earlier quoted context omitted.
This is also compounded by the fact that LLMs are not deterministic, every response is different for the same given prompt. And people tend to judge on one off experiences.
> LLMs are not deterministic They can be. The cloud-hosted LLMs add a gratuitous randomization step to make the output seem more human. (In vein with the moronic idea of selling LLM's as sci-fi human-like assistants.) But you don't have to add those randomizations. Nothing much is lost if you don't. (Output from my self-hosted LLM's is deterministic.)
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#204The loser in the AI model competition appears to be... Microsoft. When ChatGPT was the only game in town Microsoft was seen as a leader, thanks to their wise investment in Open AI. They relied on Open AI's model and didn't develop their own. As a result Microsoft has no interesting AI products. Copilot is a flop. Bing failed to take advantage of AI, Perplexity ate their lunch. Satya Nadella last year: “Google should…
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#205Earlier quoted context omitted.
Note that Microsoft do have their own LLM team, and their own model called Phi-4. https://huggingface.co/microsoft/phi-4
Recently I was looking for a small LLM that could perform reasonably well while answering questions with low latency, for near realtime conversations running on a single RTX 3090. I settled on Microsoft’s Phi-4 model so far. However I’m not sure yet if my choice is good and open to more suggestions!
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#206The loser in the AI model competition appears to be... Microsoft. When ChatGPT was the only game in town Microsoft was seen as a leader, thanks to their wise investment in Open AI. They relied on Open AI's model and didn't develop their own. As a result Microsoft has no interesting AI products. Copilot is a flop. Bing failed to take advantage of AI, Perplexity ate their lunch. Satya Nadella last year: “Google should…
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#207The loser in the AI model competition appears to be... Microsoft. When ChatGPT was the only game in town Microsoft was seen as a leader, thanks to their wise investment in Open AI. They relied on Open AI's model and didn't develop their own. As a result Microsoft has no interesting AI products. Copilot is a flop. Bing failed to take advantage of AI, Perplexity ate their lunch. Satya Nadella last year: “Google should…
When my parent speak about AI, they call it Copilot. Mircosoft has a big Advantage that they can integrate AI in many daily used products, where it is not competing with their core product like Google
these days it seems like everyone is trying to get their AI to be the standard.
i wonder how things will look in 10 years.
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#208The loser in the AI model competition appears to be... Microsoft. When ChatGPT was the only game in town Microsoft was seen as a leader, thanks to their wise investment in Open AI. They relied on Open AI's model and didn't develop their own. As a result Microsoft has no interesting AI products. Copilot is a flop. Bing failed to take advantage of AI, Perplexity ate their lunch. Satya Nadella last year: “Google should…
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#209Earlier quoted context omitted.
Recently I was looking for a small LLM that could perform reasonably well while answering questions with low latency, for near realtime conversations running on a single RTX 3090. I settled on Microsoft’s Phi-4 model so far. However I’m not sure yet if my choice is good and open to more suggestions!
I've been using claude running via Ollama (incept5/llama3.1-claude) and I've been happy with the results. The only annoyance I have is that it won't search the internet for information because that capability is disabled via flag.
Re: Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison
#210I had huge success letting Gemini 2.5 oneshot whole codebases in a single text file format and then split it up with a script. It's putting in work for like 5 minutes and spits out a working codebase, I also asked it to show of a little bit and it almost one shotted a java cloud service to generate pdf invoices from API calls, (made some minor mistakes but after feeding them back it fixed them) I basically use two sc…
Can you please expound on this? You’re using this approach to turn an existing codebase into a single file and then asking Gemini to make changes/enhancements? Does it also handle breaking the files back out? Would love more info!
When providing the flat format it was able to replicate it without much instructions for a blank prompt i had success with the prompt below
===FILE=== Index: 1 Path: src/main/java/com/example/myapp/Greeter.java Length: 151 Content: package com.example.myapp;
public class Greeter { public String getGreeting() { return "Hello from the Greeter class!"; } } ===ENDFILE=== ===FILE=== Index: 2 Path: src/main/java/com/example/myapp/Main.java Length: 222 Content: package com.example.myapp;
public class Main { public static void main(String[] args) { Greeter greeter = new Greeter(); String message = greeter.getGreeting(); System.out.println("Main app says: " + message); } } ===ENDFILE=== ===FILE=== Index: 3 Path: pom.xml Length: 659 Content: http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"> 4.0.0
com.example
my-simple-app
1.0-SNAPSHOT
17
17
UTF-8
===ENDFILE===Prompt to request the format if starting from scratch: Present the entire codebase using the following multi-file format:
The codebase should be presented as a single, monolithic text output. Inside this output, represent each file of the project individually using the following structure:
Start Marker: Each file must begin with the exact line: ===FILE===
Metadata Block: Immediately following the start marker, include these four specific metadata lines, each on its own line:
Index: (where is a sequential integer index for the file, starting from 1).
Path: (The full relative path of the file from the project's root directory, e.g., index.html, css/style.css, js/script.js, jobs.html, etc.).
Length: (where is the exact character count of the file's content that follows).
Content: (This literal line acts as a separator).
File Content: Immediately after the Content: line, include the entire raw content of the file. Preserve all original line breaks, indentation, and formatting exactly as it should appear in the actual file.
End Marker: Each file's section must end with the exact line: ===ENDFILE===
Ensure all necessary files for the project (HTML, CSS, JS) are included sequentially within the single output block according to this structure.
Crucially, enclose the entire multi-file output, starting from the very first ===FILE=== line down to the very last ===ENDFILE=== line, within a single Markdown fenced code block using exactly five backticks (`````) on the lines immediately before the first ===FILE=== and immediately after the last `===ENDFILE===`. This ensures that any triple backticks (```) within the generated file content are displayed correctly.