I’ve been experimenting with LLM’s recently and found a neat tool to easily run LLM’s locally on your machine. It’s a python library called LLM.
Install it with pip install llm
. You can now interact with openai models, but if you
want can install gpt4all to run local models.
Install the gpt4all plugin llm install gpt4all
See available models with llm models
and pick one you’d like to try.
To download a model to your machine and run a query against it:
llm -m replit-code-v1_5-3b-q4_0 "replace all commas in a string in plain javascript"
The output is hilarious:
var str = "Hello,,world";
console.log(str); // Hello,,,,world
function replaceAllComasInStringWithoutUsingLibrariesOrFrameworksWithoutReplacingTheCommaCharacterWithAnyOtherCharacters() {
return (
String("" + arguments[0])
.split(",")
/*replace comma character with empty string*/
-> Array<string>
:: map(s => s.trim())
| join("")
//return a single value as the result of this function call, instead if you want to return an array or object then use "->" in place and put your logic between brackets "{}".
);
}
console.log("Hello,,world");
You can even create an alias for easier access:
llm aliases set replit replit-code-v1_5-3b-q4_0
To interact with OpenAI models, set your API key with: llm keys set openai
Now you can query GPT with llm "replace all commas in a string with plain javascript"
To replace all commas in a string with JavaScript, you can use the `replace()` method with
a regular expression as the first argument and the replacement character or string as the
second argument.
Here's an example:
```javascript
var str = "Hello, world, this, is, an, example";
// Replace all commas with a space
var newStr = str.replace(/,/g, ' ');
console.log(newStr); // "Hello world this is an example"
In the above example, the regular expression /,/g
matches all commas in the string
str
. The g
flag stands for “global” and ensures that all occurrences of the comma will
be replaced, not just the first one. The ' '
is the replacement string, which in this
case is a space.
```
This response is definitely way more useful.
Another neat trick is that you can analyze code with openai:
cat file.rb | llm -s "Explain this code"