[PR #21] [CLOSED] Updating model to gpt-4o-mini #105

New issue

Closed

opened 2026-03-03 13:53:06 +03:00 by kerem · 0 comments

kerem commented

2026-03-03 13:53:06 +03:00

Owner

📋 Pull Request Information

Original PR: https://github.com/jehna/humanify/pull/21
Author: @bilalba
Created: 7/29/2024
Status: ❌ Closed

Base: main ← Head: main

📝 Commits (10+)

188ef95 Add npm init
26e14f8 Add TypeScript
33c35cc Add Prettier
dd2e033 Add babel
4806fc3 Add not for ignoringa step
261e919 Add gpt-3.5 step for renaming variables
7bffb40 Add the code mode readable using small transforms
0eb8b09 Add plugin, remove duplicate code
cbe8478 Add makeNumbersLonger plugin
9f526a6 Add CLI interface

📊 Changes

1 file changed (+0 additions, -4 deletions)

View changed files

📝 src/openai/openai.ts (+0 -4)

📄 Description

Inspired by https://github.com/Acters/humanify-ng/tree/GPT4o, I decided to make a PR after testing some of the changes from there.

The openai sdk is updated to version 4.52.7 addressing #19

gpt-tokenizer is used instead of gpt3-encoder addressing #5

I am seeking to address #4 with a good enough solution of using gpt-4o-mini with the assumption that it would be intelligent enough.

I have updated the model to gpt-4o-mini and removed the option to use 4k context, since gpt-4o-mini is already cheap enough and is not constrained by context length.

However, I have still capped the context at 32k. In my testing, Using a larger context would not result in enough variable name replacements. I.e. it would only change a few function and variable names.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/jehna/humanify/pull/21 **Author:** [@bilalba](https://github.com/bilalba) **Created:** 7/29/2024 **Status:** ❌ Closed **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (10+) - [`188ef95`](https://github.com/jehna/humanify/commit/188ef957e7756cccfc1769b48fb3864cd86447af) Add npm init - [`26e14f8`](https://github.com/jehna/humanify/commit/26e14f8b0f75ef8f01c2e5364b9e702137d45846) Add TypeScript - [`33c35cc`](https://github.com/jehna/humanify/commit/33c35cc2d15a8932c447f404b33ee13dfc73097b) Add Prettier - [`dd2e033`](https://github.com/jehna/humanify/commit/dd2e033f251cdfeb0c772f7a45889348dad9612f) Add babel - [`4806fc3`](https://github.com/jehna/humanify/commit/4806fc393d64c098431672bc05836f87e9e09acd) Add not for ignoringa step - [`261e919`](https://github.com/jehna/humanify/commit/261e919839ec928642de5aaff902ac76d4e63600) Add gpt-3.5 step for renaming variables - [`7bffb40`](https://github.com/jehna/humanify/commit/7bffb40364db532274c9e542bf0121c22f535f5e) Add the code mode readable using small transforms - [`0eb8b09`](https://github.com/jehna/humanify/commit/0eb8b095f47d2dbfe25ca3b9e2cf785a37beae44) Add plugin, remove duplicate code - [`cbe8478`](https://github.com/jehna/humanify/commit/cbe8478336050865495c0450e15c327ed81f3cba) Add makeNumbersLonger plugin - [`9f526a6`](https://github.com/jehna/humanify/commit/9f526a66169cd26f253cba9f63f884074924075a) Add CLI interface ### 📊 Changes **1 file changed** (+0 additions, -4 deletions) <details> <summary>View changed files</summary> 📝 `src/openai/openai.ts` (+0 -4) </details> ### 📄 Description Inspired by https://github.com/Acters/humanify-ng/tree/GPT4o, I decided to make a PR after testing some of the changes from there. The openai sdk is updated to version 4.52.7 addressing #19 gpt-tokenizer is used instead of gpt3-encoder addressing #5 I am seeking to address #4 with a good enough solution of using gpt-4o-mini with the assumption that it would be intelligent enough. I have updated the model to gpt-4o-mini and removed the option to use 4k context, since gpt-4o-mini is already cheap enough and is not constrained by context length. However, I have still capped the context at 32k. In my testing, Using a larger context would not result in enough variable name replacements. I.e. it would only change a few function and variable names. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>