AI & machine learning2021 · shared on Hacker News

Evaluating Large Language Models Trained on Code

Explore Evaluating Large Language Models Trained on Code, its original source and Hacker News discussions. Discover community perspectives and save it to your reading list.

The conversation around the idea.

Links found in the discussion

All public GitHub code was used in training Copilot

GitHub Copilot, with “public code” blocked, emits my copyrighted code

Mercury: Ultra-fast language models based on diffusion

AI is stifling new tech adoption?

OpenAI Codex

GitHub Copilot Generated Insecure Code in 40% of Circumstances During Experiment

Z3 approach to discover that “q_rsqrt” is in Copilot's slur list

GPT-3 APIs are now in public beta

A neural network solves and generates mathematics problems by program synthesis

AlphaFold won’t revolutionise drug discovery

ML code generation vs. coding by hand: what we think programming will look like

DeepSeekMath 7B achieved 51.7% on MATH benchmark

Qwen1.5-110B

Solving Advent of Code Puzzles with GitHub Copilot: Day 1

Does RL Incentivize Reasoning in LLMs Beyond the Base Model?

Show HN: Phind Model beats GPT-4 at coding, with GPT-3.5 speed and 16k context

A new way to predict when software jobs will become automated

Evaluating Large Language Models Trained on Code

Evaluating Large Language Models Trained on Code

Evaluating Large Language Models Trained on Code (paper about GH copilot model)

Evaluating Large Language Models Trained on Code(GitHub Copilot)

Replit's new Code LLM: Open Source, 77% smaller than Codex, trained in 1 week

GPT-Migrate converts repos from one lang/framework to another

Refact LLM: New 1.6B code model reaches 32% HumanEval and is SOTA for the size

Evaluating Large Language Models Trained on Code

What OpenAI Really Wants

One idea leads to another.