Today's LLM & Language Models: Fastest-Growing Projects — April 20, 2026
Today's the LLM & Language Models space, we've seen a surge in tools that aim to make large language models more accessible and user-friendly. Many repositories are gaining traction by providing innovative solutions for working with LLMs, from making them talk like normal people to compiling personal knowledge bases.
The fastest-growing repository this week is arman-bd/guppylm, with a growth score of 87.27 and over 2,955 stars. This ~9M parameter LLM talks like a small fish, and its unique personality has captured the attention of developers and users alike. Its high growth rate suggests that there's a strong interest in experimenting with language models that have distinct tones and styles.
hexiecs/talk-normal is another repository that's gaining popularity, with a growth score of 74.04 and over 1,333 stars. This tool makes any LLM talk like a normal person by removing AI slop, which has obvious applications in chatbots and virtual assistants. Its high commit activity (56 commits in the last 30 days) indicates that developers are actively working to improve its functionality.
sdyckjq-lab/llm-wiki-skill is growing steadily, with a growth score of 57.73 and over 904 stars. This repository provides a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. Its high commit activity (71 commits in the last 30 days) suggests that it's an actively maintained project.
amitshekhariitbhu/llm-internals is also worth mentioning, with a growth score of 48.25 and over 574 stars. This repository provides step-by-step explanations of LLM internals, from tokenization to attention to inference optimization. Its steady growth rate indicates that there's a strong interest in understanding the inner workings of language models.
Other notable repositories include kessler/gemma-gem (growth score: 39.30, over 792 stars), which runs Google's Gemma 4 model entirely on-device via WebGPU; xoai/sage-wiki (growth score: 39.09, over 435 stars), an LLM-compiled personal knowledge base that compiles papers and articles into a structured wiki; and lucasastorian/llmwiki (growth score: 32.62, over 561 stars), an open-source implementation of Karpathy's LLM Wiki.
Lastly, Pratiyush/llm-wiki (growth score: 32.46, over 128 stars) is another repository that implements the LLM Wiki pattern, providing a knowledge base from Claude Code and other sessions. Its high commit activity (100 commits in the last 30 days) suggests that it's an actively maintained project.
Overall, Today's trends suggest that there's a growing interest in making language models more accessible and user-friendly, as well as understanding their inner workings.
The fastest-growing repository this week is arman-bd/guppylm, with a growth score of 87.27 and over 2,955 stars. This ~9M parameter LLM talks like a small fish, and its unique personality has captured the attention of developers and users alike. Its high growth rate suggests that there's a strong interest in experimenting with language models that have distinct tones and styles.
hexiecs/talk-normal is another repository that's gaining popularity, with a growth score of 74.04 and over 1,333 stars. This tool makes any LLM talk like a normal person by removing AI slop, which has obvious applications in chatbots and virtual assistants. Its high commit activity (56 commits in the last 30 days) indicates that developers are actively working to improve its functionality.
sdyckjq-lab/llm-wiki-skill is growing steadily, with a growth score of 57.73 and over 904 stars. This repository provides a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. Its high commit activity (71 commits in the last 30 days) suggests that it's an actively maintained project.
amitshekhariitbhu/llm-internals is also worth mentioning, with a growth score of 48.25 and over 574 stars. This repository provides step-by-step explanations of LLM internals, from tokenization to attention to inference optimization. Its steady growth rate indicates that there's a strong interest in understanding the inner workings of language models.
Other notable repositories include kessler/gemma-gem (growth score: 39.30, over 792 stars), which runs Google's Gemma 4 model entirely on-device via WebGPU; xoai/sage-wiki (growth score: 39.09, over 435 stars), an LLM-compiled personal knowledge base that compiles papers and articles into a structured wiki; and lucasastorian/llmwiki (growth score: 32.62, over 561 stars), an open-source implementation of Karpathy's LLM Wiki.
Lastly, Pratiyush/llm-wiki (growth score: 32.46, over 128 stars) is another repository that implements the LLM Wiki pattern, providing a knowledge base from Claude Code and other sessions. Its high commit activity (100 commits in the last 30 days) suggests that it's an actively maintained project.
Overall, Today's trends suggest that there's a growing interest in making language models more accessible and user-friendly, as well as understanding their inner workings.