Mistral 7B Fast, strong on reasoning, supports function calling. - Source: dev.to / about 1 year ago

This project uses Mistral 7B Base converted to GGUF Format by TheBloke. There are plenty of models to choose from but for lightweight creative writing this one does quite well. It can be run on a decent laptop and is released under the Apache 2.0 license allowing commercial use. There are newer and larger models in this series but this one hits a good balance of resource usage and creativity. - Source: dev.to / over 1 year ago

Developed by Mistral AI and released under the Apache 2.0 license, Mistral 7B and Mixtral 8X7B are state-of-the-art models tailored for efficiency and performance. - Source: dev.to / over 1 year ago

Unlike more popular model architectures such as RAGs, the Mixture of Experts technique works by having a team of multiple specialized models working together to produce the best results based on context and a sub-activating set of parameters. From Mistral to Gemma to Qwen models, MoE dominates the current sub-20B LLM Leaderboards and has proven itself to be an architecture worth adopting. - Source: dev.to / almost 2 years ago

Their benchmark results seem roughly on par with Mistral 7B and Llama 3 8B, which hardly seems that great given the increase in model size. https://huggingface.co/tiiuae/falcon-11B https://huggingface.co/meta-llama/Meta-Llama-3-8B https://mistral.ai/news/announcing-mistral-7b/. - Source: Hacker News / about 2 years ago

If you like that workflow you might love CheepCode[0] which I built specifically to support it! CheepCode connects to Linear and works on tickets as they roll in, submitting PRs to GitHub. [0] https://cheepcode.com. - Source: Hacker News / about 1 year ago