jouniPretty wild night with LLMs. You don't need Gitlab Copilot, there is offline running solutions ready: https://continue.dev/ - alternative to GPT4ALL https://ollama.ai/ - check the models! and heartwarming full-c++ implementation of LLaMA : LLaMA.cpp https://github.com/ggerganov/llama.cpp (works on OpenCL and Radeon GPUs). #llm #largelanguagemodel #largelanguagemodels #llama #llama2
@opny721
- 4 Followers
- 14 Following
- 7 Posts
Mistral 7B
Grouped query attention and sliding window attention increase efficiency and performance of LLMs. Source of training data not disclosed, but the model is released under a permissible license.
My summary on HFPapers: https://huggingface.co/papers/2310.06825#6527a2be0ef49cfb784b936f
@derekvanvliet doing the same, failed often. Keep trying. Keep it going!
@adrienne this is really funny indeed...wtf
@parismarx this will benefit just who can afford A100 GPU or who produce them. Luckily opens source alternatives are growing
@ninokadic yeah I do agree, also the quality of the content was degrading. Let's see how it goes here, seems is starting well 
@ninokadic hey did the same, that X hit badly