The DeepSeek v3 paper (and are out, after yesterday's mysterious launch of Loads of interesting particulars in right here. More analysis results may be found right here. This is probably solely model specific, so future experimentation is required right here. This model is a high quality-tuned 7B parameter LLM on the Intel Gaudi 2 processor from the Intel/neural-chat-7b-v3-1 on the meta-math/MetaMathQA dataset. The Intel/neural-chat-7b-v3-1 was originally fine-tuned from mistralai/Mistral-7B-v-0.1. 1.3b-instruct is a 1.3B parameter model initialized from deepseek-coder-1.3b-base and fine-tuned on 2B tokens of instruction knowledge.