Bilibili open sourced its lightweight AI language model Index-1.9B, which attracted widespread attention in the industry. The model contains multiple versions designed to meet the needs of different application scenarios and provides a comparative study of benchmark models. The move of open source makes it easier for developers to conduct research and application, and promotes technological development in the field of AI. The launch of the Index-1.9B model provides developers with more choices and makes it possible for AI technology to be applied in more fields.
Station B announced that it has open sourced the Index-1.9B model. The launch of this lightweight AI language model has attracted widespread attention. The Index-1.9B model contains several versions, including a pedestal model, a control group, and a conversation model. 1.9B includes: Index-1.9B base: base model, with 1.9 billion non-word embedding parameters, pre-trained on 2.8T Chinese and English corpus, and is ahead of models of the same level on multiple evaluation benchmarks Index- 1.9B pure : The control group of the base model has the same parameters and training strategy as the base. The difference is that the data related to all instructions in this version of the corpus are strictly filtered to verify the impact of the instructions on the benchmark. Index-1.9B chat: based on index-1.9B base dialogue model aligned with SFT and DPO. Since more Internet community corpus is introduced in pre-training, chatting is significantly more interesting. Index-1.9B character: RAG is introduced based on SFT and DPO to realize role-play customization of fewshots
Index-1.9B product entrance: https://top.aibase.com/tool/index-1-9b
The open source of the Index-1.9B model provides strong support for the popularization and application of AI technology. Its different versions also meet the needs of different users. For example, the dialogue model focuses more on fun, and the role-playing model focuses more on personalized customization. This move by Station B has undoubtedly contributed to the AI community, and it is worth looking forward to subsequent updates and developments.