LLM-jp-3 13B Base
LLM-jp-3 13B Base is the 13B pretrained text model in LLM-jp's generation-3 size series. The underlying model was exposed to 2.1 trillion pretraining tokens.
Research release
The size changes the architecture
The 40-layer model has hidden width 5,120 and 40 attention heads. Pretraining includes Japanese web and archive material, English Dolma sources and code from The Stack.LLM-jp-3 13B Base official model card
Use the size as an experimental control
Practical guidance. Use this as the larger Apache-licensed dense generation-3 baseline. Keep the exact revision in evaluation records because the instruction releases use different alignment steps.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | llm-jp-3-13b |
| Context window | 4,096 tokens Maximum sequence length stated in the model card. LLM-jp-3 13B Base official model card |
| Architecture | 40 layers; hidden size 5,120; 40 attention heads.LLM-jp-3 13B Base official model card |
| Parameters | 13,707,924,480 total, from the card's embedding and non-embedding counts.LLM-jp-3 13B Base official model card |
| License | Apache 2.0; research weights.LLM-jp-3 13B Base official model card |
Related models
- LLM-jp-3 13B Instruct3
LLM-jp / NII LLMC ยท LLM-jp-3
LLM-jp-3 13B Instruct3 is a preference-aligned instruction release with 4,096-token context and Apache 2.0 weights.
Launch date not confirmedResearch releaseDownloadable