LLM-jp-3 172B beta1
LLM-jp-3 172B beta1 preserves the beta1 stage of LLM-jp's 172B model. The card records 700 billion seen tokens and a 4,096-token context.
Restricted access
A documented intermediate training stage
The beta release uses Japanese, English and code pretraining sources. It provides an intermediate training point before the final 172B release. Its gate should not be confused with a removed repository.LLM-jp-3 172B beta1 official model card
Keep the training stage in research records
Practical guidance. Use beta1 when the experiment needs that training stage. A later tokenizer, final checkpoint or different aligned version would change the comparison. Review the specific beta terms before distributing derived work.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Restricted access |
| Access | Downloadable |
| Model identifiers | llm-jp-3-172b-beta1 |
| Context window | 4,096 tokens Maximum sequence length stated in the model card. LLM-jp-3 172B beta1 official model card |
| Architecture | 96 transformer layers; hidden size 12,288; 96 attention heads.LLM-jp-3 172B beta1 official model card |
| Access | Hugging Face access gate requires acceptance of conditions and sharing profile information.LLM-jp-3 172B beta1 official model card |
| License | Custom LLM-jp-3 172B beta1 terms of use.LLM-jp-3 172B beta1 official model card |
| Approval | The official release index requires approval and restricts redistribution and certain uses.LLM-jp official release index |
Related models
- LLM-jp-3 172B Base
LLM-jp / NII LLMC · LLM-jp-3 172B
The final 2.1-trillion-token 172B base checkpoint uses application-based access and custom terms.
Launch date not confirmedRestricted accessDownloadable - LLM-jp-3 172B beta1 Instruct
LLM-jp / NII LLMC · LLM-jp-3 172B
LLM-jp-3 172B beta1 Instruct is a gated 172B research checkpoint with 700 billion pretraining tokens.
Launch date not confirmedRestricted accessDownloadable