ERNIE 4.5 300B A47B
The large ERNIE 4.5 text model uses a sparse expert network for general language tasks. This entry covers the post-trained checkpoint.
Model design and access
The card identifies this release as the post-trained text model and provides chat examples. Its overview table contains a conflicting pretraining label; the base checkpoint has a separate repository. The checkpoint has a 131,072-token window.ERNIE-4.5-300B-A47B-PT official model card
Practical use
Practical guidance. Evaluate the post-trained checkpoint on evidence-heavy tasks before committing a large inference cluster. Compare useful completed answers per unit of compute. Treat the full expert weights and long-context cache as separate memory requirements.ERNIE-4.5-300B-A47B-PT official model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | ERNIE-4.5-300B-A47B-PT |
| Context window | 131,072 tokens Official checkpoint context window. Input and generation share this limit. ERNIE-4.5-300B-A47B-PT official model card |
| Parameters | 300B total parameters; 47B active per tokenERNIE-4.5-300B-A47B-PT official model card |
| Architecture | 54 layers; 64 query heads and eight key/value headsERNIE-4.5-300B-A47B-PT official model card |
| License | Apache 2.0ERNIE-4.5-300B-A47B-PT official model card |
Related models
- ERNIE 4.5 300B A47B Base
Baidu ยท ERNIE 4.5
ERNIE 4.5 base language model with 300 billion total parameters and 47 billion active parameters.
Launch date not confirmedDownloadable