Pythia 1B Deduped
Pythia 1B Deduped is the deduplicated-Pile counterpart of Pythia 1B. It preserves the architecture while changing the training data.
Research release
A sixteen-layer deduplicated counterpart
This release has the same nominal size and architecture as standard Pythia 1B. It belongs to the deduplicated-data suite. The two variants should not be treated as the same checkpoint when reporting experiments or loading a saved revision.pythia-1b-deduped model cardpythia-1b model configuration
Weights and training revisions
The release includes 154 saved training steps. The data order is shared with other sizes in the same suite. The main branch provides the final model; a step revision selects an earlier checkpoint. This base model has not been trained as a conversational assistant.pythia-1b-deduped model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Launch date | |
| Access | Downloadable |
| Model identifiers | EleutherAI/pythia-1b-deduped |
| Context window | 2,048 tokens Maximum position count in the released model configuration. pythia-1b model configuration |
| Nominal parameters | 1B including embeddingspythia-1b-deduped model card |
| Network | 16 layers; hidden width 2,048; 8 attention headspythia-1b model configuration |
| Training data | Deduplicated Pilepythia-1b-deduped model card |
| License | Apache 2.0pythia-1b-deduped model card |
Related models
- Pythia 1B
EleutherAI ยท Pythia
A wide, sixteen-layer Pythia model trained with bfloat16 precision.
Launch date Research releaseDownloadable