Nous ResearchPuffin
Redmond Puffin 13B
Redmond Puffin 13B uses a curated set of about 3,000 examples, many of them multi-turn human and GPT-4 conversations.
Training and model behavior
The release also uses selected science and mathematics examples. The card recommends Puffin for creative, extended interaction. It provides corrected prompt prefixes and suffixes after an earlier formatting error.Redmond Puffin 13B model card
Working with this model
Practical guidance. Check recall of details across multiple turns. Use the corrected card template, and include both coherent and contradictory conversation histories in your review.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | July 2023 |
| Access | Downloadable |
| Model identifiers | NousResearch/Redmond-Puffin-13B |
| Context window | 4,096 tokens Configured position limit; practical serving limits depend on the runtime. Redmond Puffin 13B configuration |
| Base model | Meta Llama 2 13BRedmond Puffin 13B model card |
| Repository license label | mitRedmond Puffin 13B model card |
Related models
- Nous Puffin 70B
Nous Research ยท Puffin
A Llama 2 fine-tune focused on extended conversations.
Launch date 2023Downloadable