Llama 3.1 8B TFree-HAT SFT
The 8B TFree-HAT SFT checkpoint adds supervised instruction training to Aleph Alpha's tokenizer-free adaptation of Llama 3.1.
Research release
Model design
The architecture combines character-level encoding and decoding with the pretrained word-level backbone. It targets English and German instructions. The research card does not position it as a specialized mathematics or code model.Llama 3.1 8B TFree-HAT SFT model card
Practical use
Practical guidance. Use SFT to isolate the effect of instruction training from later preference optimization. Keep the prompt template fixed when comparing variants. Review actual response usefulness rather than relying only on text-compression measurements.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | Aleph-Alpha/llama-3_1-8b-tfree-hat-sft |
| License | Open Aleph License for non-commercial research and education.Llama 3.1 8B TFree-HAT SFT model card |
Related models
- Llama 3.1 8B TFree-HAT BASE
Aleph Alpha · TFree-HAT
Tokenizer-free Llama 3.1 8B research adaptation with continued pretraining in English and German.
Launch date not confirmedResearch releaseDownloadable - Llama 3.1 8B TFree-HAT DPO
Aleph Alpha · TFree-HAT
English-and-German TFree-HAT instruction model further trained with direct preference optimization.
Launch date not confirmedResearch releaseDownloadable