license, base_model, pipeline_tag, library_name, language, tags
| license | base_model | pipeline_tag | library_name | language | tags | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| llama3.1 | meta-llama/Llama-3.1-8B-Instruct | text-generation | transformers |
|
|
Keystone GPT v12.3.1
Domain-adapted Llama 3.1 8B Instruct, LoRA fine-tuned then merged. Targets community / regional bank financial analysis: call-report interpretation, RAG over banking statutes and regulations, regulator- facing memos. v12.3.1 adds polish-pass training (off-topic redirect, clean calculation output, regulatory humility, no-fake-data templates).