pansophic-1-preview 🦝 is an 8-billion parameter large language model that was trained on a mix of English and Romanian instructions. The prompt format used is ChatML. — Model type: An 8B parameter GPT-like model fine-tuned and aligned with APO-ZERO. — Language(s) (NLP): Romanian and English — License: llama3.1 — Finetuned from model: meta-llama The model is compared on benchmarks created by the OpenLLMRo Community: !Benchmarks Results The model is optimized for agentic behavior when it has access to external tools. In the example below, we inform the model that it has access to a function that can generate images: And this is the generated image using FLUX model. !Flux Image The start of the function call is denoted by the special token , and the token marks the end of the function call. The model is optimized to adhere strictly to system prompts: The model is optimized for context usage. The model generates special tokens to retrieve information from the context before formulating an answer, ensuring that the response is grounded in the appropriate context. The model will use special tokens to denote the information retrieval from the context. «`text context = «»»Document:0…
Модальности:
Генерация текста
Области применения:
Диалог / чат
Задача: Генерация текста
Автор: pansophic
Теги: llama, conversational, ro, en, text-generation-inference, endpoints_compatible
Лайков: 3 | Загрузок: 48
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
