SMITH-Exp is an instruction-tuned Mixture-of-Experts checkpoint based on GigaChat 3.1 Lightning (10B total parameters, 1.8B active, BF16). It is optimized for multi-hop retrieval through GigaChat 3 tool calls: the model iteratively invokes searchindex, then returns an answer with supportingcorpusids`. For reproducible results, use the checkpoint with the SMITH retrieval harness or an equivalent implementation of the same tool-calling loop. The evaluation results below assume this execution protocol; behavior and quality are not guaranteed in general-purpose chat frameworks or integrations that use different prompts, tools, parsing, or stopping logic. The included gigachat3guideddecoding.py plugin integrates with vLLM’s GigaChat 3 tool parser and constrains tool calls to the expected structure. This reduces malformed calls and allows the harness to parse searchindex and final answer` actions reliably. Use the plugin with the supplied chat template, as shown below. MuSiQue answerable validation (bdsaglam/musique), 100 queries mixing the three four-hop topologies: 50 4hop1, 25 4hop2, and 25 4hop3. Ranking is against gold supporting paragraphs at k=10 on a text-deduplicated…
Модальности:
Генерация текста
Области применения:
Следование инструкциям Диалог / чат Вызов функций (Tool use) RAG (поиск + генерация) Мультиязычность
Задача: Генерация текста
Автор: ai-forever
Теги: deepseek_v3, instruct, moe, multilingual, bf16, long-context, tool-use, retrieval
Лайков: 3 | Загрузок: 168
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.