11,567 Turkish conversation examples translated from the Hermes 2 Pro training dataset. This dataset, created by Tuguberk, is a machine-translated version of the NousResearch/hermes-function-calling-v1 dataset, designed to teach language models structured output and tool-use capabilities. Over half (52.9%) of the conversations are multi-turn, with an average of 5.6 turns per dialogue.
Use Cases
- Fine-tuning Turkish LLMs for tool and API usage based on the structured function-calling examples.
- Training conversational agents to handle multi-turn dialogues in Turkish, leveraging the 6,120 multi-turn conversations.
- Benchmarking machine translation quality for technical NLP datasets, as it was translated using the DeepSeek V4 Flash model.
- Developing instruction-following models for the Turkish language based on the translated Hermes-style prompts.
Strengths
- Contains 11,567 translated conversation examples.
- 52.9% (6,120) of the conversations are multi-turn, providing context for dialogue modeling.
- Has a clear lineage, being a translation of the established NousResearch/hermes-function-calling-v1 dataset.
- Released under the permissive Apache 2.0 license.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is known, but the specific data structure and file formats are unknown.
- As a machine-translated dataset, it may contain translation artifacts or errors not present in the original English data.
Provenance
- Source
- NousResearch/hermes-function-calling-v1
- Collection Method
- Machine translation using DeepSeek V4 Flash (deepseek-chat).
- Freshness
- Last updated 2026-05-24 11:40:44; freshness should be verified.