Nopm/Opus_WritingStruct
Opus Writing Instruct 6k Synthetically generated creative writing data using Claude 3 Opus, by Anthropic, filtered and cleaned using automated means. Focus was placed on having as many genres as possible represented in the data, and to have Claude more openly use its excellent prose. It also contains question-answer instruction pairs related to the topic of writing. Dataset Details Curated by: Nopm License: Apache 2 Credits: The entire SillyTilly community for… See the full description on the dataset page: https://huggingface.co/datasets/Nopm/Opus_WritingStruct.
| Author / organization | Nopm |
|---|---|
| License (as declared) | apache-2.0 |
| Downloads | 731 (as reported by Hugging Face) |
| Likes | 39 |
| Created | 24 July 2024 |
| Last updated at source | 24 July 2024 |
Popularity, last 2 days
731 now (+21 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
- mradermacher/EVA-Qwen2.5-32B-v0.1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-72B-v0.2-i1-GGUFModel · used to train
- mradermacher/EVA-LLaMA-3.33-70B-v0.1-i1-GGUFModel · used to train
- mradermacher/deepRP-nemo-12b-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-72B-v0.1-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-72B-v0.0-i1-GGUFModel · used to train
- mradermacher/L3.1-70B-Celeste-V0.1-BF16-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.1-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-1.5B-v0.0-GGUFModel · used to train
- mradermacher/natsumura-storytelling-rp-1.0-llama-3.1-8b-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-72B-v0.2-padded-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.2-i1-GGUFModel · used to train
- mradermacher/deepRP-nemo-12b-i1-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-32B-v0.1-GGUFModel · used to train
- Felladrin/Minueza-3-95M-RPModel · used to train
- bartowski/EVA-Qwen2.5-72B-v0.2-GGUFModel · used to train
- mradermacher/ChatWaifu_32B_reasoning-i1-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-14B-v0.2-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-7B-v0.0-i1-GGUFModel · used to train
- mradermacher/EVA-LLaMA-3.33-70B-v0.0-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-72B-v0.1-Padded-i1-GGUFModel · used to train
- tensorblock/EVA-Qwen2.5-32B-v0.1-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-7B-v0.0-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-14B-v0.0-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-7B-v0.1-GGUFModel · used to train
- EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2Model · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.0-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-32B-v0.1-i1-GGUFModel · used to train
- mradermacher/natsumura-storytelling-rp-1.0-llama-3.1-8b-i1-GGUFModel · used to train
- backyardai/EVA-Qwen2.5-32B-v0.2-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-7B-v0.1-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.1-GGUFModel · used to train
- bartowski/EVA-Qwen2.5-32B-v0.2-GGUFModel · used to train
- mradermacher/ChatWaifu_72B_v2.2-i1-GGUFModel · used to train
- tohur/natsumura-storytelling-rp-1.0-llama-3.1-8b-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.2-GGUFModel · used to train
- EVA-UNIT-01/EVA-Qwen2.5-14B-v0.1Model · used to train
- bartowski/EVA-Qwen2.5-32B-v0.0-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-14B-v0.0-i1-GGUFModel · used to train
- mradermacher/EVA-Qwen2.5-32B-v0.2-GGUFModel · used to train
Recent changes
- 16 Oct 2026 Description: Opus Writing Instruct 6k Synthetically generated creative writing data using Claude 3 Opus, by Anthropic, filtered and cleaned using automated means. Focus was placed on having as many genres as possible represented in the data, and to have → Opus Writing Instruct 6k Synthetically generated creative writing data using Claude 3 Opus, by Anthropic, filtered and cleaned using automated means. Focus was placed on having as many genres as possible represented in the data, and to have
Changes detected by Silytics between refreshes of the source.
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.