Download README.md from Stormtrooperaim/Strata-distilled: direct link, hf CLI and curl.
- Browser
- Download file 3.8 kB
-
https://huggingface.co/datasets/Stormtrooperaim/Strata-distilled/resolve/main/README.md
- Command line
-
hf download hf://datasets/Stormtrooperaim/Strata-distilled/README.md
-
curl -L -o README.md https://huggingface.co/datasets/Stormtrooperaim/Strata-distilled/resolve/main/README.md
license: mit
task_categories:
- text-generation
language:
- en
pretty_name: Strata-distilled
size_categories:
- n<1K
tags:
- synthetic
- distillation
- reasoning
- chain-of-thought
Strata-distilled
A small test dataset (~50 conversations) of synthetic chat data with reasoning traces at several thinking levels. It can be used to fine-tune smaller thinking models.
Test release. This is a proof-of-concept run of a larger generation pipeline, not a finished dataset.
Disclaimer
- Generated by a heavily quantized model. All data was produced by
swift-1.5-iq3_xxs, an IQ3_XXS (roughly 3-bit) quantization. Heavy quantization lowers answer quality and reasoning reliability, so expect factual errors, flawed logic, weak code and inconsistent formatting. - Not checked before publishing. The samples are raw model output. They were not reviewed by a human or verified for correctness, safety or bias. Do not treat the answers or the reasoning as ground truth.
- Use it to test a training pipeline. Filter and validate it before relying on it for anything serious.
How it was made
One teacher model (swift-1.5-iq3_xxs, served locally through an OpenAI-compatible API) played both sides of every conversation:
- The user. The model was prompted to write casual, human-sounding messages. Each conversation got a random persona, writing style and message length, and a concrete topic idea brainstormed by the model, so messages read like real chat input (lowercase, slang, typos, terse or rambling).
- The assistant. The model answered each turn, with thinking turned on or off depending on the sample's thinking level.
Conversations have 1 to 4 turns, and some start with a system prompt.
Topics: coding, MMLU-style multiple-choice questions across the 57 MMLU subjects, world knowledge, math and reasoning, writing and language, practical advice, instruction following and casual chat.
Thinking levels
Each sample has a thinking_level field:
| Level | Meaning |
|---|---|
off |
No reasoning. The assistant turn begins with an empty <think> block. |
low |
Very short reasoning. |
medium |
Focused, step-by-step reasoning. |
high |
Long, thorough reasoning with double-checking. |
Levels were requested through the prompt, and samples whose reasoning length didn't fit the requested level were discarded. They are approximate labels, not guarantees.
Format
The file is a JSON array. Each element looks like this:
{
"id": "a1b2c3d4e5f6",
"category": "coding",
"subcategory": "Python",
"thinking_level": "medium",
"num_turns": 2,
"messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "..."},
{"role": "assistant", "content": "<think>\n...reasoning...\n</think>\n\nfinal answer"}
]
}
systemmessages are optional.- Assistant turns follow the Qwen3 chat format. The reasoning is inside
<think>...</think>, followed by the answer. Withoff, the block is empty (<think>\n\n</think>\n\n). messagesis the standard chat format used by most fine-tuning libraries (TRL, Unsloth, Axolotl and others).
Usage
from datasets import load_dataset
ds = load_dataset("json", data_files="swift_distill_dataset.json", split="train")
print(ds[0]["messages"])
Limitations
- Tiny sample size, so it is not suitable for real fine-tuning on its own.
- Single teacher model, so its mistakes and style biases are copied into the data.
- Reasoning traces come from a heavily quantized model and may be wrong even when the final answer is right (or the reverse).
- English only.
- Simulated users are imitations of human messages, not real user data.
License
MIT, as declared in the metadata above.