Dataset · Hugging Face

marin-community/openthoughts4-code-9168-prompts-qwen3-4b-n16-flattened-logprobs-k16

Name: openthoughts4-code-9168-prompts-qwen3-4b-n16-flattened-logprobs-k16
Creator: marin-community
License: apache-2.0
Keywords: task_categories:text-generation, language:en, license:apache-2.0, size_categories:100K<n<1M, format:parquet, modality:tabular, modality:text, modality:timeseries, library:datasets, library:dask, library:polars, library:mlcroissant, arxiv:2505.09388, region:us, code

OpenThoughts-4 Code SDG: Qwen3-4B (n=16, top-16 logprobs) Synthetic generations from Qwen/Qwen3-4B on the Marin OpenThoughts-4 code SDG prompt set. Each prompt is sampled n=16 times, and for every generated token the dataset stores the chosen-token log probability plus the top-16 log probabilities over the vocabulary, enabling distillation, KL-style fine-tuning, reranking, and uncertainty analysis. Generation setup Field Value Generator model… See the full description on the dataset page: https://huggingface.co/datasets/marin-community/openthoughts4-code-9168-prompts-qwen3-4b-n16-flattened-log

View on Hugging FaceReport an issue

Author / organization	marin-community
License (as declared)	apache-2.0
Downloads	511 (as reported by Hugging Face)
Likes	0
Created	5 May 2026
Last updated at source	5 May 2026

codesyntheticsdgreasoninglogprobsqwen3

Embed this entry

<iframe src="https://silytics.si/embed/item/aca47051-efec-4b05-a1a0-e7ef8c375c51" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 3 days

511 now (-51 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.

Recent changes

17 Oct 2026 Description: OpenThoughts-4 Code SDG: Qwen3-4B (n=16, top-16 logprobs) Synthetic generations from Qwen/Qwen3-4B on the Marin OpenThoughts-4 code SDG prompt set. Each prompt is sampled n=16 times, and for every generated token the dataset stores the chos → OpenThoughts-4 Code SDG: Qwen3-4B (n=16, top-16 logprobs) Synthetic generations from Qwen/Qwen3-4B on the Marin OpenThoughts-4 code SDG prompt set. Each prompt is sampled n=16 times, and for every generated token the dataset stores the chos

Changes detected by Silytics between refreshes of the source.

More from marin-community

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.