datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ssd-qwen3-4b-instruct-data
SSD Qwen3-4B-Instruct-2507 Self-Distillation Data
Self-distillation training data generated as part of an attempted reproduction of:
Embarrassingly Simple Self-Distillation Improves Code Generation
Ruixiang Zhang, Richard He Bai, Huangjie Zheng, Navdeep Jaitly, Ronan Collobert, Yizhe Zhang
Apple, 2026 — arXiv:2604.01193
Overview
These are raw, unverified solutions sampled from the frozen Qwen3-4B-Instruct-2507 base model on competitive programming problems. The whole… See the full description on the dataset page: https://huggingface.co/datasets/Ayushnangia/ssd-qwen3-4b-instruct-data.RLVE-Eval20-Qwen3-1.7B-SSD-N20-SFT-Train
RLVE-Eval20-Qwen3-1.7B-SSD-N20-SFT-Train
Parquet for supervised fine-tuning (SFT) in a Simple Self-Distillation (SSD) style setup on the RLVE eval20 filtered pool.
What this is
Questions: 800 prompts from rlve_eval20_filtered (20 harder RLVE environments × 40 seeds).
Continuations: For each question, N = 20 completions were sampled from a frozen Qwen3-1.7B student (including the initial greedy / first sample and additional stochastic samples), then packed for SFT.
Rows:… See the full description on the dataset page: https://huggingface.co/datasets/CL-From-Nothing/RLVE-Eval20-Qwen3-1.7B-SSD-N20-SFT-Train.RLVE-Eval20-Qwen3-4B-SSD-N20-SFT-Train
RLVE-Eval20 Qwen3-4B SSD N=20 SFT Train
Self-generated SFT corpus for Simple Self-Distillation (SSD) with Qwen/Qwen3-4B.
800 RLVE Eval20 (filtered) prompts × 20 self-samples = 16,000 rows
Sampled from frozen Qwen3-4B (vLLM, max_tokens=16384, thinking enabled).
Stored as VERL MultiTurnSFTDataset parquet with a messages column.
Companion 1.7B dataset: CL-From-Nothing/RLVE-Eval20-Qwen3-1.7B-SSD-N20-SFT-Train.
ssd-math-v1p1-e06m-fixed-batch-16
SSD Math V1.1-E06M fixed batch
Tiny 16-record SFT JSONL fixture used for the V1.1-E06M 4B single-batch LR diagnostic.
The records are generated math-reasoning traces from the local v0 overfit fixture artifacts/overfit/v0_unique_batch_16.jsonl.