This student shows no measurable capability gain over its base model.

A later trajectory study (exp_student_trajectory_2026-08-20) compared this checkpoint with its own base model on the full GSM8K test set: 70.8 → 71.7 (+0.8 points), McNemar paired test over 1,319 questions p=0.47 — not significant at α=0.05. 233 of 1,319 answers changed (111 right → wrong, 122 wrong → right) for a net of +11.

Fine-tuning did change the model's behaviour; it did not change how often it is right. That may still be useful for studying style transfer independent of capability, but do not treat this as an improved model.

DistillDetect-Llama-3.2-3B-Instruct-from-Qwen3-8B-s1

Unofficial reproduction of a distilled student model from the paper Reference-Based Distillation Detection in LLMs (Rawat et al., arXiv:2607.09692), retrained with the authors' released code and teacher-generated data (github.com/RajatRawat-creator/DistillDetect, MIT). The original authors did not release student checkpoints; this repo is an independent reproduction and is not affiliated with the authors.

Base (student) model: meta-llama/Llama-3.2-3B-Instruct
Teacher: Qwen/Qwen3-8B
Training data: s1 (1K prompts) — 1000 teacher-generated responses, shipped verbatim in the authors' repo (data/training/Teacher=Qwen-3-8B_Data=S1_Template=Chat.jsonl)
Prompt template: model's native chat template (apply_chat_template), supervision on assistant turn only

Training

SFT with the authors' released training/ scripts (paper Appendix A recipe): 3 epochs, LR 1e-5, cosine schedule, 5% warmup, per-device batch 4 x grad-accum 4 (effective batch 16), block size 4096, bf16, gradient checkpointing, loss on response tokens only (prompt masked -100). Teacher responses were pre-truncated to 2,048 tokens in the released data. Trained on 1x H100 (paper used 2x H200; hyperparameters identical), transformers 4.55.4 / trl 0.19.1, seed 42 (HF default; paper seed unknown). One compatibility patch: trl renamed max_seq_length to max_length, so the block size is passed explicitly (no behavioral change).

Evaluation (ours vs. paper Table 9)

Greedy decoding, template matched to training, scored with math_verify. GSM8K 8-shot; MATH500 zero-shot. Few-shot counts were calibrated so the base models reproduce their Table 9 baselines. The paper does not document its eval protocol, so treat cross-paper comparisons as approximate.

Benchmark	This reproduction	Paper Table 9	Gen. budget	Hit budget cap
GSM8K	71.72	n/a (config not retained in paper)	4096 tok	1.6%
MATH500	22.80	n/a (config not retained in paper)	16384 tok	82.4%

A high "hit budget cap" fraction means the model was still generating when the token budget ran out, so that accuracy is a lower bound — these students are SFT'd on teacher traces that were themselves truncated at 2,048 tokens, which makes some of them generate very long self-checking traces.

Base-model reference (our protocol / paper): GSM8K 68.69 / 76.57, MATH500 42.00 / 42.00.