| --- |
| license: apache-2.0 |
| task_categories: [feature-extraction] |
| tags: [interpretability, mechanistic-interpretability, induction-heads, training-dynamics, pythia, controls] |
| pretty_name: Induction-head emergence across Pythia training |
| size_categories: [1K<n<10K] |
| --- |
| |
| # Induction-head emergence across Pythia training |
|
|
| Per-head **induction**, **previous-token**, and **in-context-learning** scores across the full |
| training checkpoint sequence of Pythia models — plus the **causal ablation control** that turns |
| a correlational score into a mechanism. |
|
|
| **This is a data layer, not a new finding.** Induction-head emergence on Pythia has been studied |
| before (see *Prior work*); what has never been published is the computed scores themselves, as a |
| tidy downloadable table. Olsson et al. (2022), the origin of the phase-change result, used 34 |
| **internally-trained** models — no weights, no checkpoints, no per-head data were ever released. |
|
|
| ## Contents |
|
|
| | file | rows | what | |
| |---|---|---| |
| | `induction_pythia-160m.jsonl` | **154 checkpoints** | all of `step0…step143000`, incl. the log-spaced `step1–512` region | |
| | `induction_pythia-160m-seed{1..9}.jsonl` | 11 ckpts × 9 | **PolyPythias seed axis** | |
| | `induction_pythia-{410m,1b,1.4b}.jsonl` | 11 ckpts each | scale axis | |
| | `induction_pythia-{1b,1.4b}_fp32.jsonl` | 11 each | fp32 re-runs (dtype control) | |
| | `induction_pythia-160m_bf16.jsonl` | 11 | bf16/fp32 comparison | |
| | `ablation_pythia-160m.jsonl` | 5 ckpts | causal control | |
|
|
| Each row: `revision`, `step`, `dtype`, `batch`, `seqlen`, `icl_score_mean`, `icl_score_seeds`, |
| and a `heads` list of `{layer, head, induction_mean, induction_std, prev_token_mean}` over |
| 3 stimulus seeds. |
|
|
| ## What the data shows |
|
|
| **A sharp phase change**, flat until step 512 then complete by step 1000 (induction 0.035 → |
| 0.963; ICL −0.02 → −9.97). Bounded by Pythia's checkpoint spacing — no checkpoints exist between |
| 512 and 1000, so that interval is the finest timing statement this model permits. |
|
|
| **Invariant timing, arbitrary implementation.** All **10 seeds** and all **4 sizes** cross in |
| the same interval. Since Pythia uses identical data order and batch size across sizes, same step |
| = same tokens, so the transition is **data-determined**. Yet the top induction head is a |
| *different* (layer, head) in every one of the 10 seeds, spanning layers 4–8. |
|
|
| **A visible precursor**: at step 512 the previous-token score is 10–17× the induction score in |
| every seed — the prerequisite circuit forms first. |
|
|
| **The ablation control** tracks the phase change: ablating the top-5 induction heads costs |
| **−0.003** at step 512 but **+9.48** at step 1000 (+10.1 at step 143000); five random heads |
| ≈ 0 throughout. Before the transition the top-*scoring* heads have no causal role at all. |
|
|
| ## Caveats |
|
|
| - **dtype matters at the final checkpoint.** bf16 and fp32 agree to ±0.001 through the phase |
| change, but at step 143000 on 160m, 22/144 heads shift by >0.05 and the argmax head changes. |
| 1b and 1.4b show no such effect. Use the fp32 files for mature-checkpoint claims. |
| - A late, gradual ICL improvement (−12.3 → −18.8 from ~step 105000 on 160m) coincides with the |
| tail of Pythia's cosine LR decay — **confounded**, and it is also dtype-sensitive. |
| - Induction score is correlational; the ablation file is what makes it causal. |
| - Yin & Steinhardt argue *function-vector* heads, not induction heads, drive few-shot ICL. |
|
|
| ## Prior work (please cite these too) |
|
|
| Olsson et al., *In-context Learning and Induction Heads* (2022) · Tigges, Hanna, Yu & Biderman, |
| *LLM Circuit Analyses Are Consistent Across Training and Scale* (NeurIPS 2024) · Yin & |
| Steinhardt, *Which Attention Heads Matter for In-Context Learning?* · Feucht et al., *Dual-Route |
| Model of Induction* (nearest prior artifact: 7 of 154 checkpoints, copying scores) · Aoyama, |
| Wilcox & Schneider, *Predicting the Emergence of Induction Heads* (ICML 2026). |
|
|
| ## Reproduce |
|
|
| `scripts/sweep_induction.py` and `scripts/ablate_induction.py`. Probe is seconds per checkpoint; |
| the whole sweep is bandwidth-bound and cost under $20. |
|
|
| Part of *Controls & Trajectories* — publishing the null distributions and developmental |
| trajectories that interpretability papers rely on but rarely ship. |
| [Curriculum](https://github.com/m9h/spinning-up-in-mech-interp) · Morgan Hough, Orthogonal |
| Research and Education Lab (OREL). |
|
|
