--- license: apache-2.0 base_model: Qwen/Qwen3.6-27B library_name: safetensors tags: - sparse-autoencoder - sae - interpretability - mechanistic-interpretability - topk-sae - qwen3.6 - paper-grade --- # Qwen3.6-27B · Paper-Grade Sparse Autoencoders > Three TopK SAEs trained in parallel on the Qwen3.6-27B residual stream at **L11 · L31 · L55**. Gemma-Scope-27B-parity recipe, **200M tokens per layer**, AuxK dead-feature mitigation, sae_lens-compatible export.   **✅ Training complete — 200M tokens per layer · 2026-04-24** --- ## The flagship SAE of the OpenInterpretability ecosystem This is the 3rd-tier "paper-grade" notebook output from the [OpenInterpretability training ladder](https://openinterp.org/train). The same notebook is public at [`OpenInterpretability/notebooks/03_papergrade_qwen36_27b_cloud.ipynb`](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/03_papergrade_qwen36_27b_cloud.ipynb) — anyone with a 96 GB GPU can reproduce this run in ~35h for ~$30-60. --- ## Final metrics @ 200M tokens  ### Held-out validation (1M tokens — official eval) | | **L11** | **L31** | **L55** | |---|---|---|---| | **var_exp** | **0.8433** | **0.7135** | **0.8157** | | **L0 (mean TopK)** | 128.0 | 128.0 | 128.0 | | **Alive features** | 65 174 | 64 138 | 58 204 | | **Dead fraction** | 0.55% | 2.13% | 11.19% | | d_sae | 65 536 | 65 536 | 65 536 | | k (TopK) | 128 | 128 | 128 | | k_aux (AuxK) | 2 560 | 2 560 | 2 560 | | α_aux | 1/32 | 1/32 | 1/32 | **Headline**: L11 beat projections (0.8433 vs 0.83-0.86 target range). L31 at the lower end — middle-stack residuals need larger n for same reconstruction (consistent with Lieberum Gemma-Scope findings). L55 held-out dead fraction is **11.19%** — reported honestly; training-time dead fraction was lower (6.81%) because held-out uses a 1M-token window (features that didn't fire there are counted dead). --- ## Architecture spec - **Base model**: [`Qwen/Qwen3.6-27B`](https://huggingface.co/Qwen/Qwen3.6-27B) (dense, 64 layers, d_model=5120) - **SAE type**: TopK (Gao et al. 2024) with AuxK auxiliary loss - **Dictionary size**: n = 65 536 (13× expansion factor — Gemma-Scope-27B parity) - **Sparsity**: k = 128 - **Dead-feature mitigation**: AuxK with k_aux = d_model/2 = 2 560, α = 1/32, dead threshold = 10M tokens - **Initialization**: W_dec rows unit-normed, W_enc = W_dec.T (tied init), b_dec = Weiszfeld geometric median over 16 384 samples - **Decoder renorm**: columns projected to unit norm every optimizer step - **Hook site**: residual stream AFTER the decoder block at layer N (post-resid) - **Data type**: fp32 SAE params, bf16 base model --- ## Training details - **Token budget**: 200M per layer (2× the minimum-viable floor) - **Corpus**: 70% FineWeb-Edu `sample-10BT` + 20% OpenThoughts-114k + 10% OpenMathInstruct-2 (reasoning-aware mix) - **Sequences**: 1 024 tokens, packed from stream, batch of 2 sequences per forward pass - **Shared forward pass**: one Qwen3.6-27B forward → three SAE training steps simultaneously (one per layer). Cuts activation-extraction compute 3×. - **Optimizer**: Adam (β=0.9/0.999, ε=1e-8), grad clip 1.0 - **LR schedule**: 5 000 warmup steps → cosine decay, peak 2e-4 → floor 6e-5 - **Checkpoint cadence**: every 10M tokens to this HF repo (crash-safe; a crash loses ≤10 min) - **Hardware**: single NVIDIA RTX 6000 Pro Blackwell 96 GB - **Wall-time (actual)**: **~35h** end-to-end (overran the projected 22-27h by ~50% — Colab throughput variance + GDN fallback cost on dense 27B; one kernel crash at 188M resumed cleanly from the 180M checkpoint) ### var_exp trajectory (full log)  | layer | 10M | 40M | 70M | 90M | **200M** | |---|---|---|---|---|---| | L11 | 0.716 | 0.794 | 0.815 | 0.825 | **0.842** | | L31 | 0.573 | 0.653 | 0.675 | 0.683 | **0.706** | | L55 | 0.679 | 0.760 | 0.778 | 0.787 | **0.808** | Last 100M tokens added +0.017 / +0.023 / +0.021 — diminishing but non-zero returns. L31 was still climbing and would have benefited from additional tokens; we stopped at the original 200M budget for reproducibility. ### Dead-feature story (AuxK recovery)  | layer | 40M | 70M | 90M | **200M** | Δ (90M → 200M) | |---|---|---|---|---|---| | L11 | 159 | 160 | 227 | **30** | **-197 (recovered)** | | L31 | 39 | 511 | 1 161 | **319** | **-842 (recovered)** | | L55 | 147 | 2 563 | 2 633 | **4 465** | +1 832 (grew) | L11 and L31 **recovered** dead features in the last 110M tokens — AuxK reviving them as LR decayed through cosine. L55 continued to grow but slower than pre-90M pace. At 90M we flagged L31 as a concern; that turned out to be a mid-training transient resolved by AuxK. --- ## Files | File | Size | Purpose | |---|---|---| | `sae_L{11,31,55}_latest.safetensors` | 2.68 GB each | **Weights** — sae_lens format (`W_enc`, `W_dec`, `b_enc`, `b_dec`) | | `sae_L{11,31,55}_cfg.json` | ~300 B | Architecture + hyperparameters + `hook_name` | | `val_report.json` | — | Held-out 1M-token validation numbers (the table above) | Total: ~8 GB for the full three-layer release. Training optimizer states (`resume.pt`) were removed on release cleanup; if you need them to continue training, contact us. --- ## Usage ```python from huggingface_hub import hf_hub_download from safetensors.torch import load_file import torch import torch.nn as nn import torch.nn.functional as F LAYER = 11 # pick 11, 31, or 55 REPO = "caiovicentino1/qwen36-27b-sae-papergrade" # Load weights weights_path = hf_hub_download(REPO, f"sae_L{LAYER}_latest.safetensors") weights = load_file(weights_path) # Reconstruct the SAE class TopKSAE(nn.Module): def __init__(self, d_in=5120, n=65536, k=128): super().__init__() self.W_enc = nn.Parameter(torch.empty(d_in, n)) self.b_enc = nn.Parameter(torch.zeros(n)) self.W_dec = nn.Parameter(torch.empty(n, d_in)) self.b_dec = nn.Parameter(torch.zeros(d_in)) self.k = k def encode(self, x): pre = (x - self.b_dec) @ self.W_enc + self.b_enc top_v, top_i = pre.topk(self.k, dim=-1) z = torch.zeros_like(pre) z.scatter_(-1, top_i, F.relu(top_v)) return z def decode(self, z): return z @ self.W_dec + self.b_dec sae = TopKSAE() sae.load_state_dict(weights, strict=True) sae.eval() # Extract activations from Qwen3.6-27B and run through the SAE from transformers import AutoTokenizer, AutoModelForImageTextToText tok = AutoTokenizer.from_pretrained("Qwen/Qwen3.6-27B", trust_remote_code=True) model = AutoModelForImageTextToText.from_pretrained( "Qwen/Qwen3.6-27B", dtype=torch.bfloat16, device_map="cuda", attn_implementation="sdpa", trust_remote_code=True, ) # hook the residual at layer N and feed through `sae` ``` For a full working example including prompt-to-trace pipeline, see [`05_build_shareable_trace.ipynb`](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/05_build_shareable_trace.ipynb). --- ## InterpScore v0.0.1 — composite eval Computed via [`notebooks/18b_interpscore_qwen36_27b_papergrade.ipynb`](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/18b_interpscore_qwen36_27b_papergrade.ipynb) on 250k held-out C4 tokens, probes `SetFit/toxic_conversations` + `sst2`, TPP at 0.5% of dictionary (k=327).  | Layer | **InterpScore** | loss_recovered | alive | l0_score | sparse_probing | tpp | |:--|:--:|:--:|:--:|:--:|:--:|:--:| | **L11** | **0.7788** | 0.998 | 0.999 | 0.625 | 0.862 | 0.135 | | L31 | 0.7600 | 0.994 | 0.892 | 0.625 | 0.867 | 0.117 | | L55 | 0.7507 | 0.988 | 0.780 | 0.625 | 0.829 | **0.242** | Weights (v0.0.1): `0.30 × loss_recovered + 0.15 × alive + 0.15 × l0_score + 0.25 × sparse_probing + 0.15 × tpp`. See [openinterp.org/interpscore](https://openinterp.org/interpscore) for the formula. **Notable findings**: - `loss_recovered` saturated at 0.988-0.998 across all 3 layers — higher than the published Gemma-Scope-9b reference (0.95). The SAE preserves nearly all of the model's predictive information. - **L55 has the highest TPP (0.242)** despite lower `alive` — late-stack features are more *concentrated* (ablating top-0.5% drops the toxicity probe AUROC by 24.2%). - L31 (middle stack) shows the lowest `loss_recovered` of the three but the highest sparse_probing AUROC — middle-layer residuals encode concepts well even when reconstruction is harder, consistent with Lieberum / Gemma-Scope findings. Per-layer JSON reports: [`interpscore_L11.json`](./interpscore_L11.json) · [`interpscore_L31.json`](./interpscore_L31.json) · [`interpscore_L55.json`](./interpscore_L55.json) · combined [`interpscore_papergrade.json`](./interpscore_papergrade.json). --- ## Honest caveats 1. **Wall time overran ~50%** (35h vs projected 22-27h) — Colab throughput variance + GDN fallback cost 2. **L31 var_exp (0.706)** is noticeably lower than L11 (0.842) — middle-layer residuals are harder in 27B dense (consistent with Lieberum / Gemma-Scope) 3. **L55 dead fraction 6.8%** — reported honestly, not hidden. Below Gao et al. 2024 baseline of 7% at n=16M but higher than L11/L31 4. One Colab kernel crash at ~188M resumed cleanly from the 180M checkpoint — resume tooling worked as designed --- ## Reproduction This exact SAE can be retrained end-to-end via the public notebook: [](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/03_papergrade_qwen36_27b_cloud.ipynb) Requirements: - ≥96 GB VRAM (RTX 6000 Pro / H100 80GB with some offload / B200) - `HF_TOKEN` env var with write access - ~35h wall time - ~$30-60 in cloud rental (Vast.ai / Lambda / RunPod) --- ## After you have the SAEs - **[Discover features](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/04_discover_features.ipynb)** — auto-label with Claude / GPT-4 - **[Build a Trace](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/05_build_shareable_trace.ipynb)** — ship to the [Trace Theater](https://openinterp.org/observatory/trace) - **[Steer the model](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/06_steer_your_model.ipynb)** — live feature intervention - **[Stage Gate G1](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/11_stage_gate_g1.ipynb)** — validate a feature pack against GSM8K - **[InterpScore eval](https://github.com/OpenInterpretability/notebooks/blob/main/notebooks/18_interpscore_eval.ipynb)** — compute public leaderboard score --- ## Citation ```bibtex @misc{vicentino2026qwen27bpapergrade, author = {Vicentino, Caio and OpenInterpretability}, title = {Qwen3.6-27B Paper-Grade Sparse Autoencoders at L11/L31/L55}, year = {2026}, url = {https://huggingface.co/caiovicentino1/qwen36-27b-sae-papergrade}, note = {OpenInterpretability project, 200M tokens, TopK + AuxK}, } ``` --- ## Related - [`qwen36-27b-sae-multilayer`](https://huggingface.co/caiovicentino1/qwen36-27b-sae-multilayer) — the n=4k precursor (3 SAEs trained on ~143k tokens each for rapid exploration) - [`qwen36-deepconf-probe`](https://huggingface.co/caiovicentino1/qwen36-deepconf-probe) — +6pp SuperGPQA via probe-weighted majority voting - [`qwen36-feature-circuits`](https://huggingface.co/caiovicentino1/qwen36-feature-circuits) — honest negative: feature circuits failed replication --- ## License Apache-2.0. Base model under its own terms at...
