Upload README.md with huggingface_hub

README.md

@@ -59,11 +59,26 @@ on positions 30-128 is on near-zero-weight tail and doesn't affect output.
 **ANN at K ≥ 256 produces lower perplexity than full attention** — the
 sparse-attention denoising effect. Full softmax is forced to spread small
 amounts of weight over a long tail of irrelevant keys; truncating to top-K
-and renormalizing puts the weight where it matters. The smooth monotonic
-trend (no discontinuous jumps) is consistent with this explanation, and the
-sanity checks (same input sequences for `ppl_full` vs `ppl_ann`, intact
-causal mask in retrieval, single-softmax renormalization with no wrapper
-leakage between iterations) confirm the result is real.
+and renormalizing puts the weight where it matters.
+
+The novelty is not "ANN beats attention" (denoising is a property of any
+hard top-K selection over softmax, see Top-k Attention / Reformer). What's
+distinctive is that this method produces the denoised top-K at *sub-linear
+cost* via off-the-shelf FAISS HNSW, vs. computing all `O(L²)` scores first.
+
+Sanity checks pass: same input sequences for `ppl_full` vs `ppl_ann`,
+intact causal mask in retrieval, single-softmax renormalization with no
+wrapper leakage between iterations.
+
+### Deployment knobs (L = 4096)
+
+| Use case | K | PPL gap | Attention compute reduction |
+|---|---|---|---|
+| Quality-improving | 256 | −0.79% | ~16× |
+| Quality-improving | 512 | −2.89% | ~8× |
+| Quality-preserving | 128 | +0.82% | ~32× |
+| Aggressive | 64 | +2.42% | ~64× |
+| Speed-only | 32 | +4.51% | ~128× |
 
 Note: the K-sweep recall numbers (24–41%) are not directly comparable to the
 in-training `evaluate()` recall (50.9% at K=128). Same checkpoint, same K,