Rebuild Numbers on the base model with descriptive prompts; add the prompt-distribution result

README.md

@@ -89,38 +89,37 @@ audio = ae.decode(ae.encode(waveform))
 
 ## Numbers
 
-Four generations of 2 minutes, made with a DiT LoRA on `medium` and screened
-from 12 candidates by tonality runaway. Each latent is decoded twice, adapter off and on, and each chain
-re-encodes in place, so dN is N round trips of that decoder config.
+Four base-model generations of 2 minutes on ordinary descriptive prompts,
+screened from 12 candidates by tonality runaway. Each latent is decoded twice,
+adapter off and on, and each chain re-encodes in place, so dN is N round trips of
+that decoder config. The adapter is the only variable.
 
-| depth | tonality p95 | tonal frames | 12-16 kHz | SI-SDR vs own d0 |
+| depth | tonality p95 | tonal frames | 16-22 kHz | SI-SDR vs own d0 |
 |---|---|---|---|---|
-| d0, fresh generation | 11.25 → 11.33 | 780 → 933 | 20.7 → 21.4 | |
-| d1, one continuation | 12.46 → **11.59** | 972 → 1074 | 19.9 → **21.2** | 13.5 → **17.5** |
-| d2 | 14.42 → **11.81** | 1492 → **1246** | 19.3 → **21.0** | 8.9 → **12.8** |
-| d3 | 17.03 → **12.06** | 2453 → **1411** | 18.6 → **20.9** | 6.7 → **10.4** |
-
-Stock tonality climbs **+5.79 dB** over three round trips. With the adapter it
-climbs **+0.73 dB**, and that holds on all four generations individually (stock
-+4.98 to +6.97, adapter +0.18 to +1.40). The artifact stops compounding.
-
-The frame count rises at d0 and d1 because the adapter puts high-frequency energy
-back and the detector counts peakiness, so restored harmonics read as tonal. It
-is also dominated by two of the four generations, which start near 1000 and 2100
-frames because glitchcore genuinely is tonal. The other two start at zero and
-stay near zero with the adapter (0 → 39 and 0 → 8 at d3) while stock reaches 1038
-and 268. Read p95.
-
-The adapter holds more 12-16 kHz than stock at every depth and the gap widens
-with re-encoding (+0.70, +1.31, +1.76, +2.30 dB at d0 to d3), while SI-SDR
-against its own d0 runs 3.6 to 3.9 dB better. It drifts less rather than erasing
-what drifts.
-
-Run again on the base model with no DiT LoRA, the same ladder gives a stock climb
-of +6.73 dB against the adapter's +1.81 dB, so this is the autoencoder's
-behaviour rather than something a LoRA introduces. An earlier unscreened run
-covering a second, quite different LoRA, at 30 s and 120 s, showed the same
-direction at depth. All of these are fp16 at 120 s.
+| d0, fresh generation | 9.80 → **9.10** | 87 → **62** | 3.16 → **4.34** | |
+| d1, one continuation | 11.82 → **9.73** | 462 → **64** | 0.95 → **3.76** | 16.9 → **21.0** |
+| d2 | 14.79 → **10.61** | 2111 → **133** | −1.36 → **3.32** | 12.4 → **16.2** |
+| d3 | 17.85 → **11.74** | 4279 → **453** | −3.36 → **2.96** | 10.1 → **13.6** |
+
+The adapter is ahead at every depth, including a fresh generation. Stock tonality
+climbs +8.05 dB over three round trips against the adapter's +2.64, and by d3
+stock carries 4279 tonal frames against 453.
+
+It is not doing this by sanding the top end. After three round trips the
+adapter's 16-22 kHz sits 0.2 dB from a **fresh** stock render, at unchanged RMS,
+while stock has lost 6.5 dB of it. Stock's 12-16 kHz meanwhile *rises* with
+depth: the invented energy is replacing the air.
+
+Prompt distribution matters more here than anything else. The same ladder run on
+terse genre-tag prompts, which are out of distribution for the base model, starts
+at **1649** tonal frames instead of 87. With a DiT LoRA attached, which makes
+those tags in-distribution again, it starts at 780. The artifact is worst where
+the prompt sits furthest from what the model expects, and re-encoding compounds
+whatever it starts with.
+
+A DiT LoRA does not change the shape: on that run stock climbs +5.79 dB against
+the adapter's +0.73, and the adapter ends d3 holding more 12-16 kHz and 16-22 kHz
+than a fresh stock render. All runs are fp16 at 120 s.
 
 ## Limits
 
@@ -140,10 +139,10 @@ direction at depth. All of these are fp16 at 120 s.
   wrong.
 - **The gain scales with re-encode depth**, so it is small for one-shot
   generation and compounds across chained continuations.
-- **The squeaks are worst on out-of-distribution prompts**, which is easy to hit
-  when a DiT LoRA has pulled the model toward a narrow style. We have not
-  measured with-LoRA against without-LoRA on matched content, and on its own the
-  base model runs away slightly more rather than less.
+- **The squeaks are worst on out-of-distribution prompts.** Genre-tag prompts on
+  the base model start with 19x more tonal frames than descriptive ones. A DiT
+  LoRA makes such tags in-distribution again and lands in between. This adapter
+  reduces the artifact; it does not make prompt choice irrelevant.
 - **One checkpoint, one dataset**, step 2000, chosen on re-encode behaviour and
   listening. Your material may want a different one.
 - **Neither half of the autoencoder is bit-deterministic** (encoder 1.5 %