Correct what-trains section; add caveats on training length and alpha/r

README.md

@@ -82,11 +82,14 @@ English orthography — Hebrew `י` phonemized as `j` was read as the English le
 to the mean of the symbol's original BPE pieces. The output vocabulary is
 untouched — these tokens are never predicted, only read.
 
-**What trains.** LoRA r=32, α=16 on `attention` + `mlp` of the slow transformer,
-plus the IPA embedding table. The acoustic/fast stack is frozen, so voice cloning
-and the codec side are exactly the base model's. The residual-codebook loss is
-down-weighted to 0.3 (Qwen3-TTS's sub-talker coefficient) so the gradient stays on
-the text→semantic mapping.
+**What trains.** LoRA r=32, α=16 on `attention` + `mlp`, plus the IPA embedding
+table — 66.9M parameters total: 60.2M in the slow transformer, 6.7M in the fast
+transformer, 0.03M IPA embeddings. Frozen are the direct interfaces to codebook
+space — `fast_embeddings`, `fast_output`, and the tied slow embeddings/output —
+which is what keeps timbre close to the base model. Note α/r = 0.5, not the usual
+2.0; see *Caveats*. The residual-codebook loss is down-weighted to 0.3
+(Qwen3-TTS's sub-talker coefficient) so the gradient stays on the text→semantic
+mapping.
 
 **Training.** 279,476 Hebrew utterances (~10 speakers, WER ≤ 0.1), reference-
 conditioned on a same-speaker utterance 80% of the time so training prompts match
@@ -126,6 +129,22 @@ applies to any S2-Pro fine-tune, Hebrew or not.
 - `04_yod_BASE.wav` / `05_yod_LORA.wav` — the `י` → English *jay* failure, before and after atomic IPA tokens
 - `07_clone_LORA_ranlevi.wav` — voice clone from a real Hebrew speaker reference
 
+## Caveats — this checkpoint is early, not final
+
+- **Undertrained, and stopped by hand.** 2,200 optimizer steps ≈ 53k utterances
+  seen, under 20% of one epoch over the 279k-row set. Train loss was still falling
+  (3.66 → 2.70 base CE) and val loss was still improving monotonically at every
+  checkpoint (2.934 → 2.844 → 2.820 → 2.807). Nothing had plateaued; the run was
+  simply halted.
+- **α/r = 0.5 is a workaround for a bug that no longer exists.** The unusual
+  scaling was chosen empirically because at α=64 the delta destroyed free-running
+  generation — which we now know was the embedding-scale bug above, not the LoRA
+  strength. That rationale is obsolete post-fix, and the standard α = 2r was never
+  re-tried. It may well be better.
+- The pitch homogenization noted below is the symptom you would expect from
+  putting LoRA on the fast transformer at all. Freezing `fast_layers` entirely is
+  the obvious next experiment.
+
 ## Known limitations
 
 - **Emotion tags (`[whisper]`, `[excited]`, …) do not work** — and this is not a