Point links at the fork branch, add origin note, add base-model ladder

README.md

@@ -36,7 +36,7 @@ transform re-encodes its input, so a chained session compounds them.
 | **neurofunk**, after 3 round trips | <audio controls src="https://huggingface.co/thepatch/same-l-decoder-lora/resolve/main/samples/neurofunk_stock_d3.wav"></audio> | <audio controls src="https://huggingface.co/thepatch/same-l-decoder-lora/resolve/main/samples/neurofunk_v3_d3.wav"></audio> |
 
 15 s clips, same latent and seed, decoder adapter the only variable, peak-matched
-to −1 dBFS. Generated with the `koan` DiT LoRA on `medium`. These three prompts
+to −1 dBFS, generated on `medium` with a DiT LoRA attached. These three prompts
 were not picked for artifact content, and the fresh-generation rows are there to
 show the progression rather than because much is audible in them.
 
@@ -51,12 +51,21 @@ pair where it carries. What changes here is air rather than squeak count: on thi
 pair tonality moves 0.08 dB, which is nothing, while 16-22 kHz comes back
 +2.4 dB.
 
+## How this came about
+
+I do a lot of transformations and continuations using my guitar as input audio.
+When I first noticed these squeaks in the outputs, I mentioned that my trick was
+to stem separate in Ableton and then apply the erosion effect to the percussion
+layer. Claude used that as a guide for refining the methodology. The technical
+version is below.
+
 ## Usage
 
-Requires `target: "decoder"` support in
-[Stability-AI/stable-audio-3](https://github.com/Stability-AI/stable-audio-3).
-The `target` field in the checkpoint config routes it; there is no special
-loading path. Ship at strength **1.0**, where it was trained and auditioned.
+Requires the `target: "decoder"` support on the
+[`decoder-lora` branch of betweentwomidnights/stable-audio-3](https://github.com/betweentwomidnights/stable-audio-3/tree/decoder-lora).
+It is not upstream yet, so that branch is currently the only place a decoder LoRA
+will load. The `target` field in the checkpoint config routes it; there is no
+special loading path. Ship at strength **1.0**, where it was trained and auditioned.
 
 ```bash
 python run_gradio.py --model medium-base --lora-ckpt-path squeakfix_v3.safetensors
@@ -80,8 +89,8 @@ audio = ae.decode(ae.encode(waveform))
 
 ## Numbers
 
-Four `koan` generations of 2 minutes, screened from 12 candidates by tonality
-runaway. Each latent is decoded twice, adapter off and on, and each chain
+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.
 
 | depth | tonality p95 | tonal frames | 12-16 kHz | SI-SDR vs own d0 |
@@ -107,8 +116,11 @@ 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.
 
-This table is `koan` at 120 s and fp16. An earlier unscreened run covering `kev`
-as well, at 30 s and 120 s, showed the same direction at depth.
+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.
 
 ## Limits
 
@@ -128,8 +140,10 @@ as well, at 30 s and 120 s, showed the same direction at depth.
   wrong.
 - **The gain scales with re-encode depth**, so it is small for one-shot
   generation and compounds across chained continuations.
-- **The squeaks are partly a prompt-distribution problem.** Out-of-distribution
-  prompts produce far more of the offending frames.
+- **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.
 - **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 %
@@ -158,7 +172,7 @@ drift), and latents sampled from the base DiT. No discriminator, since
 adversarial training makes a decoder invent *more* detail.
 
 Full method, losses, ablations and traps: [Decoder LoRA
-workflow](https://github.com/Stability-AI/stable-audio-3/blob/main/docs/workflows/decoder-lora.md).
+workflow](https://github.com/betweentwomidnights/stable-audio-3/blob/decoder-lora/docs/workflows/decoder-lora.md).
 Provenance is embedded in the file's `lora_config` metadata. `v1` (identical
 recipe, `--lambda_patch 0`) is an internal control, not published.