Update README.md

README.md

@@ -45,6 +45,17 @@ This repo contains the following configurations under `./models/`:
     + Seems to be a decent foundation for "distillation", at the very least for LoRA training.
     	- Addendum: it seems to serve fine for patch-training a few extra tweaks, to non-unified position IDs, split classifier heads, and para-parallel decoding for the AR.
 
+* `config.llama-tts+stt.yaml` / `ar+nar-tts+stt-llama-8`: The above, but with partially trained for STT.
+    + These weights use the above weights but with additional training for the default `tts` task and a new `stt` task (at a 3:1 ratio).
+    + Initially was trained with `duration_range: [3.0, 60.0]` and `sample_shuffle: True` for a few hours, but then pivoted to my standard  `duration_range: [3.0, 12.0]` and `sample_shuffle: False`
+      + Will need the former training to "undo" any issues with durations, as it usually came up before.
+    + `stt` task simply takes a piece of audio and outputs a transcription using IPA phonemes (that the model already is trained against for its text inputs).
+      + Can be done with `--task=stt` and an empty (`""`) text input through the CLI interface or the `Speech-to-Text` tab in the web UI.
+    + This mainly serves as a stepping stone before pivoting towards SpeechX tasks.
+      + I first need a good mechanism to make sure I *can* extend existing weights with additional tasks, but with a simple enough task.
+      + This also *maybe* seems to help bolster the initial TTS task by helping the model have a better internal state (or something to that tune).
+    + STT is not perfect against voices that aren't close to a normal speaking voice (as per the dataset), unlike TTS where you can easily have "sounds close enough" and room for errors.
+
 Some additional configurations have been explored with, but experiments have not been fruitful:
 * Exotic wrappers like `BitNet` seemed to yield little gains in inferencing, somehow. The memory savings is pretty much unneccessary as the models are already manageable at ~200M parameters.
 * Mamba / Mamba2-based models have shown that it's ***really*** hard to have an AR+NAR model. I really do not want to bother throwing the compute at another ~~meme~~ arch I can't easily make use of all the other tech to throw at.