Upload README.md with huggingface_hub

README.md

@@ -227,11 +227,12 @@ the uninstrumented one, so what the lab displays is what the model computed.
 
 ## Part of a family
 
-Four repos from one project on small models doing narrow tasks, each with the
+Five repos from one project on small models doing narrow tasks, each with the
 measurements behind it rather than around it:
 
 - [`gpt-alpha-bg-91m`](https://huggingface.co/glassbox/gpt-alpha-bg-91m) - the 91M flagship: a fluency judge that ties a frontier model on Bulgarian error/correction pairs
 - **`gpt-alpha-bg-restorer`** - you are here
+- [`gpt-alpha-bg-corrector`](https://huggingface.co/glassbox/gpt-alpha-bg-corrector) - the flagship fine-tuned to rewrite Bulgarian rather than judge it, and four documented ways of failing to improve it
 - [`gpt-alpha-bg-14m-onnx`](https://huggingface.co/glassbox/gpt-alpha-bg-14m-onnx) - the 13.77M LM as 17 MB of int8 ONNX, with the quantization cost measured
 - [`gpt-alpha-molgen-selfies-0.8m`](https://huggingface.co/glassbox/gpt-alpha-molgen-selfies-0.8m) - a sub-1M molecule generator, and evidence that the search wrapper is the lever