Slim AAIT-86M main repo and move GGUF artifacts to dedicated repo

README.md

@@ -60,10 +60,6 @@ Action logit order:
   - anchor head checkpoint
 - `AAIT-86M.safetensors`
   - combined self-contained release artifact
-- `AAIT-86M_q8_0.gguf`
-  - combined GGUF export, higher-fidelity quantization
-- `AAIT-86M_q5_1.gguf`
-  - combined GGUF export, smaller quantization
 - `config.json`
 - `load_aait86m.py`
 - `example_inference.py`
@@ -110,15 +106,11 @@ The retrieval checkpoint stores the trained projection heads and runtime config
 
 ## GGUF Note
 
-The GGUF files in this repo are quantized exports of the combined `AAIT-86M` package using the custom `triembed` architecture metadata.
+GGUF exports for this model live in the separate repository:
 
-They are useful for:
+- `augmem/AAIT-86M-GGUF`
 
-- compact storage
-- transport
-- custom runtime integration work
-
-They are not generic llama.cpp text-model artifacts.
+Those artifacts are quantized exports of the combined `AAIT-86M` package using the custom `triembed` architecture metadata. They are not generic llama.cpp text-model artifacts.
 
 ## Operational Caveats