Card: checkpoints/1217308k (step 62,895)

README.md

@@ -46,6 +46,16 @@ This card is written by program from the checkpoint's own record, at step
 
 ## Architecture
 
+Nirca was inspired by [mini-AGI](https://github.com/volotat/mini-AGI), and its architecture is
+deliberately derived from it: Nirca began as a port of mini-AGI to MLX.
+
+- **Kept:** bytes, no tokenizer; latent recurrence; one shared pool of experts
+  every pass routes into; a halting head; first pretraining text mini-AGI's
+  corpus.
+- **Changed:** ternary weights, trained quantization-aware from step 14,655 of its lineage (its format before that is not recorded);
+  tuned on chats; chat format: ChatML instead of mini-AGI's own tags;
+  a fixed 32-expert pool where mini-AGI grows and prunes its own.
+
 ![Nirca Mini's architecture](architecture.svg)
 
 The two dense layers run once. The recurrent block then runs again and again,
@@ -121,11 +131,10 @@ It is a small research model, and it is often wrong or incoherent. It may
 repeat biases in its training data. It is trained mostly on English, and it has
 no knowledge of events beyond what its data held. Verify anything it says.
 
-## Inspiration and starting point
+## Citation
 
-Nirca Mini began as a port of [mini-AGI](https://github.com/volotat/mini-AGI) to MLX: its latent
-recurrence, shared expert pool, and halting come from there, and its first
-pretraining text was mini-AGI's corpus.
+Nirca Mini's architecture is derived from [mini-AGI](https://github.com/volotat/mini-AGI) (see
+Architecture):
 
 ```bibtex
 @misc{volotat_miniagi,