Aether step 40000 (weights only; architecture proprietary)

README.md

@@ -4,18 +4,18 @@ tags: [aether, anima, reasoning]
 library_name: aether
 ---
 
-# FatStinkyPanda · Aether — step 23,000
+# FatStinkyPanda · Aether — step 40,000
 
 A from-scratch reasoning LLM by **FatStinkyPanda** (Daniel A. Bissey), part of the **Anima** project.
 
 > **Proprietary architecture.** The model architecture and training innovations are NOT public. This repo hosts the trained **weights + tokenizer only**; the config + modeling code live in a private repo. Benchmarks are run on **neutral free compute (Kaggle)** and published openly so the numbers are verifiable without exposing how the model works.
 
 ## Latest held-out benchmarks (acc_norm; neutral compute)
-- **hellaswag**: 0.32
-- **arc_easy**: 0.32
-- **arc_challenge**: 0.24
-- **winogrande**: 0.52
-- **openbookqa**: 0.25
-- **train_slice_ppl**: 20.95
+- **hellaswag**: 0.33
+- **arc_easy**: 0.35
+- **arc_challenge**: 0.27
+- **winogrande**: 0.57
+- **openbookqa**: 0.29
+- **train_slice_ppl**: 12.02
 
 Created & released by **FatStinkyPanda**. For access or collaboration, contact the creator.

latest.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:349348e1dfb632a23172c44470022e70046c2ba79527d636861d97bbcc58e141
+oid sha256:c009998e077014e43a1bfd875f4fff644fe3293cadf753a3a81dbbb3ed72efbd
 size 1303007612

metadata.json

@@ -3,6 +3,6 @@
   "developer": "Daniel A. Bissey",
   "model": "FatStinkyPanda/Aether",
   "project": "Anima",
-  "step": 23000,
+  "step": 40000,
   "config_name": "aether-100m-128k.yaml"
 }