Aether step 2000 (weights only; architecture proprietary)

README.md

@@ -1,12 +1,21 @@
 ---
 license: other
 tags: [aether, anima, reasoning]
+library_name: aether
 ---
 
-# FatStinkyPanda · Aether
+# FatStinkyPanda · Aether — step 2,000
 
 A from-scratch reasoning LLM by **FatStinkyPanda** (Daniel A. Bissey), part of the **Anima** project.
 
-> **Proprietary architecture.** Architecture + training innovations are NOT public. This repo hosts the trained **weights + tokenizer only**. Benchmarks run on neutral free compute (Kaggle) and are published openly so the numbers are verifiable without exposing how the model works.
+> **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**: 0.29
+- **arc_easy**: 0.25
+- **arc_challenge**: 0.17
+- **winogrande**: 0.58
+- **openbookqa**: 0.2
+- **train_slice_ppl**: 38.26
 
 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 sha256:024f576096e018886b73a8a4287bb873b9073aaafefd14e55e392be79a7633a6
-size 1284657420
+oid sha256:08160f2c9aec7c0608ed7bae4f8ff3ce0b2f3e6befcc9a314671b7a4128bd87a
+size 1303007612

metadata.json

@@ -3,5 +3,6 @@
   "developer": "Daniel A. Bissey",
   "model": "FatStinkyPanda/Aether",
   "project": "Anima",
-  "step": 90
+  "step": 2000,
+  "config_name": "aether-100m-128k.yaml"
 }