Update README.md

README.md

@@ -14,6 +14,7 @@ datasets:
 - BEE-spoke-data/fineweb-1M_longish
 language:
 - en
+inference: false
 ---
 
 # jamba-900M-v0.13-KIx2
@@ -22,14 +23,18 @@ language:
   <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/>
 </a>
 
-This is a pretraining experiment on the `jamba` arch as a "smol MoE". Details:
+> The API widget is off as it isn't supported by hf yet - try the Colab
+
+This is a pretraining experiment on the `jamba` arch as a "smol MoE". 
+
+Details:
 
 - pretrained at context length 16384
 - seen approx 20b tokens
 - uses Claude3 tokenizer (as hf GPT2 tokenizer)
 - hidden size 1024, 12 layers, 8 experts
 
-most recent dataset, achieves the following results on the evaluation set:
+achieves the following results on the evaluation set (_ of the latest dataset_):
 - Loss: 3.0366
 - Accuracy: 0.4514
 - Num Input Tokens Seen: 1975517184