From 08d5053663a4e75989e5106c7a157e4d50a8f396 Mon Sep 17 00:00:00 2001
From: Cameron Otsuka <cameron@otsuka.haus>
Date: Fri, 7 Feb 2025 18:18:13 -0800
Subject: [PATCH] standardize figure captions

---
 content/articles/minimum-utxo-value/index.md               | 4 ++--
 content/articles/r1-vs-o1-ai-as-commodity-or-moat/index.md | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/content/articles/minimum-utxo-value/index.md b/content/articles/minimum-utxo-value/index.md
index 15018a1..aac8383 100644
--- a/content/articles/minimum-utxo-value/index.md
+++ b/content/articles/minimum-utxo-value/index.md
@@ -25,7 +25,7 @@ The highest fee periods in Bitcoin history have been 2018, 2021, and 2023:
 
 <figure>
     {% image "./minimum-utxo-value_mempool_vbytes_history.svg", "Historical graph of Bitcoin mempool by vBytes" %}
-    <figcaption>Source: <a href="https://mempool.space/graphs/mempool#all">mempool.space</a></figcaption>
+    <figcaption>Source: <a href="https://mempool.space/graphs/mempool#all">mempool.space</a>.</figcaption>
 </figure>
 
 I've selected 4 blocks to deep dive, based on their proximity to mempool peaks in those three years, as well as a more "normal" current block:
@@ -36,7 +36,7 @@ I've selected 4 blocks to deep dive, based on their proximity to mempool peaks i
 
 <figure>
     {% image "./minimum-utxo-value_box-whisker-overall.svg", "Box and whisker plot of fee rates across blocks" %}
-    <figcaption>Outliers removed</figcaption>
+    <figcaption>Outliers removed.</figcaption>
 </figure>
 
 | **Fee Rate (sats/vB)** | **504000** | **680000** | **782400** | **801171** |
diff --git a/content/articles/r1-vs-o1-ai-as-commodity-or-moat/index.md b/content/articles/r1-vs-o1-ai-as-commodity-or-moat/index.md
index 59fa2ae..9aa5535 100644
--- a/content/articles/r1-vs-o1-ai-as-commodity-or-moat/index.md
+++ b/content/articles/r1-vs-o1-ai-as-commodity-or-moat/index.md
@@ -27,21 +27,21 @@ DeepSeek trained R1 for an estimated $5.6MM on A100 and H800 GPUs (not the lates
 
 <figure>
     {% image "./input-output-pricing.jpg", "Input/Output Pricing for o1-Class Inference Models ($/1M Tokens)" %}
-    <figcaption>Source: <a href="https://x.com/RnaudBertrand/status/1881709223152878000/photo/1">@RnaudBertrand</a></figcaption>
+    <figcaption>Source: <a href="https://x.com/RnaudBertrand/status/1881709223152878000/photo/1">@RnaudBertrand</a>.</figcaption>
 </figure>
 
 Despite less capital expenditure, R1 has comparable performance to o1 due to several technical innovations. I don't pretend to deeply understand everything on this front, so instead I'll direct you [to this paper](https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf), which breaks down what's happening under the hood. Ultimately, R1 is essentially a drop-in replacement for o1.
 
 <figure>
     {% image "./r1-o1-benchmarks.jpg", "Accuracy / Percentile (%) by Benchmark" %}
-    <figcaption>Source: <a href="https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf">DeepSeek-AI</a></figcaption>
+    <figcaption>Source: <a href="https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf">DeepSeek-AI</a>.</figcaption>
 </figure>
 
 This is incredible for the little guy: run your own model, privately, without the need for massive infrastructure.
 
 <figure>
     {% image "./r1-on-iphone-localghost.gif", "R1 running locally on an iPhone" %}
-    <figcaption>A distilled version of R1 running locally on an iPhone. Source: <a href="https://x.com/localghost/status/1882109711732154387">@localghost</a></figcaption>
+    <figcaption>A distilled version of R1 running locally on an iPhone. Source: <a href="https://x.com/localghost/status/1882109711732154387">@localghost</a>.</figcaption>
 </figure>
 
 ## Is OpenAI Cooked?
-- 
2.51.0