Skip to content

Commit 240ac36

Browse files
committed
rsi
1 parent f662fe9 commit 240ac36

File tree

5 files changed

+549
-132
lines changed

5 files changed

+549
-132
lines changed

docs/posts/2025-09-13-recursive-self-improvement-explosion-optimization.html

Lines changed: 61 additions & 132 deletions
Original file line numberDiff line numberDiff line change
@@ -331,7 +331,7 @@ <h1>Summary (OLD)</h1>
331331
<li><strong>Measuring capability.</strong></li>
332332
<li><strong>Bottlenecks.</strong> Some arguments that AI R&amp;D is bottlenecked by compute, e.g.&nbsp;see <span class="citation" data-cites="whitfill2025bottlenecks">Whitfill and Wu (<a href="#ref-whitfill2025bottlenecks" role="doc-biblioref">2025</a>)</span>.</li>
333333
<li><strong>Scale-dependent algorithmic progress.</strong> <span class="citation" data-cites="gundlach2025algorithmicprogressai">Gundlach et al. (<a href="#ref-gundlach2025algorithmicprogressai" role="doc-biblioref">2025</a>)</span> argue that algorithmic progress has contributed much less.</li>
334-
<li><strong>Data contribution.</strong> Berren Millidge argues <a href="https://www.beren.io/2025-08-02-Most-Algorithmic-Progress-is-Data-Progress/">“Most Algorithmic Progress is Data Progress”</a>. Data has been growing slower than compute.</li>
334+
<li><strong>Data contribution.</strong> Berren Millidge argues <a href="https://www.beren.io/2025-08-02-Most-Algorithmic-Progress-is-Data-Progress/">“Most Algorithmic Progress is Data Progress”</a></li>
335335
</ul></li>
336336
</ol>
337337
<div class="columns" style="height: 100vh; gap: 2rem;">
@@ -1082,63 +1082,25 @@ <h2 class="anchored" data-anchor-id="jones1995rd-rd-based-models-of-economic-gro
10821082
&amp;&amp; \text{(allocation)}
10831083
\end{aligned}
10841084
\]</span></p>
1085-
<p>Balanced-growth implications: <span class="math display">\[\begin{aligned}
1086-
g_A &amp;= \frac{\lambda}{\beta}n \\
1087-
g_y &amp;= \sigma g_A = \frac{\lambda\sigma}{\beta}n = \gamma n
1088-
\end{aligned}
1089-
\]</span></p>
1090-
<p>Parameters:</p>
1091-
<table class="caption-top table">
1092-
<colgroup>
1093-
<col style="width: 24%">
1094-
<col style="width: 75%">
1095-
</colgroup>
1096-
<tbody>
1097-
<tr class="odd">
1098-
<td style="text-align: center;"><span class="math inline">\(\beta \approx 3\)</span></td>
1099-
<td>how fast proportional improvements are getting harder to find. Jones writes that aggregate data are roughly consistent with <span class="math inline">\(\beta \approx 3\)</span> if <span class="math inline">\(\lambda = 1\)</span>.</td>
1100-
</tr>
1101-
<tr class="even">
1102-
<td style="text-align: center;"><span class="math inline">\(\lambda \approx 1\)</span></td>
1103-
<td>elasticity of idea production with respect to research effort. <span class="math inline">\(\lambda = 1\)</span> means no duplication effect; <span class="math inline">\(\lambda &lt; 1\)</span> allows duplication/congestion.</td>
1104-
</tr>
1105-
<tr class="odd">
1106-
<td style="text-align: center;"><span class="math inline">\(\sigma &gt; 0\)</span></td>
1107-
<td>how strongly nonrival ideas raise final output; this is the degree of increasing returns in goods production. He doesn’t separately calibrate.</td>
1108-
</tr>
1109-
<tr class="even">
1110-
<td style="text-align: center;"><span class="math inline">\(\gamma \equiv \frac{\lambda\sigma}{\beta} \approx 1/3\)</span></td>
1111-
<td>the overall degree of increasing returns in this simple semi-endogenous setup. Jones says <span class="math inline">\(\gamma = 1/3\)</span> is consistent with data when research intensity has been rising.</td>
1112-
</tr>
1113-
</tbody>
1114-
</table>
1115-
<p><strong>Visually:</strong></p>
1116-
<ol type="1">
1117-
<li><p>Basic model with research effort but no knowledge term. If <span class="math inline">\(R\)</span> is constant, then <span class="math inline">\(\dot{A}\)</span> is constant, so <span class="math inline">\(A\)</span> rises linearly and the growth rate <span class="math inline">\(g_A = \dot{A}/A\)</span> declines toward zero: <span class="math display">\[\begin{gathered}
1118-
\dot{A}=R^\lambda\\
1119-
\xymatrix{*++[F]{R\&amp;D} \ar[r]|(0.4)\lambda &amp; *++[F]{\Delta knowledge}\ar[r] &amp; *++[F]{knowledge}}
1120-
\end{gathered}
1121-
\]</span></p></li>
1122-
<li><p>Add the knowledge term: <span class="math display">\[\begin{gathered}
1123-
\dot{A}=R^\lambda A^{1-\beta}\\
1124-
\xymatrix{*++[F]{R\&amp;D} \ar[r]|(0.4)\lambda &amp; *++[F]{\Delta knowledge}\ar[r] &amp; *++[F]{knowledge}\ar@/^2em/[l]|{1-\beta}}
1125-
\end{gathered}
1126-
\]</span></p>
1127-
<p>Then <span class="math inline">\(g_A = \frac{\dot{A}}{A} = R^\lambda A^{-\beta}.\)</span></p>
1128-
<p>So if <span class="math inline">\(R\)</span> is constant, <span class="math inline">\(g_A\)</span> declines as <span class="math inline">\(A\)</span> rises.<a href="#fn1" class="footnote-ref" id="fnref1" role="doc-noteref"><sup>1</sup></a> If <span class="math inline">\(R\)</span> grows at rate <span class="math inline">\(g_R\)</span> along a balanced growth path, then <span class="math inline">\(g_A = \frac{\lambda}{\beta}g_R.\)</span></p></li>
1129-
<li><p>Recursive self-improvement, where knowledge directly raises research input: <span class="math display">\[\begin{gathered}
1130-
R=A^\kappa\\
1131-
\dot{A}=(A^\kappa)^\lambda A^{1-\beta}=A^{\lambda\kappa+1-\beta}\\
1132-
\xymatrix{*++[F]{R\&amp;D} \ar[r]|(0.4)\lambda &amp; *++[F]{\Delta knowledge}\ar[r] &amp; *++[F]{knowledge}\ar@/^2em/[l]|{1-\beta}\ar@/^4em/[ll]|\kappa}
1133-
\end{gathered}
1134-
\]</span></p>
1135-
<p>This yields:</p>
1085+
<p>He also says he’ll sometimes write <span class="math inline">\(\dot{A}_t=R^\gamma_t A_t^\psi\)</span>.</p>
1086+
<blockquote class="blockquote">
1087+
<p>“The parameter <span class="math inline">\(\lambda\)</span> allows for thepossibility of duplicatione ffects,so that doubling the number of researchers at a point in time may potentially less than double the innovation rate; however, any λ&gt;0, including λ=1, is allowed.”</p>
1088+
</blockquote>
1089+
<blockquote class="blockquote">
1090+
<p>“The parameter β &gt; 0 captures the rate at which ideas—that is,proportional improvements in productivity—are getting harder to find.</p>
1091+
</blockquote>
1092+
<p>On balanced growth path: <span class="math display">\[g_y = \sigma g_A = \frac{\lambda\sigma}{\beta}n=\gamma n\]</span></p>
1093+
<p>On knife edge assumptions: he says the root of all exponential growth is exponential population growth (but seems a bit fishy).</p>
1094+
<p>Estimates of parameters:</p>
11361095
<ul>
1137-
<li>if <span class="math inline">\(\lambda\kappa &lt; \beta\)</span>, growth slows over time;</li>
1138-
<li>if <span class="math inline">\(\lambda\kappa = \beta\)</span>, you get constant exponential growth;</li>
1139-
<li>if <span class="math inline">\(\lambda\kappa &gt; \beta\)</span>, the model implies a finite-time singularity.</li>
1140-
</ul></li>
1141-
</ol>
1096+
<li>p136: <span class="math inline">\(\gamma = 1/3\)</span>, so output growth is 1/3 growth in ideas. He also predicts that long-run growth is much lower.</li>
1097+
</ul>
1098+
<p>Description of <span class="citation" data-cites="aghion2019artificial">Aghion, Jones, and Jones (<a href="#ref-aghion2019artificial" role="doc-biblioref">2019</a>)</span>. Adjust the ideas production function:</p>
1099+
<p><span class="math display">\[\dot{A}_t=A_t^{1-\beta}R_t^{1-\alpha}K_t^{\alpha}\]</span></p>
1100+
<p>with constant capital-output ratio you get: <span class="math display">\[\dot{A}_t=\kappa A_t^{1-(\beta-\alpha)}L_t\]</span></p>
1101+
<blockquote class="blockquote">
1102+
<p>“if the fraction of tasks that are automated (α) rises to reach the rate at which ideas are getting harder to find (β), we get a singularity! In particular, once α ≥ β, the model exhibits sufficiently strong increasing returns that there is no balanced growth path. Instead, the growth rate rises rapidly over time and the economy reaches infinite knowledge and income in finite time,assuming that is possible.”</p>
1103+
</blockquote>
11421104
</section>
11431105
<section id="aghion2019artificial" class="level2">
11441106
<h2 class="anchored" data-anchor-id="aghion2019artificial"><span class="citation" data-cites="aghion2019artificial">Aghion, Jones, and Jones (<a href="#ref-aghion2019artificial" role="doc-biblioref">2019</a>)</span></h2>
@@ -1457,6 +1419,7 @@ <h2 class="anchored" data-anchor-id="davidson2025howquickandbigwo"><span class="
14571419
<li>The paper calibrates primarily in terms of <span class="math inline">\(p\)</span> and <span class="math inline">\(r\)</span>, not by directly estimating <span class="math inline">\(\beta\)</span>.</li>
14581420
<li>Their reported medians are <span class="math inline">\(p=0.3\)</span> and <span class="math inline">\(r=1.2\)</span>, and one representative decomposition is <span class="math inline">\(\alpha=0.5\)</span>, <span class="math inline">\(\lambda=0.6\)</span>, and <span class="math inline">\(\beta=0.25\)</span>.</li>
14591421
</ul>
1422+
<p>then they note <span class="math inline">\(r=\lambda \alpha / \beta\)</span>, and this is critical</p>
14601423
</section>
14611424
<section id="kokotajlo2025aifuturesmodel-ai-futures-model-dec-2025-update" class="level2">
14621425
<h2 class="anchored" data-anchor-id="kokotajlo2025aifuturesmodel-ai-futures-model-dec-2025-update"><span class="citation" data-cites="kokotajlo2025aifuturesmodel">Kokotajlo et al. (<a href="#ref-kokotajlo2025aifuturesmodel" role="doc-biblioref">2025</a>)</span> “AI Futures Model: Dec 2025 Update”</h2>
@@ -1534,72 +1497,61 @@ <h2 class="anchored" data-anchor-id="davidson2026automatingairesearch-when-does-
15341497
<li>You gradually automate some of the things, so they can be done with capital.</li>
15351498
<li>You have spillovers between different processes.</li>
15361499
</ul></li>
1500+
<li>Estimates of <span class="math inline">\(\beta\)</span>:
1501+
<ul>
1502+
<li>Bloom overall <span class="math inline">\(\beta=3\)</span></li>
1503+
<li>For software R&amp;D they estimate <span class="math inline">\(\beta=3\)</span></li>
1504+
<li>they say in software <span class="math inline">\(\beta=0.1\)</span></li>
1505+
</ul></li>
1506+
<li>Questions:
1507+
<ul>
1508+
<li>Any historical domain where we’ve seen regimes of <span class="math inline">\(\beta&lt;0\)</span>?, &amp; so explosion.</li>
1509+
</ul></li>
15371510
</ul>
15381511
<p>Implications / thresholds:</p>
15391512
<ul>
1540-
<li>In the one-sector software model, the key threshold is <span class="math inline">\(r = \lambda\alpha/\beta_S\)</span>.</li>
1541-
<li>In the multi-sector model, explosive growth can arise from the combined automation of goods production, software R&amp;D, hardware progress, and aggregate innovation, even when no single channel is decisive on its own.</li>
1513+
<li><p>Q: how to think about the growth effect of uplift vs automation?</p></li>
1514+
<li><p>It seems to me AI typically makes things effectively <em>free</em>, rather than being bottlenecked by capital. I think this is a problem for Cobb-Douglas production.</p></li>
15421515
</ul>
15431516
</section>
1517+
<section id="kokotajlo2025aifuturesmodel-ai-futures-model" class="level2">
1518+
<h2 class="anchored" data-anchor-id="kokotajlo2025aifuturesmodel-ai-futures-model"><span class="citation" data-cites="kokotajlo2025aifuturesmodel">Kokotajlo et al. (<a href="#ref-kokotajlo2025aifuturesmodel" role="doc-biblioref">2025</a>)</span> AI futures model</h2>
1519+
<p>https://www.timelinesmodel.com</p>
1520+
<p>They assume AI automates some fraction of R&amp;D tasks, proportional to the time horizon.</p>
1521+
</section>
15441522
<section id="kwa2026simpleraitimelines-tkwa-model" class="level2">
15451523
<h2 class="anchored" data-anchor-id="kwa2026simpleraitimelines-tkwa-model"><span class="citation" data-cites="kwa2026simpleraitimelines">Kwa (<a href="#ref-kwa2026simpleraitimelines" role="doc-biblioref">2026</a>)</span> tkwa&nbsp; model</h2>
1546-
<p>Equations:</p>
1547-
<p><span class="math display">\[
1548-
\begin{aligned}
1549-
S'(t) &amp;= R(t) S^{1 - \beta} = \left(\frac L {1-f}\right)^\alpha C^\zeta S^{1 - \beta}
1550-
&amp;&amp; \text{(accumulation of software ideas)}\\
1524+
<p>Short: AI automates some fraction of R&amp;D tasks, which causes an effective multiplier on R&amp;D labor: “parallel uplift and 1/(1-f) are equivalent in the simple model”.</p>
1525+
<p>https://github.com/tkwa/ai-takeoff-model/blob/main/notes.md https://www.lesswrong.com/posts/uy6B5rEPvcwi55cBK/research-note-a-simpler-ai-timelines-model-predicts-99-ai-r</p>
1526+
<p><span class="math display">\[\begin{aligned}
1527+
S'(t) &amp;= R(t) S^{1 - \beta} = \left(\frac L {1-f}\right)^\alpha C^\zeta S^{1 - \beta}\\
15511528
f(t) &amp;= \sigma(v(\log C(t)S(t) - \log E_{hac}))
1552-
&amp;&amp; \text{(automation of tasks)}
15531529
\end{aligned}
15541530
\]</span></p>
1555-
<p>Automating a fraction <span class="math inline">\(f\)</span> of R&amp;D tasks multiplies effective R&amp;D labor by <span class="math inline">\(1/(1-f)\)</span> (i.e.&nbsp;tasks are perfect complements).</p>
1531+
<p>My simpler version:</p>
15561532
<p><span class="math display">\[\begin{aligned}
15571533
\dot{S} &amp;= L^\alpha C^\gamma S^{1-\beta} \\
15581534
\end{aligned}
15591535
\]</span></p>
1560-
<p>Parameters:</p>
1561-
<table class="caption-top table">
1562-
<colgroup>
1563-
<col style="width: 19%">
1564-
<col style="width: 80%">
1565-
</colgroup>
1566-
<tbody>
1567-
<tr class="odd">
1568-
<td><span class="math inline">\(\beta\in [0.3,1]\)</span></td>
1569-
<td>software-difficulty exponent.</td>
1570-
</tr>
1571-
<tr class="even">
1572-
<td><span class="math inline">\(\alpha\)</span></td>
1573-
<td>not pinned down separately; together with <span class="math inline">\(\zeta\)</span>, Kwa uses <span class="math inline">\(\alpha/(\alpha+\zeta)\in [0.12,0.35]\)</span> and <span class="math inline">\(\alpha+\zeta\in [0.8,1]\)</span>, implying roughly <span class="math inline">\(\alpha\in [0.10,0.35]\)</span>.</td>
1574-
</tr>
1575-
<tr class="odd">
1576-
<td><span class="math inline">\(\zeta\)</span></td>
1577-
<td>not pinned down separately; implied by the same calibration above, roughly <span class="math inline">\(\zeta\in [0.52,0.88]\)</span>.</td>
1578-
</tr>
1579-
<tr class="even">
1580-
<td><span class="math inline">\(f(t)\)</span></td>
1581-
<td>fraction of R&amp;D tasks automated; Kwa sets current <span class="math inline">\(f\)</span> in Jan 2026 to lie in <span class="math inline">\([0.25,0.5]\)</span>.</td>
1582-
</tr>
1583-
<tr class="odd">
1584-
<td><span class="math inline">\(v\)</span></td>
1585-
<td>automation velocity; Kwa uses <span class="math inline">\(1/v\in [1.5,4.2]\)</span>, so <span class="math inline">\(v\in [0.24,0.67]\)</span>.</td>
1586-
</tr>
1587-
<tr class="even">
1588-
<td><span class="math inline">\(E_{hac}\)</span></td>
1589-
<td>effective compute level of the half-automated coder; not directly estimated here, but defined as the point where automation reaches 50%.</td>
1590-
</tr>
1591-
</tbody>
1592-
</table>
1593-
<p>Relevant notes on interpretation / sources:</p>
1594-
<blockquote class="blockquote">
1595-
<p><span class="math inline">\(E_{hac}\)</span> is the effective compute level of the half-automated coder”</p>
1596-
</blockquote>
1597-
<blockquote class="blockquote">
1598-
<p>“v is the automation velocity: S must increase by factor of e^(1/v) to get from 50% to 73% automation”</p>
1599-
</blockquote>
1600-
<blockquote class="blockquote">
1601-
<p>“alpha/(alpha + zeta) is between 0.12 and 0.35 … This range is based on Yafah’s (Epoch) recommendation to calibrate from lab spending ratios of labor vs capital.”</p>
1602-
</blockquote>
1536+
<p>We can show:</p>
1537+
<p><span class="math display">\[S^{\beta} = S_0^{\beta} + \frac{\beta\,L_0^\alpha C_0^\gamma}{\alpha g_L+\gamma g_C}\Big(e^{(\alpha g_L+\gamma g_C)t}-1\Big).\]</span></p>
1538+
<p>In the limit we get <span class="math inline">\(g_s=\frac{\alpha g_L+\gamma g_C}{\beta}\)</span></p>
1539+
</section>
1540+
<section id="david-rein-notes" class="level2">
1541+
<h2 class="anchored" data-anchor-id="david-rein-notes">David Rein notes</h2>
1542+
<ul>
1543+
<li>Model:
1544+
<ul>
1545+
<li><span class="math inline">\(dA = Q^q A^{1-\beta}\)</span></li>
1546+
<li><span class="math inline">\(Q\)</span> is the quality of cognitive labour, measured in units of time horizon—this is our AI researcher (e.g.&nbsp;Opus 4.5, GPT-5.2, Gemini 3)</li>
1547+
<li><span class="math inline">\(A\)</span> is the current level of algorithms, also measured in units of time horizon (e.g.&nbsp;GPT-2, GPT-3, Pythia, etc.)</li>
1548+
<li><span class="math inline">\(\beta\)</span> is the ideas getting harder to find parameter</li>
1549+
<li>The explosion condition here is <span class="math inline">\(q &gt; \beta\)</span></li>
1550+
<li>(Must also assume that <span class="math inline">\(Q=cA\)</span>.</li>
1551+
</ul></li>
1552+
<li><a href="https://docs.google.com/document/d/16Ugl7g3GL1Ao9-3UluRENp6e9UeEMAqhh5jgXDi3kjo/edit?tab=t.a5tu3xadb63o#heading=h.ribcf6q6wbps">doc</a></li>
1553+
</ul>
1554+
<hr>
16031555
<div class="cell" data-layout-align="default">
16041556
<div class="cell-output-display">
16051557
<div>
@@ -1657,23 +1609,6 @@ <h2 class="anchored" data-anchor-id="kwa2026simpleraitimelines-tkwa-model"><span
16571609
</div>
16581610
</div>
16591611
</div>
1660-
</section>
1661-
<section id="david-rein-2025-model" class="level2">
1662-
<h2 class="anchored" data-anchor-id="david-rein-2025-model">David Rein (2025) model</h2>
1663-
<p>The core idea: idea production depends on quality:</p>
1664-
<p><span class="math display">\[\begin{aligned}
1665-
dA &amp;= Q^q A^{1-\beta}
1666-
&amp;&amp; \text{(idea production depends on quality of cognitive labor)}\\
1667-
Q &amp;= cA
1668-
&amp;&amp; \text{(quality depends on ideas)}
1669-
\end{aligned}
1670-
\]</span></p>
1671-
<p>Parameters: | | | | ——- | —————————————————————————————————————————– | | <span class="math inline">\(Q\)</span> | quality of cognitive labour, measured in units of time horizon; this is our AI researcher (e.g.&nbsp;Opus 4.5, GPT-5.2, Gemini 3). | | <span class="math inline">\(A\)</span> | current level of algorithms, also measured in units of time horizon (e.g.&nbsp;GPT-2, GPT-3, Pythia, etc.). | | <span class="math inline">\(q\)</span> | elasticity of idea production with respect to cognitive-labor quality. | | <span class="math inline">\(\beta\)</span> | ideas-get-harder-to-find parameter. |</p>
1672-
<p>Implications / thresholds:</p>
1673-
<ul>
1674-
<li>The explosion condition is <span class="math inline">\(q &gt; \beta\)</span>.</li>
1675-
</ul>
1676-
<p><a href="https://docs.google.com/document/d/16Ugl7g3GL1Ao9-3UluRENp6e9UeEMAqhh5jgXDi3kjo/edit?tab=t.a5tu3xadb63o#heading=h.ribcf6q6wbps">source</a></p>
16771612

16781613

16791614

@@ -1721,9 +1656,6 @@ <h2 class="anchored" data-anchor-id="david-rein-2025-model">David Rein (2025) mo
17211656
<div id="ref-EpochAITrends2026" class="csl-entry" role="listitem">
17221657
———. 2026b. <span>“Trends in Artificial Intelligence.”</span> <a href="https://epoch.ai/trends">https://epoch.ai/trends</a>.
17231658
</div>
1724-
<div id="ref-erdil2025automatingrd" class="csl-entry" role="listitem">
1725-
Erdil, Ege, and Matthew Barnett. 2025. <span>“Most AI Value Will Come from Broad Automation, Not from r&amp;d.”</span> <a href="https://epoch.ai/gradient-updates/most-ai-value-will-come-from-broad-automation-not-from-r-d">https://epoch.ai/gradient-updates/most-ai-value-will-come-from-broad-automation-not-from-r-d</a>.
1726-
</div>
17271659
<div id="ref-erdil2024explosive" class="csl-entry" role="listitem">
17281660
Erdil, Ege, and Tamay Besiroglu. 2024. <span>“Explosive Growth from AI Automation: A Review of the Arguments.”</span> <a href="https://arxiv.org/abs/2309.11690">https://arxiv.org/abs/2309.11690</a>.
17291661
</div>
@@ -1745,9 +1677,6 @@ <h2 class="anchored" data-anchor-id="david-rein-2025-model">David Rein (2025) mo
17451677
<div id="ref-ho2024algorithmicprogresslm" class="csl-entry" role="listitem">
17461678
Ho, Anson, Tamay Besiroglu, Ege Erdil, David Owen, Robi Rahman, Zifan Carl Guo, David Atkinson, Neil Thompson, and Jaime Sevilla. 2024. <span>“Algorithmic Progress in Language Models.”</span> <a href="https://doi.org/10.48550/arXiv.2403.05812">https://doi.org/10.48550/arXiv.2403.05812</a>.
17471679
</div>
1748-
<div id="ref-ho2025explosionexperiments" class="csl-entry" role="listitem">
1749-
Ho, Anson, and Parker Whitfill. 2025. <span>“The Software Intelligence Explosion Debate Needs Experiments.”</span> <a href="https://epoch.ai/gradient-updates/the-software-intelligence-explosion-debate-needs-experiments">https://epoch.ai/gradient-updates/the-software-intelligence-explosion-debate-needs-experiments</a>.
1750-
</div>
17511680
<div id="ref-jones2025aird" class="csl-entry" role="listitem">
17521681
Jones, Benjamin F. 2025. <span>“Artificial Intelligence in Research and Development.”</span> NBER Working Paper 34312. National Bureau of Economic Research. <a href="https://doi.org/10.3386/w34312">https://doi.org/10.3386/w34312</a>.
17531682
</div>

docs/site_libs/bootstrap/bootstrap-d5e7c60e6424aa6ccf163f01508596ce.min.css

Lines changed: 12 additions & 0 deletions
Some generated files are not rendered by default. Learn more about customizing how changed files appear on GitHub.

0 commit comments

Comments
 (0)