Найден корректный robots.txt
Файл управления сканированием сайта роботами
?
Robots.txt указывает поисковым роботам, какие страницы сканировать, а какие — нет. Ошибки в файле могут случайно закрыть важные разделы от индексации.
Кол-во редиректов для файла robots.txt 3 слишком большое! Это может вызывать затруднение при индексации поисковыми роботами!
Проверяемая страница не запрещена в robots.txt.
Цепочка редиректов для файла robots.txt:
http://goncalofaria.com/robots.txt
301 MovedPermanently
https://goncalofaria.com/robots.txt
301 MovedPermanently
https://www.goncalofaria.com/robots.txt
200 OK
Показать содержимое robots.txt
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
<title>Gonçalo Faria</title>
<link rel="stylesheet" type="text/css" href="style.css">
<script type="module">
// Import the functions you need from the SDKs you need
import { initializeApp } from "https://www.gstatic.com/firebasejs/9.17.1/firebase-app.js";
import { getAnalytics } from "https://www.gstatic.com/firebasejs/9.17.1/firebase-analytics.js";
const firebaseConfig = {
apiKey: "AIzaSyC1igVn6HUX_AbTWxtR2NCmwYkSWnBorUU",
authDomain: "website-31941.firebaseapp.com",
projectId: "website-31941",
storageBucket: "website-31941.appspot.com",
messagingSenderId: "295277225868",
appId: "1:295277225868:web:257c1df026f976b5a418a9",
measurementId: "G-QD7FKYYH3K"
};
// Initialize Firebase
const app = initializeApp(firebaseConfig);
const analytics = getAnalytics(app);
</script>
</head>
<body>
<!-- idea: have a javascript animated loss function converging in the header? :p -->
<div id="dhead" class="container">
<div class="row">
<div id="dpic">
<img src="assets/me.jpeg" class="ppic" />
</div>
<div id="ddesc">
<h1>Gonçalo <strong>Faria</strong></h1>
<div id="status">
<span class="text-success highlight">Ph.D. Student</span> @
<a href="https://noahs-ark.github.io/">Noah's ARK</a>,
<a href="https://www.cs.washington.edu/">UW CSE</a><br>
</div>
<div id="email">
<span class="text-success">gfaria@cs.washington.edu</span><br>
</div>
<div id="social">
<a href="https://twitter.com/goncalorafaria">
<img src="assets/ctwitter.svg" class="iico" alt="Twitter">
</a>
<a href="https://github.com/goncalorafaria">
<img src="assets/cgithub.svg" class="iico" alt="GitHub">
</a>
<a href="https://linkedin.com/in/goncalorafaria/">
<img src="assets/clinkedin.svg" class="iico" alt="LinkedIn">
</a>
<a href="assets/cv.pdf" class="iico">CV</a>
</div>
</div>
</div>
</div>
<div id="history" class="container">
<div class="large-8 large-offset-1 columns">
<div class="metext">
<p>
I am currently an PhD student at <strong>University of Washington</strong>, under the guidance of Professor
<a href="https://nasmith.github.io/">Noah Smith</a>. I studied at the <strong>University of Minho</strong> for my undergrad and at <strong>Instituto Superior
Técnico</strong> for my master's <strong>(Portugal)</strong>.
My master's dissertation, advised by <a href="https://andre-martins.github.io/">André Martins</a> and <a
href="http://www.lx.it.pt/~mtf/">Mário Figueiredo</a>, focused on Differentiable Causal Discovery.</p>
</div>
</div>
</div>
<!---
<div style=" background-color: #eee; padding-top: 1px; margin-top: 10px;">
<div id="featured-talks" class="container">
<div class="ctitle">Projects</div>
<div class="row">
<a href="https://www.questdecoding.com" class="iico">
<div class="card" style="display: flex; align-items: center; justify-content: center; width: 100px; height: 100px; background-color: #000; color: #fff; text-align: center;">
QUEST
</div>
</a>
Increasing test-time computation has emerged as a promising direction for improving language model performance, particularly in scenarios where model finetuning is impractical or impossible due to computational constraints or private model weights. However, existing test-time search methods using a reward model (RM) often degrade in quality as compute scales, due to the over-optimization of what are inherently imperfect reward proxies. We introduce QUEST-RLHF, a new test-time alignment approach. As we scale test-time compute, QUEST-RLHF converges to sampling from the optimal aligned distribution for each individual prompt. By adopting recent advances in Markov chain Monte Carlo for text generation, our method enables better-aligned outputs without modifying the underlying model or even requiring logit access. We demonstrate the effectiveness of \rlhfmethod on mathematical reasoning benchmarks (GSM8K and GSM-Symbolic) using a task-specific RM, showing consistent improvements over existing test-time compute methods like best-of-<i>n</i> and majority voting. Furthermore, when applied with more realistic RMs trained on the Tulu 3 preference dataset, QUEST-RLHF outperforms direct preference optimization (DPO), best-of-<i>n</i>, majority voting, and weighted majority voting on a diverse range of datasets (GSM8K, MATH500, IFEval, MMLU-Redux, and TruthfulQA). A practical solution to aligning language models at test time using additional computation without degradation, our approach expands the limits of the capability that can be obtained from off-the-shelf language models without further training.
</div> -->
<br>
<div class="container">
<div class="ctitle">Publications</div>
<div class="pub highlighted">
<div class="pub-title"><a href="https://www.questdecoding.com/qalign">Sample, Don't Search: Rethinking Test-Time Alignment for Language Models</a></div>
<div class="pub-venue">TBD</div>
<div class="pub-web"><a href="https://www.questdecoding.com/qalign">Website</a> </div>
<div class="pub-paper"><a href="https://arxiv.org/pdf/2504.03790">Paper</a></div>
<div class="pub-authors"> <b>Gonçalo R. A. Faria</b>, Noah Smith</div>
</div>
<div class="pub">
<div class="pub-title"><a href="https://arxiv.org/abs/2410.07779">Modeling User Preferences with Automatic Metrics: Creating a High-Quality Preference Dataset for Machine Translation</a></div>
<div class="pub-venue">EMNLP 2024</div>
<div class="pub-paper"><a href="https://arxiv.org/pdf/2410.07779">Paper</a></div>
<div class="pub-authors">Sweta Agrawal, José G. C. de Souza, Ricardo Rei, António Farinhas,<b> Gonçalo Faria</b>, Patrick Fernandes, Nuno M Guerreiro, Andre Martins</div>
</div>
<div class="pub" >
<div class="pub-title"><a href="https://arxiv.org/abs/2406.00049">QUEST: Quality-Aware Metropolis-Hastings
Sampling
for Machine Translation</a></div>
<div class="pub-venue">NEURIPS 2024</div>
<div class="pub-web"><a href="https://www.questdecoding.com/mt">Website</a> </div>
<div class="pub-paper"><a href="https://arxiv.org/pdf/2406.00049">Paper</a></div>
<div class="pub-authors"><b>Gonçalo R. A. Faria</b>, Sweta Agrawal, António Farinhas, Ricardo Rei, José G. C. de Souza,
André
F. T. Martins</div>
</div>
<div class="pub" href="https://www.goncalofaria.com/lcd.html">
<div class="pub-title"><a href="https://arxiv.org/abs/2203.02336">Differentiable Causal Discovery Under Latent
Interventions</a></div>
<div class="pub-venue">CLEAR 2022</div>
<div class="pub-web"><a href="https://www.goncalofaria.com/lcd.html">Website</a> </div>
<div class="pub-paper"><a href="https://arxiv.org/abs/2203.02336">Paper</a></div>
<div class="pub-authors"><b>Gonçalo R. A. Faria</b>, André F. T. Martins, Mário A. T. Figueiredo</div>
</div>
<div>
<br>
Also on <a href="https://scholar.google.com/citations?user=caTuELoAAAAJ&hl=en&oi=sra">Google Scholar</a>
</div>
</div>
<br><br>
</body>
</html>
