Skip to content

Commit f28ab4b

Browse files
author
alexrame
committed
wip
1 parent b9e6ab4 commit f28ab4b

File tree

12 files changed

+61
-6
lines changed

12 files changed

+61
-6
lines changed

files/resume_alexrame.pdf

27 Bytes
Binary file not shown.

home/index.xml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -18,7 +18,7 @@
1818
<pubDate>Wed, 20 Apr 2016 00:00:00 +0200</pubDate>
1919

2020
<guid>/home/about/</guid>
21-
<description>Research scientist at Google DeepMind, in Paris. My thesis, pursued at Sorbonne University under the supervision of Professor Matthieu Cord, received the 2024 SSFAM award of the best French PhD in ML. My works initially showed how model merging and invariance can improve robustness and generalization in deep learning. I am now exploring how model merging and reinforcement learning can improve the alignment of AIs, with the world, in all its diversity.</description>
21+
<description>Research scientist at Google DeepMind, in Paris. My thesis, pursued at Sorbonne University under the supervision of Professor Matthieu Cord, received the 2024 SSFAM award of the best French PhD in ML. My works initially showed how model merging and invariance can improve robustness and generalization in deep learning. I am now exploring how model merging and reinforcement learning can improve the alignment of AIs with the world in all its diversity.</description>
2222
</item>
2323

2424
<item>

index.html

Lines changed: 21 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -363,7 +363,8 @@ <h3 itemprop="jobTitle">Deep learning scientist</h3>
363363
<p>Research scientist at Google DeepMind, in Paris.
364364
My thesis, pursued at Sorbonne University under the supervision of Professor Matthieu Cord, received the 2024 SSFAM award of the best French PhD in ML.
365365
My works initially showed how model merging and invariance can improve robustness and generalization in deep learning.
366-
I am now exploring how model merging and reinforcement learning can improve the alignment of AIs, with the world, in all its diversity.</p>
366+
I am now exploring how model merging and reinforcement learning can improve the alignment of AIs with the world in all its diversity.
367+
Will be pleased to catch up at ICML 2024.</p>
367368

368369

369370
<div class="row">
@@ -663,6 +664,15 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
663664

664665

665666

667+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warp/warp-slides.pdf" target="_blank"
668+
rel="noopener">
669+
Slides
670+
</a>
671+
672+
673+
674+
675+
666676

667677
</div><br />
668678
</div>
@@ -740,6 +750,15 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
740750

741751

742752

753+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warm/warm-poster.pdf" target="_blank"
754+
rel="noopener">
755+
Poster
756+
</a>
757+
758+
759+
760+
761+
743762

744763
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warm/warm-slides.pdf" target="_blank"
745764
rel="noopener">
@@ -828,7 +847,7 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
828847

829848

830849

831-
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/phd/slides_phd.pdf" target="_blank"
850+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/phd/phd-slides.pdf" target="_blank"
832851
rel="noopener">
833852
Slides
834853
</a>

publication/index.html

Lines changed: 19 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -550,6 +550,15 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
550550

551551

552552

553+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warp/warp-slides.pdf" target="_blank"
554+
rel="noopener">
555+
Slides
556+
</a>
557+
558+
559+
560+
561+
553562

554563
</div>
555564

@@ -637,6 +646,15 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
637646

638647

639648

649+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warm/warm-poster.pdf" target="_blank"
650+
rel="noopener">
651+
Poster
652+
</a>
653+
654+
655+
656+
657+
640658

641659
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/warm/warm-slides.pdf" target="_blank"
642660
rel="noopener">
@@ -735,7 +753,7 @@ <h3 class="article-title mb-0 mt-0" itemprop="name">
735753

736754

737755

738-
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/phd/slides_phd.pdf" target="_blank"
756+
<a class="btn btn-primary btn-outline my-1 mr-1 btn-sm" href="/publication/phd/phd-slides.pdf" target="_blank"
739757
rel="noopener">
740758
Slides
741759
</a>

publication/phd/index.html

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -452,7 +452,7 @@ <h1 itemprop="name">Diverse and Efficient Ensembling of Deep Networks</h1>
452452

453453

454454

455-
<a class="btn btn-primary btn-outline my-1 mr-1" href="/publication/phd/slides_phd.pdf" target="_blank"
455+
<a class="btn btn-primary btn-outline my-1 mr-1" href="/publication/phd/phd-slides.pdf" target="_blank"
456456
rel="noopener">
457457
Slides
458458
</a>

publication/phd/phd-slides.pdf

4.84 MB
Binary file not shown.

publication/phd/slides_phd.pdf

-13.2 MB
Binary file not shown.

publication/warm/cite.bib

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
@inproceedings{rame2024warm,
22
title = {WARM: On the Benefits of Weight Averaged Reward Models},
3-
author = {Ramé, Alexandre and Vieillard, Nino and Hussenot, L{\'e}onard and Dadashi, Robert and Cideron, Geoffrey and Bachem, Olivier and Ferret, Johan},
3+
author = {Ramé, Alexandre and Vieillard, Nino and Hussenot, Léonard and Dadashi, Robert and Cideron, Geoffrey and Bachem, Olivier and Ferret, Johan},
44
booktitle = {ICML},
55
year = {2024}
66
}

publication/warm/index.html

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -487,6 +487,15 @@ <h1 itemprop="name">WARM: On the Benefits of Weight Averaged Reward Models</h1>
487487

488488

489489

490+
<a class="btn btn-primary btn-outline my-1 mr-1" href="/publication/warm/warm-poster.pdf" target="_blank"
491+
rel="noopener">
492+
Poster
493+
</a>
494+
495+
496+
497+
498+
490499

491500
<a class="btn btn-primary btn-outline my-1 mr-1" href="/publication/warm/warm-slides.pdf" target="_blank"
492501
rel="noopener">

publication/warm/warm-poster.pdf

601 KB
Binary file not shown.

0 commit comments

Comments
 (0)