-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
470 lines (454 loc) · 27.8 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
<!doctype html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>agentic learning ai lab</title>
<meta name="description" content="" />
<!-- <link
rel="shortcut icon"
href="./assets/logo.png"
type="image/x-icon"
/> -->
<!-- Open Graph / Facebook -->
<meta property="og:title" content="agentic learning ai lab" />
<meta property="og:description" content="agentic learning ai lab" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://agenticlearning.ai/" />
<!--Replace with the current website url-->
<meta property="og:image" content="" />
<link rel="stylesheet" href="./css/tailwind-runtime.css" />
<!-- <link rel="stylesheet" href="./css/tailwind-build.css" /> -->
<link rel="stylesheet" href="css/index.css" />
<link
rel="stylesheet"
href="https://fonts.googleapis.com/icon?family=Material+Icons"
/>
<link
rel="stylesheet"
href="https://cdnjs.cloudflare.com/ajax/libs/bootstrap-icons/1.11.3/font/bootstrap-icons.min.css"
integrity="sha512-dPXYcDub/aeb08c63jRq/k6GaKccl256JQy/AnOq7CAnEZ9FzSL9wSbcZkMp4R26vBsMLFYH4kQ67/bbV8XaCQ=="
crossorigin="anonymous"
referrerpolicy="no-referrer"
/>
<script src="https://www.google.com/recaptcha/api.js" async defer></script>
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-44MVTGBV0D"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-44MVTGBV0D');
</script>
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/@emailjs/browser@4/dist/email.min.js"></script>
<script type="text/javascript">
(function() {
// https://dashboard.emailjs.com/admin/account
emailjs.init({
publicKey: "y6ebNBhEpEzmyqS8F",
});
})();
</script>
<script
src="https://cdnjs.cloudflare.com/ajax/libs/gsap/3.12.0/gsap.min.js"
integrity="sha512-B1lby8cGcAUU3GR+Fd809/ZxgHbfwJMp0jLTVfHiArTuUt++VqSlJpaJvhNtRf3NERaxDNmmxkdx2o+aHd4bvw=="
crossorigin="anonymous"
referrerpolicy="no-referrer"
></script>
<script
src="https://cdnjs.cloudflare.com/ajax/libs/gsap/3.12.0/ScrollTrigger.min.js"
integrity="sha512-AY2+JxnBETJ0wcXnLPCcZJIJx0eimyhz3OJ55k2Jx4RtYC+XdIi2VtJQ+tP3BaTst4otlGG1TtPJ9fKrAUnRdQ=="
crossorigin="anonymous"
referrerpolicy="no-referrer"
></script>
</head> <body class="tw-flex tw-min-h-[100vh] tw-flex-col tw-bg-[#fff] tw-font-mono">
<header class="tw-absolute tw-top-0 tw-z-20 tw-flex tw-h-[120px] tw-w-full tw-bg-opacity-0 tw-px-[5%] max-lg:tw-mr-auto max-lg:tw-px-4 lg:tw-justify-around">
<a class="tw-h-[120px] tw-p-[4px] tw-w-[100px] tw-text-2xl tw-font-medium" href="./index.html">
agentic learning <br/> ai lab
</a>
<div class="collapsible-header animated-collapse max-lg:tw-shadow-md"
id="collapsed-header-items">
<div class="tw-flex tw-h-full tw-w-max tw-gap-5 tw-text-base tw-text-black max-lg:tw-mt-[30px] max-lg:tw-flex-col max-lg:tw-place-items-end max-lg:tw-gap-5 lg:tw-mx-auto lg:tw-place-items-center">
<a class="header-links" href="./index.html"> home </a>
<a class="header-links" href="./research.html"> research </a>
<a class="header-links" href="./people.html"> people </a>
<a class="header-links" href="./contact.html"> contact </a>
</div>
<div class="tw-mx-4 tw-flex tw-place-items-center tw-justify-end tw-gap-[20px] tw-text-base">
<div class="tw-flex tw-h-[30px] tw-w-[30px] tw-rounded-full tw-bg-white tw-font-semibold tw-text-black tw-place-items-center ">
<a href="">
<i class="bi bi-search"></i>
</a>
</div>
</div>
</div>
<button
class="bi bi-list tw-absolute tw-right-3 tw-top-3 tw-z-50 tw-text-3xl tw-text-black lg:tw-hidden"
onclick="toggleHeader()"
aria-label="menu"
id="collapse-btn"
></button>
</header>
<section class="tw-relative tw-flex tw-w-full tw-max-w-[100vw] tw-min-w-[350px] tw-flex-col tw-overflow-hidden tw-mt-[150px] tw-px-[5%] max-md:tw-px-4 max-lg:tw-px-4 lg:tw-justify-around xl:tw-justify-around">
<div class="tw-flex tw-w-full tw-place-content-center tw-gap-6 max-xl:tw-place-items-center max-lg:tw-flex-col">
<div class="image-container">
<img src="./assets/images/background/bg1.jpg" alt="app" class="reveal-hero-img tw-z-[1] tw-h-full tw-w-full tw-object-contain"/>
<div class="text-overlay">
<div class="tw-flex tw-flex-wrap tw-text-7xl tw-font-semibold tw-leading-[85px] max-lg:tw-text-4xl max-md:tw-leading-snug" style={position: relative}>
<span class="reveal-hero-text">
agentic learning <br/> ai lab
</span>
</div>
<div class="reveal-hero-text tw-mt-2 tw-max-w-[600px] tw-p-2 tw-text-justify max-lg:min(tw-max-w-full,tw-max-w-[600px])">
Agentic Learning AI Lab is a research lab in New York
University founded in 2022. We innovate learning algorithms that enable future agentic AI to learn and adapt flexibly in the real world.
</div>
</div>
</div>
<!-- <div class="tw-flex tw-flex-col tw-place-content-center"> -->
<!-- </div> -->
</div>
</section>
<section class="tw-mt-5 tw-flex tw-w-full tw-flex-col tw-p-[5%] max-lg:tw-p-4">
<h3 class="reveal-up text-left tw-text-4xl tw-font-medium max-md:tw-text-2xl">
research areas
</h3>
<div class="reveal-up tw-my-4 max-md:tw-h-[3px] max-md:tw-w-[40px] tw-h-[5px] tw-w-[60px] tw-bg-gray-300 ">
</div>
<div class="reveal-up tw-mt-10 tw-space-y-8 tw-gap-10 sm:tw-columns-1 md:tw-columns-3 lg:tw-columns-3 xl:tw-columns-3">
<!-- tw-place-content-center tw-justify-around -->
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300">
<div class="tw-h-[200px] tw-w-full tw-overflow-hidden tw-rounded-md">
<img
src=./assets/images/home/visual_experience.webp
alt="article image"
class="tw-h-full tw-w-full tw-object-cover tw-scale-150"
srcset=""
/>
</div>
<h3 class="tw-text-2xl tw-font-semibold max-md:tw-text-xl">
learning from visual experience
</h3>
<p class="tw-mt-2 tw-text-gray-600">
Embodied AI; Self-supervised learning; Continual learning; Representation learning; Multimodal learning.
</p>
<!-- <span>
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</span> -->
</div>
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300">
<div class="tw-h-[200px] tw-w-full tw-overflow-hidden tw-rounded-md">
<img
src=./assets/images/home/adaptive_foundation_models.webp
alt="article image"
class="tw-h-full tw-w-full tw-object-cover tw-scale-150"
srcset=""
/>
</div>
<h3 class="tw-text-2xl tw-font-semibold max-md:tw-text-xl">
adaptive foundation models
</h3>
<p class="tw-mt-2 tw-text-gray-600">
Personalized AI; Few-shot, continual, meta learning; Memorization and forgetting; Future news forecasting.
</p>
<!-- <span>
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</span> -->
</div>
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300">
<div class="tw-h-[200px] tw-w-full tw-overflow-hidden tw-rounded-md">
<img
src=./assets/images/home/concept_abstraction.webp
alt="article image"
class="tw-h-full tw-w-full tw-object-cover tw-scale-150"
srcset=""
/>
</div>
<h3 class="tw-text-2xl tw-font-semibold max-md:tw-text-xl">
concept learning and abstraction
</h3>
<p class="tw-mt-2 tw-text-gray-600">
Few-shot concept learning; Visual reasoning; Relational abstraction; Hierarchical abstraction and planning.
</p>
<!-- <span>
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</span> -->
</div>
</div>
</section>
<section class="tw-mt-8 tw-flex tw-w-full tw-flex-col tw-px-[5%] max-lg:tw-p-4 tw-overflow-hidden">
<h3 class="reveal-up text-left tw-text-4xl tw-font-medium max-md:tw-text-2xl">
recent works
</h3>
<div class="reveal-up tw-my-4 max-md:tw-h-[3px] max-md:tw-w-[40px] tw-h-[5px] tw-w-[60px] tw-bg-gray-300 ">
</div>
<!-- tw-shadow-lg -->
<div class="tw-mt-8 tw-gap-10 tw-space-y-reverse sm:tw-columns-1 md:tw-columns-2 lg:tw-columns-3 xl:tw-columns-4 tw-items-start">
<a href="https://arxiv.org/abs/2411.08324">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/are_llms_prescient.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">Are LLMs Prescient? A Continuous Evaluation using Daily News as Oracle</h3>
<p class="tw-text-gray-600 tw-mt-4">
Our new benchmark, Daily Oracle, automatically generates question-answer (QA) pairs from daily news, challenging LLMs to predict “future” events based on pre-training data.
</p>
<p class="tw-mt-4">
Publish Date: 2024-11-13
</p>
<a href=https://arxiv.org/abs/2411.08324 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2408.11208">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/poodle.webp
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">PooDLe: Pooled and Dense Self-Supervised Learning from Naturalistic Videos</h3>
<p class="tw-text-gray-600 tw-mt-4">
We propose PooDLe, a self-supervised learning method that combines an invariance-based objective on pooled representations with a dense SSL objective that enforces equivariance to optical flow warping.
</p>
<p class="tw-mt-4">
Publish Date: 2024-08-20
</p>
<a href=https://arxiv.org/abs/2408.11208 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2408.02226">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/procreate.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">ProCreate, Don't Reproduce! Propulsive Energy Diffusion for Creative Generation</h3>
<p class="tw-text-gray-600 tw-mt-4">
ProCreate is a simple and easy-to-implement method to improve sample diversity and creativity of diffusion-based image generative models and to prevent training data reproduction.
</p>
<p class="tw-mt-4">
Publish Date: 2024-08-05
</p>
<a href=https://arxiv.org/abs/2408.02226 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2404.19132">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/osiris.webp
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">Integrating Present and Past in Unsupervised Continual Learning</h3>
<p class="tw-text-gray-600 tw-mt-4">
We formulate Osiris, a unifying framework for unsupervised continual learning (UCL), which disentangles learning objectives that encompass stability, plasticity, and cross-task consolidation.
</p>
<p class="tw-mt-4">
Publish Date: 2024-04-29
</p>
<a href=https://arxiv.org/abs/2404.19132 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2403.15362">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/college.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">CoLLEGe: Concept Embedding Generation for Large Language Models</h3>
<p class="tw-text-gray-600 tw-mt-4">
CoLLEGe is a meta-learning framework capable of generating flexible embeddings for new concepts using a small number of example sentences or definitions.
</p>
<p class="tw-mt-4">
Publish Date: 2024-03-22
</p>
<a href=https://arxiv.org/abs/2403.15362 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2403.09613">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/reawakening.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">Reawakening Knowledge: Anticipatory Recovery from Catastrophic Interference via Structured Training</h3>
<p class="tw-text-gray-600 tw-mt-4">
We discover a curious and remarkable property of LLMs fine-tuned sequentially in this setting: they exhibit anticipatory behavior, recovering from the forgetting on documents before encountering them again.
</p>
<p class="tw-mt-4">
Publish Date: 2024-03-14
</p>
<a href=https://arxiv.org/abs/2403.09613 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2402.00300">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/ssl_childs_perspective.webp
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">Self-Supervised Learning of Video Representations from a Child's Perspective</h3>
<p class="tw-text-gray-600 tw-mt-4">
We train self-supervised video models on longitudinal, egocentric headcam recordings collected from a child over a two year period in their early development.
</p>
<p class="tw-mt-4">
Publish Date: 2024-02-01
</p>
<a href=https://arxiv.org/abs/2402.00300 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2312.12736">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/learning_and_forgetting_llm.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">Learning and Forgetting Unsafe Examples in Large Language Models</h3>
<p class="tw-text-gray-600 tw-mt-4">
We explore the behavior of LLMs finetuned on noisy custom data containing unsafe content and propose a simple filtering algorithm for detecting harmful content based on the phenomenon of selective forgetting.
</p>
<p class="tw-mt-4">
Publish Date: 2023-12-20
</p>
<a href=https://arxiv.org/abs/2312.12736 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
<a href="https://arxiv.org/abs/2312.12736">
<div class="safari-padding-fix">
<div class="reveal-up tw-flex tw-h-fit tw-break-inside-avoid tw-flex-col tw-gap-2 tw-rounded-lg tw-bg-[#f3f3f3b4] tw-p-4 max-lg:tw-w-full max-lg:tw-max-w-[400px] hover:tw-shadow-lg tw-transition-shadow tw-duration-300 tw-overflow-hidden">
<div class="tw-flex tw-place-items-center tw-gap-3">
<div class="tw-h-[300px] tw-w-full tw-overflow-hidden tw-rounded-lg">
<img src=./assets/images/papers/lifelong_memory.png
class="tw-h-full tw-w-full tw-object-cover"
alt="design"/>
</div>
<!-- tw-transition-transform tw-duration-300 tw-transform hover:tw-scale-110 -->
</div>
<div class="tw-flex tw-flex-col tw-gap-2">
<h3 class="tw-text-xl tw-font-medium tw-mt-4">LifelongMemory: Leveraging LLMs for Answering Queries in Long-form Egocentric Videos</h3>
<p class="tw-text-gray-600 tw-mt-4">
LifelongMemory is a new framework for accessing long-form egocentric videographic memory through natural language question answering and retrieval.
</p>
<p class="tw-mt-4">
Publish Date: 2023-12-07
</p>
<a href=https://arxiv.org/abs/2312.12736 class="tw-mt-4">
<span>Learn more</span>
<i class="bi bi-arrow-right"></i>
</a>
</div>
</div>
</div>
</a>
</div>
</section>
<hr class="tw-mt-4" />
<footer class="tw-mt-auto tw-flex tw-p-[5%] tw-min-h-[100px] tw-w-full tw-place-items-center tw-gap-3 tw-text-black
">
<div class="tw-flex tw-gap-6 tw-text-sm">
60 Fifth Ave
<br />
New York, NY
<br />
</div>
<div class="tw-flex tw-gap-6 tw-text-2xl tw-justify-end tw-ml-auto tw-px-2">
<a href="https://github.com/agentic-learning-ai-lab" aria-label="Github">
<i class="bi bi-github"></i>
</a>
<a href="https://x.com/agentic_ai_lab" aria-label="X">
<i class="bi bi-twitter-x"></i>
</a>
</div>
</footer> </body>
<script src="./index.js"></script>
</html>