Ahmad Beirami ✈️ ICLR2024 @abeirami Twitter profile | Pikagi

Pikagi

Ahmad Beirami ✈️ ICLR2024

@abeirami

4,448

Followers

2,485

Following

74

Media

2,017

Statuses

Building safe, helpful, and scalable generative AI @Google | ex-{ @AIatMeta , @EA , @MIT , @Harvard , @DukeU } | @GeorgiaTech PhD | زن زندگی آزادی | opinions my own

{NYC, BOS, YYZ}

https://t.co/sd6RyVPWmk

Joined December 2018

Don't wanna be here? Send us removal request.

Pinned Tweet

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

Giving a new talk tomorrow at @USC 𝑳𝒂𝒏𝒈𝒖𝒂𝒈𝒆 𝑴𝒐𝒅𝒆𝒍 𝑨𝒍𝒊𝒈𝒏𝒎𝒆𝒏𝒕: 𝑻𝒉𝒆𝒐𝒓𝒚 & 𝑷𝒓𝒂𝒄𝒕𝒊𝒄𝒆 hosted by @mahdisoltanol While the talk is mostly an 𝑎𝑙𝑖𝑔𝑛𝑚𝑒𝑛𝑡 tutorial, I will also touch upon some of our recent work👇

Tweet media one

5

14

144

Last Seen Profiles

@carnecrudaradio

@NbhdFoodHub

@roccoseleven

@miladymaker

@ikgallas

@halukawa

@nottpadu

@nan2000web

@_DJDON

@AHISKAMURAD

@Tsuki_Seiba

@discordfora

@EnfieldCounty

@kayo911harusaki

@Nemesis_1133

@devilfruitbat

@mfhstheatre

@ChaosCouncilOOC

@billcode

@jandakembangstw

@kagoya_kagu

@AnneZMurphy

@nickfoard

@raon_leee

@fightwithmemes

@WillBlevjr

@PAMPKlN

@daabrattt_z

@Lewis_AI_Art

@ThugProdux

@taizn25

@AlexBonnemaison

@DuneSeaGarrison

@onspire3

@5533alonso63164

@GreatThiefAcey

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

The question that a reviewer should ask themselves is: Does this paper take a gradient step in the right direction? Is the community better off with this paper published? If the answer is yes, then the recommendation should be to accept.

8

31

335

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

[ #eacl2024 paper] TL;DR We introduce 𝗴𝗿𝗮𝗱𝗶𝗲𝗻𝘁-𝗯𝗮𝘀𝗲𝗱 𝗿𝗲𝗱 𝘁𝗲𝗮𝗺𝗶𝗻𝗴 (𝗚𝗕𝗥𝗧), an effective method for triggering language models to produce unsafe responses, even when the LM is finetuned to be safe through 𝑎𝑙𝑖𝑔𝑛𝑚𝑒𝑛𝑡.

Tweet media one

3

35

294

@abeirami

Ahmad Beirami ✈️ ICLR2024

6 months

⚠️ #Internship2024 Alert⚠️ Are you a 𝐏𝐡𝐃 𝐬𝐭𝐮𝐝𝐞𝐧𝐭 excited about #ResponsibleAI in foundation models? Do you have experience training/evaluating them? Team members: @PreethiLahoti @infoxiao @ananthbshankar @sroy_subhrajit @aradsinha @YaoQinUCSD (RADML at @GoogleResearch )

11

60

259

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

Please RT Are you a *PhD* student conducting research on generative models? Are you excited about #ResponsibleAI aspects of generation? We are looking to host a student researcher/ #internship in 2023. Please get in touch via *email*. P.S. I'll be at #NeurIPS2022 and #emnlp2022

3

80

207

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 month

Have you been perplexed by the surprising performance of 𝗯𝗲𝘀𝘁-𝗼𝗳-𝗻 in alignment compared to SOTA method (𝗣𝗣𝗢/𝗗𝗣𝗢/𝗜𝗣𝗢)? We have theory that explains this phenomenon.

Tweet media one

3

17

186

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

Hey @emnlpmeeting : is this a joke or a metareview because I classified as joke initially but then changed to metareview. #emnlp2023

Tweet media one

8

3

161

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

Both @icmlconf and @NeurIPSConf held in the US in 2022-23! The US is one of the most visa unfriendly states (appointment wait times 6+ months, processing another 6+ months), this is significantly hurting diversity & inclusion. We should strive to do better! #ICML2023 #NeurIPS2023

8

23

161

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

We publish a ton of our core work on language models. Here are a few samples just from our team released in the last few weeks: - - - - How's that "giving almost nothing back"?

Tweet card media

Controlled Decoding from Language Models

KL-regularized reinforcement learning (RL) is a popular alignment framework to control the language model responses towards high reward outcomes. We propose a modular solver for this RL objective,...

@soldni

Luca Soldaini 🎀 @ ICLR 2024

5 months

I can't help being a bit sad in thinking how much the Gemini release is taking from the academic and open LMs communities, while giving almost nothing back I can't help but wish our relationship with commercial players would be more of a two way channel

4

13

112

2

11

157

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

Have you been compiling reward-KL tradeoffs to compare different alignment methods? Have you been using 𝐛𝐞𝐬𝐭-𝐨𝐟-𝐧 as a baseline? Have you wondered about the analytical formula that claims this formula? 𝐾𝐿 (𝑏𝑒𝑠𝑡-𝑜𝑓-𝑛 || 𝑏𝑎𝑠𝑒) = 𝑙𝑜𝑔(𝑛) - (𝑛-1)/𝑛

Tweet media one

4

22

152

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

I left Meta AI a few weeks ago. I am filled with gratitude for the opportunity to “think big” and propose Project CAIRaoke to redefine the future of #ConversationalAI ! I am also humbled to have worked with so many amazing people along the way!

1

0

148

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

[PSA] Please use \𝗰𝗶𝘁𝗲𝘁{} and \𝗰𝗶𝘁𝗲𝗽{} correctly or the paper will be hard to read. \𝗰𝗶𝘁𝗲𝘁{}: Author name is intended to be part of the sentence: X et al. (2024) \𝗰𝗶𝘁𝗲𝗽{}: Citation appears in parentheses and not read in the sentence: (X et al., 2024)

3

6

135

@abeirami

Ahmad Beirami ✈️ ICLR2024

8 months

The list of #NeurIPS2023 accepted papers are available here:

0

28

129

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

Off policy algorithms (even the simplest ones) are indeed useful for RLHF and can train a value function on massive amounts of data and 𝗼𝘂𝘁𝗽𝗲𝗿𝗳𝗼𝗿𝗺 𝗜𝗣𝗢/𝗗𝗣𝗢/𝗣𝗣𝗢.

Tweet media one

@natolambert

Nathan Lambert

3 months

A brief summary on what REINFORCE is in terms of RLHF and history of RL. The algorithm known as REINFORCE is really just the vanilla policy gradient approach. The name comes from Williams 1992, "Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement

3

17

123

3

17

129

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

[2022 #Internships at the @facebookai Conversational AI Research (CAIR) team] The CAIR team is seeking to recruit multiple PhD student interns to work with us ( @SatwikKottur , @Chinnadhurai , @abeirami , and others) on different aspects of #ConversationalAI and #NLProc . 1/5

1

37

125

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Excited to share that our paper titled “Winning is not everything: enhancing game development with intelligent agents,” @IEEETxnOnGames , June 2020, has been selected to receive the 2023 Outstanding Paper Award by @ieeecis Awards Committee.

3

7

126

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

A periodic reminder to reviewers: If you ask authors for more experiments, then you need to communicate a clear hypothesis you're trying to verify with those (e.g., effectiveness on imbalanced data, generalization beyond a certain modality, scalability, etc). Otherwise don't!

2

8

123

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

As for next steps, I am excited to share that I have joined #GoogleResearch to lead efforts around robust and fair development of core machine learning techniques. I am also moving to New York City, and excited to be back to the east coast!

25

0

114

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

"Oh, you're with G Research. Are you a quant?" #NeurIPS2023

2

0

103

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

If you reviewed for #NeurIPS2023 , please take a moment to read the authors' rebuttal, other reviews, and briefly respond to the authors (even if the response is that I need more time to process your rebuttal). The authors put a lot of time into this!

2

11

99

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

I've seen variants of this terrible idea proposed before. Not only it's plain wrong and unethical to not provide recommendation letters to top students, it also doesn't work! Students will find other ways to leave, and then they will not even look back!

2

3

97

@abeirami

Ahmad Beirami ✈️ ICLR2024

6 months

When I first started reviewing ML papers, I was fighting to reject bad papers. These days I find myself fighting to accept good papers. The change in perspective has also made my reviews more constructive even in cases where the recommendation has to be reject.

0

1

87

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

An O(n) algorithm (including random sketching) to check out the posters is no longer feasible! Does anyone know a sublinear algorithm for retrieval/learning from an unsorted knowledge base? #NeurIPS2023

Tweet media one

13

6

85

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

Here is what I "speculate" might have happened at #emnlp2023 : After decisions finalized by AC/SACs, the @emnlpmeeting PCs likely felt there were "too many" accepted papers (in findings), and scrambled to reject 100s of papers to lower the acceptance rate to keep "prestige".

4

10

84

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 years

The Conversational AI Research team at @facebookai is seeking to recruit multiple PhD student interns who are doing research in the conversational AI domain! Let's meet and chat tomorrow at the Conversational AI workshop! #neurips2019

0

22

81

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

It was an honor to receive the @IEEETxnOnGames Outstanding Paper Award for our 2020 paper titled "Winning Is Not Everything" at @ieee_cog 2023, from @togelius tonight!

Tweet media one

Tweet media two

10

1

82

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

If you are at #EACL2024 , on Friday 4pm Malta time, I will give a (virtual) talk on 𝗖𝗼𝗻𝘁𝗿𝗼𝗹𝗹𝗲𝗱 𝗗𝗲𝗰𝗼𝗱𝗶𝗻𝗴 𝗳𝗿𝗼𝗺 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹𝘀 at the 1𝘴𝘵 𝘗𝘦𝘳𝘴𝘰𝘯𝘢𝘭𝘪𝘻𝘢𝘵𝘪𝘰𝘯 𝘰𝘧 𝘎𝘦𝘯𝘦𝘳𝘢𝘵𝘪𝘷𝘦 𝘈𝘐 (𝘗𝘌𝘙𝘚𝘖𝘕𝘈𝘓𝘐𝘡𝘌) 𝘸𝘰𝘳𝘬𝘴𝘩𝘰𝘱.

Tweet media one

1

8

81

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

Dear authors: It is NOT okay to bully the reviewers who have engaged with you in a discussion and don't agree that your paper should be published! Thanks! #NeurIPS2021

4

1

80

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 years

If you are a *PhD student* interested in Conversational AI Research, and would like to do an internship at @facebookai in Fall 2020, please reach out via email! #internship #ConversationalAI

2

37

78

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

There are two types of empirical researchers: (1) Devise a marketable narrative. Cook up experiments to support the narrative. Write a paper! (2) Devise a hypothesis. Design experiments to verify/disprove the hypothesis. Write the narrative based on results or drop the project.

7

7

77

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

Headed to Vancouver to #AAAI2024 , giving 2 talks & serving on a panel! Hope to see you there! 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹 𝗔𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁: 𝗧𝗵𝗲𝗼𝗿𝘆 & 𝗣𝗿𝗮𝗰𝘁𝗶𝗰𝗲 While the talk is mostly an 𝑎𝑙𝑖𝑔𝑛𝑚𝑒𝑛𝑡 tutorial, I will also touch upon some of our recent work👇

1

5

75

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

It's completely acceptable to write a purely empirical paper as long as the empirical results substantiate the claims! It's not ok to inject nonsensical/wrong math "results" into an empirical paper to give the illusion of having supporting theory! This happens way too often!

5

5

73

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

[Call for papers] #NeurIPS2023 R0-FoMo Workshop Robustness of Zero/Few-shot Learning in Foundation Models We solicit novel contributions that relate broadly to zero/few-shot learning in foundation models, with both empirical and theoretical nature. Please R/T

1

34

72

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Identify and remove the bad ACs! There is no shortage of qualified people who can AC! Let the authors and reviewers identify and report these bad ACs, and don’t invite them back!

2

4

70

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

Going to San Diego to attend the ITA workshop (), which is my favorite conference/workshop of every year! I will speak about our recent findings on 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐦𝐨𝐝𝐞𝐥 𝐚𝐥𝐢𝐠𝐧𝐦𝐞𝐧𝐭 on Friday!

Tweet media one

0

2

69

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

"Overwhelming" was the theme of #NeurIPS2023

1

1

68

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

Are you excited about your new 𝗮𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁 method with a higher win-rate against the base model? Are you ready to publish? WAIT! Unless your judge was perfect (which defeats the purpose), comparing win-rate without regard to the divergence from the base policy is flawed!

3

5

68

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

I will be at #NeurIPS2023 and look forward to meeting old and new friends! Would love to connect and chat about building safe, helpful, and scalable generative language models!

Tweet media one

0

4

67

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 month

If you reviewed for #ICML2024 , please take a moment to reflect on the author rebuttal, other reviews, and ask any further clarifying questions that authors can help answer. The authors' ability to engage in discussion ends in less than 24hrs.

6

7

65

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 years

[2021 #Internships at the @facebookai Conversational AI Research (CAIR) team] The CAIR team at @facebookai is seeking to recruit multiple PhD student interns to work with us ( @SatwikKottur , @Chinnadhurai , @abeirami , and others) on different aspects of #ConversationalAI .

3

11

64

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

Markov's inequality is very simple yet extremely powerful! Let X be a non-negative RV, then P[X ≥ a] ≤ E[X]/a.

@miniapeur

Mathieu Alain

4 months

What is your favourite mathematical result? Doesn’t have to be sophisticated or anything.

384

25

508

2

1

61

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

Using only a restaurant, tell us where you went to college

Tweet media one

@ameya_pa

Ameya Velingker | अमेय वेलिंगकर

5 months

Using only a restaurant, tell us where you went to college

Tweet media one

1

0

9

1

2

60

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

TL;DR @TmlrOrg is an excellent venue to publish and read grounded ML work, both as a reviewer and an author! See the papers here: 1/n

2

5

60

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

And the #icml2023 test-of-time award goes to "learning fair representations" which has arguably had a huge impact on ML. Congratulations to the authors!!

Tweet media one

1

2

59

@abeirami

Ahmad Beirami ✈️ ICLR2024

6 months

Check out our #EMNLP2023 paper on evaluating and improving diversity of representation in LLMs! P.S. We're looking to host interns to work on different RAI aspects of using LLMs. Stay tuned on more details!

@PreethiLahoti

Preethi Lahoti

6 months

Large language models (LLMs) have come a long way and solve many tasks. BUT diversity and inclusion in LLM generations is still an open challenge. 📢 New @emnlpmeeting paper on quantifying and improving people/culture diversity in LLMs: #EMNLP2023 1/n

Tweet media one

4

21

126

2

2

59

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

Check out Somnath's #ICLR2024 spotlight paper on how to improve fairness in online learning. TL;DR Using an oblique decision tree and imposing fairness constraints on each node helps achieve good fairness guarantees empirically/theoretically! See you in Vienna to discuss more!

@SomnathBrc

Somnath Basu Roy Chowdhury

2 months

How can we improve fairness when data arrives in an online stream? We present 𝐴𝑟𝑎𝑛𝑦𝑎𝑛𝑖, an effective approach towards achieving group fairness in online environments. 𝐒𝐩𝐨𝐭𝐥𝐢𝐠𝐡𝐭 paper at #ICLR2024 . Work from my last internship at @GoogleAI .

Tweet media one

2

7

41

3

6

58

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

Happy Women in Mathematics day everyone! May 12 is the birthday of Maryam Mirzakhani, a mathematician who was awarded the Fields medal ( highest honor in math) for her contributions to geometry and dynamical systems. [Two of my fav mathematicians: Maryam and Ingrid Daubechies]

Tweet media one

1

4

57

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 month

#icml2024 workshops list is up:

1

11

56

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Please distinguish the citizens of a country from its government. Targeting the citizens of a country for the aggression of its government is not the solution!

1

1

57

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

@JonathanBerant I'm really sorry for how you're feeling, Jonathan! I thought regardless of how one might think of Israel's government, it should be a very low bar to condemn Hamas' terrorist attack against civilians. I still think/hope that's the case for majority of the community.

1

0

57

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 month

Robustness methods 1) augment data with natural/synthetic perturbations and a consistency loss 2) reweight samples to improve generalization (like DRO) We do it differently! We show significant robustness with a simple tweak of the first layer and loss motivated by comms theory.

@BhagyashreePu13

Bhagyashree Puranik

@BhagyashreePu13

1 month

📢📢📢 Late post, but here we go...! I am thrilled to announce that our work on 𝙚𝒏𝙝𝒂𝙣𝒄𝙞𝒏𝙜 𝙤𝒖𝙩-𝙤𝒇-𝒅𝙞𝒔𝙩𝒓𝙞𝒃𝙪𝒕𝙞𝒐𝙣 𝙧𝒐𝙗𝒖𝙨𝒕𝙣𝒆𝙨𝒔 of deep neural networks has been accepted to 𝘼𝑰𝙎𝑻𝘼𝑻𝙎 2024!

Tweet media one

2

5

50

2

7

55

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

I will be at @aclmeeting next week, and look forward to connecting with old and new friends! We also have a student researcher position open in the fall (and potentially beyond) to study responsibility aspects (e.g., robustness, safety, fairness) of generative (language) models.

1

7

55

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

It's a good time to remind ourselves that a paper need not be revolutionary to be acceptable. It just needs to take a positive gradient step!

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

The question that a reviewer should ask themselves is: Does this paper take a gradient step in the right direction? Is the community better off with this paper published? If the answer is yes, then the recommendation should be to accept.

8

31

335

0

5

55

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

A good research reputation is built up one paper at a time, which takes time and effort! Ruining a research reputation can be done much faster!

0

1

54

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

Wondering what's going on in Iran? A glimpse of past 24hrs: - A protester shot in the head for dancing w/o hijab - A 3-year old detained - Water shut to Kurdish city of Sanandaj to deter protests World must stop engaging with this murderous regime NOW! #مهسا_امینی #Mahsa_Amini

2

6

53

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

#mathtwitter I wonder if this tighter Chernoff bound for *non-negative* RVs is known (or is trivial from some other known bound that I am unaware of). cc: @lugosi_gabor , @ccanonne_ , @thomasahle , @mraginsky , @CindyRush Proof is similar to Theorem 14 of

Tweet media one

8

6

52

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

Just boarded a flight with more posters on board than carry ons :-D

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

I will be at #NeurIPS2023 and look forward to meeting old and new friends! Would love to connect and chat about building safe, helpful, and scalable generative language models!

Tweet media one

0

4

67

0

0

51

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 years

[thread with @meisamrr and @pooyaan ] The @NeurIPSConf Experiment We do a simple analysis that reveals that “if the #NeurIPS2020 review process is redone with new AC/Reviewer assignments, the majority of the papers that are accepted today will get rejected.” 1/7

2

11

50

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

As #NAACL2022 came to an end, the two most interesting personal takeaways: 1. I met for the first time so many people I had been working with over the past 2 years! 2. I learnt there are so many disciplines in #NLProc that I embarrassingly didn’t even know the problem statement

0

2

48

@abeirami

Ahmad Beirami ✈️ ICLR2024

18 days

Check out Zhaofeng's work from his internship with us! TL;DR A reward model trained on language S preference data could be used to align a language T LLM. This sometimes works even better than using a reward model trained on language T preference data.

@zhaofeng_wu

Zhaofeng Wu

19 days

Want to train an aligned LM in a new language 🌏 but don’t have preference data for training the reward model (RM)? 💡 Just use a RM for another language: it often works well, sometimes even BETTER than if you had a RM in your target language! 🤯

Tweet media one

2

31

131

0

5

48

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 months

TL;DR We speed up speculative decoding (for free) using optimal verification algorithm for acceptance/rejection at block level! This is complementary to numerous works that propose better drafting algorithms as all of them use the same token-wise verification (Leviathan et al).

@SZiteng

Ziteng Sun

2 months

𝗦𝗽𝗲𝗰𝘂𝗹𝗮𝘁𝗶𝘃𝗲 𝗱𝗲𝗰𝗼𝗱𝗶𝗻𝗴 is popular for speeding up LLM inference. Many have proposed better drafters, BUT they 𝑎𝑙𝑙 use a token-by-token verification procedure (𝘛𝘰𝘬𝘦𝘯𝘝𝘦𝘳𝘪𝘧𝘺). 𝘐𝘴 𝘛𝘰𝘬𝘦𝘯𝘝𝘦𝘳𝘪𝘧𝘺 𝘰𝘱𝘵𝘪𝘮𝘢𝘭? Somewhat surprisingly, 𝗻𝗼.

Tweet media one

1

7

38

1

2

47

@abeirami

Ahmad Beirami ✈️ ICLR2024

11 days

I'll be at #AISTATS2024 later this week! With Madhow, we will co-present @BhagyashreePu13 's poster on TEXP to improve robustness with a tweak to the first layer of the network. Looking forward to meeting old and new friends!

@BhagyashreePu13

Bhagyashree Puranik

@BhagyashreePu13

1 month

📢📢📢 Late post, but here we go...! I am thrilled to announce that our work on 𝙚𝒏𝙝𝒂𝙣𝒄𝙞𝒏𝙜 𝙤𝒖𝙩-𝙤𝒇-𝒅𝙞𝒔𝙩𝒓𝙞𝒃𝙪𝒕𝙞𝒐𝙣 𝙧𝒐𝙗𝒖𝙨𝒕𝙣𝒆𝙨𝒔 of deep neural networks has been accepted to 𝘼𝑰𝙎𝑻𝘼𝑻𝙎 2024!

Tweet media one

2

5

50

1

4

48

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

I have been involved in the review process of 4 @TmlrOrg papers, and guess how many got in? . . . . All 4 have been deservedly accepted, and it feels good!

0

1

46

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

Everyone pointed this breaching the professional agreement as a reviewer. That aside, there are actual people behind the paper who put in actual work. Reviewer's job is to help them improve the paper and determine whether it's publishable. It's *never* to ridicule the authors!!

@victorckumar

Victor Kumar

5 months

Is it unprofessional to live tweet the laughably awful paper I'm reviewing?

58

1

114

0

0

46

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

We are organizing an upcoming Special Issue in IEEE Internet of Things Magazine: An end-to-end Machine Learning Perspective on Industrial IoT #industry40 #smartmanufacturing #MachineLearning

Tweet card media

Ahmad Beirami on LinkedIn: #industry40 #smartmanufacturing #machinelearning

[Call for Papers, IEEE IoT Magazine, Special Issue] An end-to-end machine learning perspective on Industrial…

www.linkedin.com

0

22

44

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

While this is outrageous, please don’t imply this should be normal in any other country to make your point. This is outrageous anywhere in the world.

@GaviBegtrup

Gavi Begtrup

2 years

A 10-year-old girl is raped. The State forces her to remain pregnant and tells her to consider it an “opportunity.” This isn’t Iran. This isn’t Gilead. This isn’t hypothetical. This happened today in Ohio.

7K

88K

367K

0

2

42

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Is there a rigorous training for Reviewers/ACs/SACs in our community? Apparently not! This Metareview is completely unprofessional and unacceptable! Why didn't the SAC flag it?

@thegautamkamath

Gautam Kamath

@thegautamkamath

2 years

I was searching through the opted-in NeurIPS rejections and found this rather rude metareview. The prior sentence is also a bit harsh, but the last sentence is totally unnecessary. Please try to be more considerate than this when reviewing.

Tweet media one

26

31

381

1

3

43

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

After reviewing for several #ML confs I had a personal reflection. I realized my bar for accept is much higher than accept rate. That led me to think harder to articulate reasons to accept, despite shortcomings! A paper need not be perfect, it should be a step in right direction.

0

3

43

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

Thrilled to announce @iclr_conf Workshop on #ResponsibleAI with Speakers & Panelists: @red_abebe , @jquinonero , @FinaleDoshi , @mrtz , @_beenkim , @hima_lakkaraju , @EricRicePhD , Cynthia Rudin, @DorsaSadigh , Richard Zemel. #ICLR2021

Tweet card media

Joining the Workshop on May 7 https://iclr.cc/virtual/2021/workshop/2132 Summary Artificial Intelligence and Machine Learning are increasingly employed by industry and government alike to make or...

sites.google.com

2

10

41

@abeirami

Ahmad Beirami ✈️ ICLR2024

8 months

One third of the rejected #NeurIPS2023 papers in my stack would likely have had a chance with some additional results, revisions, reframing, and discussion, to solidify their pitch. That is where @TmlrOrg wins!

4

2

41

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

I am excited to be attending #ICML2023 this week, and look forward to catching up with friends and making new ones. I would also love to discuss responsible/reliable/efficient foundational models. Please email me if you would like to meet 1:1 A 🧵on events I am attending Wed-Sat

1

2

42

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

Here is an idea! If you are working on #ResponsibleAI and/or LLMs and have a paper in #NeurIPS2023 poster sessions, please reply with your poster number and time slot so that I (and others interested in similar topics) don't miss it. Thank you!

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

An O(n) algorithm (including random sketching) to check out the posters is no longer feasible! Does anyone know a sublinear algorithm for retrieval/learning from an unsorted knowledge base? #NeurIPS2023

Tweet media one

13

6

85

5

4

41

@abeirami

Ahmad Beirami ✈️ ICLR2024

11 months

This paper, which has been in the making with @litian0331 since the start of the pandemic is finally out!!

@JmlrOrg

Journal of Machine Learning Research

11 months

'On Tilted Losses in Machine Learning: Theory and Applications', by Tian Li, Ahmad Beirami, Maziar Sanjabi, Virginia Smith. #tilting #minimization #risk

0

4

12

1

4

40

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

Have LLMs reached the reasoning capabilities of #emnlp2023 meta reviewers? An anecdotal study

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

Hey @emnlpmeeting : is this a joke or a metareview because I classified as joke initially but then changed to metareview. #emnlp2023

Tweet media one

8

3

161

1

2

40

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

As a #NeurIPS2023 author, you effectively have unlimited space to rebut the reviewers' response (by posting multiple 6k char comments). Do NOT do that!! If you want to win your argument, please strive to prepare a point-by-point response that is clear, convincing, and concise!

1

1

40

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Iranian people are at war with a regime that has a track record of disregard for human life. It’s heartbreaking to see new lives taken in the streets of Iran by this murderous regime. We remain hopeful for the people’s will to prevail!! Be their voice! #MahsaaAmini #مهسا_امینی

0

2

40

@abeirami

Ahmad Beirami ✈️ ICLR2024

6 months

Congratulations to all top reviewers of #neurips2023 ! The quality of the program hinges upon the volunteer work of individuals who go above and beyond to fairly assess the claims and quality of the submitted papers!

0

2

38

@abeirami

Ahmad Beirami ✈️ ICLR2024

11 months

The LM reasoning literature (e.g. CoT) is predominately focused on mathematical tasks with well defined responses. What if we want to reason through more nuanced and complex topics with no clear right or wrong answer? ***Thought experiments promoting!***

@infoxiao

Xiao Ma

11 months

Introducing 🤔 Thought Experiments prompting! ⛓️Chain-of-Thought is great but real world is messier than that, especially when answering challenging moral questions. We show self-posing and answering counterfactual questions allow LLM to improve. 1/n

3

76

344

1

7

37

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Happy women in mathematics day everyone!!

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

Happy Women in Mathematics day everyone! May 12 is the birthday of Maryam Mirzakhani, a mathematician who was awarded the Fields medal ( highest honor in math) for her contributions to geometry and dynamical systems. [Two of my fav mathematicians: Maryam and Ingrid Daubechies]

Tweet media one

1

4

57

0

2

38

@abeirami

Ahmad Beirami ✈️ ICLR2024

7 months

Excited to be talking about some of our recent efforts on controlled decoding from language models on Friday!

Oct 25-27 ML Workshop

Yale Institute for Foundations of Data Science at Kline Tower

yalefds.swoogo.com

3

5

37

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

A prominent scholar has been let gone from Sharif University for expression of views opposing the government. Shameful!

@mmahdian

mmahdian

9 months

استاد شریف vs استاد بی‌شرف

Tweet media one

4

33

415

0

2

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

Stochastic gradient descent ascent in action; the generator eventually fools the discriminator.

2

1

37

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

So many criticisms of #NeurIPS reviewer not knowing X or Y. This is mostly on the AC for not making sure that the assigned reviewers have the expertise to evaluate the work, and the SAC for not making sure that the AC has the expertise to identify such reviewers!

6

0

37

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

FERMI is the first stochastic method for group fairness with guaranteed convergence, with batch size as small as one sample! More on the theoretical/empirical results, the paper, and the code package in the thread by @BaharloueiSina

@BaharloueiSina

Sina Baharlouei

@BaharloueiSina

1 year

📢 🚨 Publication Alert🚨📢 In our recently published paper at @TmlrOrg with my fantastic collaborators Andrew Lowy, Rakesh Pavan, @meisamrr , and @abeirami , we propose the first provably "convergent" stochastic algorithm to achieve group fairness in large-scale machine learning.

Tweet media one

2

4

42

2

3

37

@abeirami

Ahmad Beirami ✈️ ICLR2024

8 months

[Unpopular opinion] 1. Do I think *CL anonymity should be abolished? Absolutely. I think it hurts the very people it claims to help; any data showing otherwise? 2. Do I think the paper should've been desk rejected? Absolutely. A rule is a rule and should be enforced uniformly

@nsaphra

Naomi Saphra

8 months

Just got a desk reject, post-rebuttals, for a paper being submitted to arxiv <30 min late for the anonymity deadline. I talk about how the ACL embargo policy hurts junior researchers and makes ACL venues less desirable for NLP work. I don’t talk about the pointless NOISE it adds.

28

47

405

6

1

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

Hello world: if a is an upper bound on x (x<a), and b is an upper bound on y (y<b), then a<b does NOT imply x<y

1

0

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 months

As reviewing committee members, we should strive to 𝐚𝐜𝐜𝐞𝐩𝐭 𝒑𝒐𝒕𝒆𝒏𝒕𝒊𝒂𝒍𝒍𝒚 𝐠𝐨𝐨𝐝 𝐩𝐚𝐩𝐞𝐫𝐬 rather than trying to reject potentially bad ones!!

@OriolVinyalsML

Oriol Vinyals

@OriolVinyalsML

3 months

Fun game. Clocking 17973 citations: "Distilling the knowledge in a neural network" @geoffreyhinton , @OriolVinyalsML , @JeffDean Reviewer 38 (NeurIPS 2014): "This work is incremental and unlikely to have much impact even though it may be technically correct and well executed."

5

47

481

3

1

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

Yao Xie and Vincent Tan are co-organizing a Social at #ICML2023 : Information Theory at ICML, Wed 26 Jul 8:45 p.m. PDT — 10:30 p.m. PDT, Ballroom C.

3

4

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

Very impressive results!! But can this "open the door to superhuman feedback"? Clearly not! And why not? Data processing inequality

@jaseweston

Jason Weston

4 months

🚨New paper!🚨 Self-Rewarding LMs - LM itself provides its own rewards on own generations via LLM-as-a-Judge during Iterative DPO - Reward modeling ability improves during training rather than staying fixed ...opens the door to superhuman feedback? 🧵(1/5)

Tweet media one

5

223

1K

3

0

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

4 months

When I find an error in the theoretical derivations which is not unusual, guess how many other reviewers catch them? The answer has been 0 always.

@shoyer

Stephan Hoyer

4 months

I love when ML researchers get excited about science, but seriously the reviewing process for scientific applications at ML conferences (e.g., ICLR) is entirely broken. Papers with glaring errors are sailing through, without a single review from somebody with domain expertise.

20

26

352

1

0

34

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

Next week, I’ll be at 1. AAAI, co-organizing @R2HCAI workshop on Mon w/ @ghandeharioun @Ali_Etemad1 @NinarehMehrabi @pritam94 @luyang1125 2. ITA, co-organizing 4 sessions on ML fairness, robustness, and differential privacy Thu & Fri w/ @th33rtha Look forward to connecting!

1

4

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

If you're at #NeurIPS2023 , @nagpalchirag will be presenting this work at the Distribution Shift workshop today!

@_akhaliq

AK

5 months

Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking paper page: Reward models play a key role in aligning language model applications towards human preferences. However, this setup creates an incentive for the language

Tweet media one

1

28

68

1

8

35

@abeirami

Ahmad Beirami ✈️ ICLR2024

10 months

As #acl2023 came to an end, my main takeaway is that this community is extremely resourceful and very welcoming!! I've made many new friends and I'm glad that I've chosen this community as my second home (after information theory) to learn and grow!

0

0

34

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

7 papers in 21 days (15 working days) -> 1 review every 2 days (16 working hours) -> 1.5-2.5hr per review (assuming spending 10-15% of one’s time on #NeurIPS2021 reviews). Why do we expect people to spend more time on our @NeurIPSConf submissions?

2

1

34

@abeirami

Ahmad Beirami ✈️ ICLR2024

3 years

If you are at @icmlconf , please come attend the ITR3 workshop on Saturday July 24! The goal of the workshop is to explore information-theoretic methods for rigorous, responsible, and reliable machine learning.

@ITR3_workshop

ITR3 Workshop @ ICML21

3 years

Workshop on Information-Theoretic Methods for Rigorous, Responsible, and Reliable Machine Learning (ITR3) will take place @icmlconf this year. Submissions due May 24th. Details: @abeirami @FlavioCalmon @BerivanISIK @hey1jeong @matthewnokleby @CindyRush

Tweet media one

1

27

87

1

6

34

@abeirami

Ahmad Beirami ✈️ ICLR2024

1 year

I’ll be at #NeurIPS2022 this week and #emnlp2022 next week! Looking forward to meeting old and new friends!

0

0

32

@abeirami

Ahmad Beirami ✈️ ICLR2024

9 months

@denny_zhou What great way to exclude minorities. I know a lot of people who cannot afford that!

1

0

32

@abeirami

Ahmad Beirami ✈️ ICLR2024

2 years

I will be at #NAACL2022 Monday-Friday! This is my first in-person #NLProc conference, and first major conference in more than 2 years! Looking forward to meeting y'all!

0

0

32

@abeirami

Ahmad Beirami ✈️ ICLR2024

5 months

In industry, we mostly care about the candidate's vision. They need their published work to support the artifacts they describe in the vision. Many papers that don't add up to a unified vision is certainly not a good sign.

@victorveitch

Victor Veitch

5 months

The gap between perceived and actual incentives is pretty large. Number of papers or citations weigh pretty lightly in hiring discussions, in my experience. (What does count are letters, which ideally say "Candidate solved important problem x, which is impressive because y")

1

0

24

1

0

31

@abeirami

Ahmad Beirami ✈️ ICLR2024

8 months

Check out Pritam's paper on uncovering how video SSL methods perform in OOD. see you in #NeurIPS2023

@pritam94

Pritam Sarkar

8 months

🎉 Excited to announce that our paper has been ACCEPTED as a Spotlight Paper @NeurIPSConf ! Interested to know how video self-supervised learning (VSSL) methods behave under different types of real-world distribution shifts (e.g., in OOD)? w/ @abeirami @Ali_Etemad1 #NeurIPS2023

Tweet media one

3

3

47

0

0

32