Griffon
@ryancallihan.bsky.social
Here for the NLP, ML and AI bants. And as much as I try, I do end up getting political.
Reposted by Griffon
All borders are arbitrary and created by humans
February 3, 2025 at 10:48 AM
All borders are arbitrary and created by humans
Drowning in Documents: Consequences of Scaling Reranker Inference
This paper conducts a simple test of the effectiveness of rerankers on large amounts of documents. It's really important to think about if you are using RAG a lot.
This paper conducts a simple test of the effectiveness of rerankers on large amounts of documents. It's really important to think about if you are using RAG a lot.
December 9, 2024 at 10:30 AM
Drowning in Documents: Consequences of Scaling Reranker Inference
This paper conducts a simple test of the effectiveness of rerankers on large amounts of documents. It's really important to think about if you are using RAG a lot.
This paper conducts a simple test of the effectiveness of rerankers on large amounts of documents. It's really important to think about if you are using RAG a lot.
Reposted by Griffon
I hope I am not late to the party (was away post-quals chilling) but here are some thoughts on why this is bad IMO:
First, a disclaimer that I am writing this as an African who is a speaker of multiple African languages, NLP researcher of African languages, and HCI researcher focusing broadly on..
First, a disclaimer that I am writing this as an African who is a speaker of multiple African languages, NLP researcher of African languages, and HCI researcher focusing broadly on..
this is a green flag for openai & meta to formally be arbitrators of our languages & mass exploit the population (& researcher that've poured their souls into low resource languages),all to throw unreliable AI that has so far proven to result in more harm than benefit
www.reuters.com/technology/a...
www.reuters.com/technology/a...
Orange enlists Meta and OpenAI to develop AI language models in Africa
Orange will enlist OpenAI and Meta to fine-tune AI large language models (LLMs) to translate regional African languages for the French telecoms operator, it said on Tuesday.
www.reuters.com
December 2, 2024 at 11:43 PM
I hope I am not late to the party (was away post-quals chilling) but here are some thoughts on why this is bad IMO:
First, a disclaimer that I am writing this as an African who is a speaker of multiple African languages, NLP researcher of African languages, and HCI researcher focusing broadly on..
First, a disclaimer that I am writing this as an African who is a speaker of multiple African languages, NLP researcher of African languages, and HCI researcher focusing broadly on..
Reposted by Griffon
Anyone saying The Left must stay on Twitter to save democracy doesn’t understand how Twitter affects our psychology. Twitter makes money by disconnecting us from social reality and making us feel shitty about ourselves and each other.
December 1, 2024 at 10:48 PM
Anyone saying The Left must stay on Twitter to save democracy doesn’t understand how Twitter affects our psychology. Twitter makes money by disconnecting us from social reality and making us feel shitty about ourselves and each other.
Reposted by Griffon
Is it Bad to leave Twitter? No. Here are 7+ years of insights from my lab’s research that explain why.
Featuring work w/ @williambrady.bsky.social @killianmcloughlin.bsky.social
🧵
Featuring work w/ @williambrady.bsky.social @killianmcloughlin.bsky.social
🧵
December 1, 2024 at 10:48 PM
Is it Bad to leave Twitter? No. Here are 7+ years of insights from my lab’s research that explain why.
Featuring work w/ @williambrady.bsky.social @killianmcloughlin.bsky.social
🧵
Featuring work w/ @williambrady.bsky.social @killianmcloughlin.bsky.social
🧵
Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluations
Love love love this. Those who know me know that I am generally grumpy about evaluation metrics for LLMs and the building culture of benchmark beating that has been going on for awhile now. More of this instead please.
Love love love this. Those who know me know that I am generally grumpy about evaluation metrics for LLMs and the building culture of benchmark beating that has been going on for awhile now. More of this instead please.
December 3, 2024 at 10:03 AM
Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluations
Love love love this. Those who know me know that I am generally grumpy about evaluation metrics for LLMs and the building culture of benchmark beating that has been going on for awhile now. More of this instead please.
Love love love this. Those who know me know that I am generally grumpy about evaluation metrics for LLMs and the building culture of benchmark beating that has been going on for awhile now. More of this instead please.
Nonmyopic Generation of Language Models for Reasoning and Planning
A big issue with LLM reasoning is poisoning downstream tasks. In a reasoning chain if 1 instruction is nonoptimal that has a huge impact on later steps and the most popular frameworks don’t have a way to account for that
A big issue with LLM reasoning is poisoning downstream tasks. In a reasoning chain if 1 instruction is nonoptimal that has a huge impact on later steps and the most popular frameworks don’t have a way to account for that
December 1, 2024 at 11:04 AM
Nonmyopic Generation of Language Models for Reasoning and Planning
A big issue with LLM reasoning is poisoning downstream tasks. In a reasoning chain if 1 instruction is nonoptimal that has a huge impact on later steps and the most popular frameworks don’t have a way to account for that
A big issue with LLM reasoning is poisoning downstream tasks. In a reasoning chain if 1 instruction is nonoptimal that has a huge impact on later steps and the most popular frameworks don’t have a way to account for that
Does domain adaptive pretraining make a difference? Possibly not. In this paper, the authors find that FT models (medical domain) outperform base models 12% of the time. Fine tuning can be fun, but is it worth it?
arxiv.org/abs/2411.04118
arxiv.org/abs/2411.04118
Medical Adaptation of Large Language and Vision-Language Models: Are We Making Progress?
Several recent works seek to develop foundation models specifically for medical applications, adapting general-purpose large language models (LLMs) and vision-language models (VLMs) via continued pret...
arxiv.org
November 28, 2024 at 8:30 PM
Does domain adaptive pretraining make a difference? Possibly not. In this paper, the authors find that FT models (medical domain) outperform base models 12% of the time. Fine tuning can be fun, but is it worth it?
arxiv.org/abs/2411.04118
arxiv.org/abs/2411.04118
AI tools can widen skill gaps: the top decile of researchers at a material science company using AI tools saw 81% more productivity, while the bottom third had less than 25%.
“AI won’t take your job, but those who use it will” feels uncomfortably true.
“AI won’t take your job, but those who use it will” feels uncomfortably true.
November 27, 2024 at 9:21 AM
AI tools can widen skill gaps: the top decile of researchers at a material science company using AI tools saw 81% more productivity, while the bottom third had less than 25%.
“AI won’t take your job, but those who use it will” feels uncomfortably true.
“AI won’t take your job, but those who use it will” feels uncomfortably true.
Ive been thinking about agents recently, specifically multi-agent systems. This paper touches on an aspect of that, specifically, assigning "expert" roles to produce a better-formed output. Sometimes LLMs can be a little unidimentional and more of this might change that
arxiv.org/abs/2411.00492
arxiv.org/abs/2411.00492
Multi-expert Prompting Improves Reliability, Safety, and Usefulness of Large Language Models
We present Multi-expert Prompting, a novel enhancement of ExpertPrompting (Xu et al., 2023), designed to improve the large language model (LLM) generation. Specifically, it guides an LLM to fulfill an...
arxiv.org
November 25, 2024 at 9:27 AM
Ive been thinking about agents recently, specifically multi-agent systems. This paper touches on an aspect of that, specifically, assigning "expert" roles to produce a better-formed output. Sometimes LLMs can be a little unidimentional and more of this might change that
arxiv.org/abs/2411.00492
arxiv.org/abs/2411.00492
Reposted by Griffon
Every time I see someone post this image it goes viral
November 19, 2024 at 8:03 PM
Every time I see someone post this image it goes viral