Reposted
🚨 Paper Alert: “RL Finetunes Small Subnetworks in Large Language Models”
From DeepSeek V3 Base to DeepSeek R1 Zero, a whopping 86% of parameters were NOT updated during RL training 😮😮
And this isn’t a one-off. The pattern holds across RL algorithms and models.
🧵A Deep Dive
From DeepSeek V3 Base to DeepSeek R1 Zero, a whopping 86% of parameters were NOT updated during RL training 😮😮
And this isn’t a one-off. The pattern holds across RL algorithms and models.
🧵A Deep Dive
May 21, 2025 at 3:50 AM
🚨 Paper Alert: “RL Finetunes Small Subnetworks in Large Language Models”
From DeepSeek V3 Base to DeepSeek R1 Zero, a whopping 86% of parameters were NOT updated during RL training 😮😮
And this isn’t a one-off. The pattern holds across RL algorithms and models.
🧵A Deep Dive
From DeepSeek V3 Base to DeepSeek R1 Zero, a whopping 86% of parameters were NOT updated during RL training 😮😮
And this isn’t a one-off. The pattern holds across RL algorithms and models.
🧵A Deep Dive
Can English-finetuned LLMs reason in other languages?
Short Answer: Yes, thanks to “quote-and-think” + test-time scaling. You can even force them to reason in a target language!
But:
🌐 Low-resource langs & non-STEM topics still tough.
New paper: arxiv.org/abs/2505.05408
Short Answer: Yes, thanks to “quote-and-think” + test-time scaling. You can even force them to reason in a target language!
But:
🌐 Low-resource langs & non-STEM topics still tough.
New paper: arxiv.org/abs/2505.05408
May 10, 2025 at 3:12 PM
Can English-finetuned LLMs reason in other languages?
Short Answer: Yes, thanks to “quote-and-think” + test-time scaling. You can even force them to reason in a target language!
But:
🌐 Low-resource langs & non-STEM topics still tough.
New paper: arxiv.org/abs/2505.05408
Short Answer: Yes, thanks to “quote-and-think” + test-time scaling. You can even force them to reason in a target language!
But:
🌐 Low-resource langs & non-STEM topics still tough.
New paper: arxiv.org/abs/2505.05408
Reposted
📣 New paper!
We observe that reasoning language models finetuned only on English data are capable of zero-shot cross-lingual reasoning through a "quote-and-think" pattern.
However, this does not mean they reason the same way across all languages or in new domains.
[1/N]
We observe that reasoning language models finetuned only on English data are capable of zero-shot cross-lingual reasoning through a "quote-and-think" pattern.
However, this does not mean they reason the same way across all languages or in new domains.
[1/N]
May 9, 2025 at 7:53 PM
📣 New paper!
We observe that reasoning language models finetuned only on English data are capable of zero-shot cross-lingual reasoning through a "quote-and-think" pattern.
However, this does not mean they reason the same way across all languages or in new domains.
[1/N]
We observe that reasoning language models finetuned only on English data are capable of zero-shot cross-lingual reasoning through a "quote-and-think" pattern.
However, this does not mean they reason the same way across all languages or in new domains.
[1/N]
Check our latest cultural survey paper presented in the #EMNLP2024 last week!
with Prof. Monojit and @sagnikmukherjee.bsky.social
with Prof. Monojit and @sagnikmukherjee.bsky.social
📢📢LLMs are biased towards Western Culture. Well, okay, but what do you mean by "Culture"?
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
November 22, 2024 at 12:42 AM
Check our latest cultural survey paper presented in the #EMNLP2024 last week!
with Prof. Monojit and @sagnikmukherjee.bsky.social
with Prof. Monojit and @sagnikmukherjee.bsky.social
Reposted
📢📢LLMs are biased towards Western Culture. Well, okay, but what do you mean by "Culture"?
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
November 21, 2024 at 10:03 PM
📢📢LLMs are biased towards Western Culture. Well, okay, but what do you mean by "Culture"?
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
In our survey of on cultural bias in LLMs, we reviewed ~90 papers. Interestingly, none of these papers define "culture" explicitly. They use “proxies”. [1/7]
[Appeared in EMNLP mains]
first skeet!
November 21, 2024 at 1:48 PM
first skeet!