标签：ai
带有 ai 标签的公开 Whisper
划选高亮2026-08-12 13:35:25
原文高亮摘录
“capable of recalling and reasoning over fine-grained information”
Whisper 随想笔记
Recalling fine-grained stuff is cool, but I'd rather see it not hallucinate the details first.
划选高亮2026-08-12 13:26:25
原文高亮摘录
“capable of recalling and reasoning over fine-grained information”
Whisper 随想笔记
10M tokens is wild, but can it actually find that one specific line in a 500-page PDF?
划选高亮2026-08-12 10:32:25
原文高亮摘录
“next generation of highly compute-efficient multimodal models”
Whisper 随想笔记
Still waiting for the open source model that does this without breaking the bank.
划选高亮2026-08-12 10:23:25
原文高亮摘录
“next generation of highly compute-efficient multimodal models”
Whisper 随想笔记
Million token context is wild, finally can throw whole codebases at it.
划选高亮2026-08-12 10:14:25
原文高亮摘录
“next generation of highly compute-efficient multimodal models”
Whisper 随想笔记
Compute-efficient but what about the energy cost for training these things?
划选高亮2026-08-11 18:38:31
原文高亮摘录
“QLoRA backprop”
Whisper 随想笔记
My 48GB card finally feels useful for something other than gaming.
划选高亮2026-08-11 03:06:31
原文高亮摘录
“General reasoning represents a long-standing and formidable challenge”
Whisper 随想笔记
The abstract is truncated, but DeepSeek-R1's RL approach sounds really promising.
划选高亮2026-08-10 14:32:38
原文高亮摘录
“a herd of language models that natively support multilingua”
Whisper 随想笔记
multilinguality is cool but tool usage is where it's really at
划选高亮2026-08-10 14:23:38
原文高亮摘录
“a herd of language models that natively support multilingua”
Whisper 随想笔记
405B params is insane, wonder how much compute that took
划选高亮2026-08-10 14:14:38
原文高亮摘录
“a herd of language models that natively support multilingua”
Whisper 随想笔记
finally a model that actually gets multiple languages without breaking a sweat
划选高亮2026-08-08 15:09:57
原文高亮摘录
“The term machine learning was coined in 1959 by Arthur Samuel”
Whisper 随想笔记
Arthur Samuel coined the term way back in 1959, that's wild.
划选高亮2026-08-08 15:09:18
原文高亮摘录
“this is sounding awfully familiar.”
Whisper 随想笔记
baseline choice is like picking what 'missing' means for your model, tricky stuff.
划选高亮2026-08-01 03:09:18
原文高亮摘录
“standard interface for models, embeddings, vector stores, and more.”
Whisper 随想笔记
Thats the core value prop right there, a standard interface for everything LLM.
划选高亮2026-07-30 21:09:18
原文高亮摘录
“not a modular toolbox of building blocks for neural nets.”
Whisper 随想笔记
That's a refreshing take, keeping it simple for researchers to hack on.
划选高亮2026-07-28 16:29:34
原文高亮摘录
“reduce the training time”
Whisper 随想笔记
Finally a normalization that works for RNNs without all that batch size fuss.
划选高亮2026-07-28 13:26:34
原文高亮摘录
“batch norma”
Whisper 随想笔记
Does this actually scale well for CNNs though? Batch norm felt more natural there.
划选高亮2026-07-28 13:23:53
原文高亮摘录
“An energy function is a macroscopic measure of a network's activation state.”
Whisper 随想笔记
Wait, so lower energy means better? That's like the opposite of what I'd expect from "energy".
划选高亮2026-07-28 13:17:34
原文高亮摘录
“batch norma”
Whisper 随想笔记
So layer norm is basically batch norm but for single examples, makes sense for RNNs.
划选高亮2026-07-28 13:14:53
原文高亮摘录
“An energy function is a macroscopic measure of a network's activation state.”
Whisper 随想笔记
So it's basically using physics jargon to describe how the network settles into a stable state. Cool.
划选高亮2026-07-28 13:04:22
原文高亮摘录
“use of consistency training on a large amount of unlabe”
Whisper 随想笔记
So basically, they found that how you add noise matters more than the amount of unlabeled data?