人工智能2022 · 分享于 Hacker News
Training Compute-Optimal Large Language Models
探索《Training Compute-Optimal Large Language Models》的原文与 Hacker News 社区讨论，查看来源并加入阅读清单。
从论文，走向讨论
看看社区如何讨论这个想法。
来自 Hacker News 的原始讨论。不同的视角、疑问，以及值得继续思考的线索。
最多展示 100 条相关讨论。论文可能出现在评论中，并非原帖投稿内容。
在讨论中发现的引用
按 HN 条目顺序展示最近的 50 条引用来源。
评论 #48560093 评论 #46324366 评论 #44808589 评论 #43937317 评论 #38677460 评论 #38662687 评论 #37937668 评论 #37550394 评论 #36883432 评论 #36777492 评论 #36700825 评论 #35810294 评论 #35810215 评论 #35737571 评论 #35631428 评论 #35577425 评论 #35508682 评论 #35332959 评论 #35299718 评论 #35110454 评论 #34944775 评论 #34936903 评论 #34861412 评论 #34751882 评论 #34682566 评论 #34146825 评论 #34049644 评论 #34028941 评论 #33730551 评论 #33342468 评论 #32945083 评论 #32527518 评论 #32462472 评论 #30995935 评论 #30982442 评论 #30981024 评论 #30914697 评论 #30912871
Yhardmaru ·
Stable Diffusion 2.0
1,320 赞同471 评论
Yiamwil ·
History LLMs: Models trained exclusively on pre-1913 texts
876 赞同410 评论
Yjasondavies ·
John Carmack's new AGI company, Keen Technologies, has raised a $20M round
461 赞同619 评论
Yhomarp ·
Pathways Language Model (PaLM): Scaling to 540B parameters
235 赞同207 评论
Yfriggeri ·
Llama 2 – Meta AI
235 赞同55 评论
Ygeorgehill ·
DeepMind’s New Language Model, Chinchilla
229 赞同142 评论
YHard_Space ·
The road to realistic full-body deepfakes
220 赞同201 评论
Yraldi ·
GPT-3 can create both sides of an Interactive Fiction transcript
132 赞同68 评论
Yjmsflknr ·
Sundar Pichai: An important next step on our AI journey
127 赞同78 评论
YBostonian ·
Why do LLMs have emergent properties?
83 赞同110 评论
Ydavidbarker ·
Stability AI Launches the First of Its StableLM Suite of Language Models
76 赞同15 评论
Ygardenfelder ·
The LLama Effect: Leak Sparked a Series of Open Source Alternatives to ChatGPT
58 赞同7 评论
YBrajeshwar ·
Estimating PaLM's Training Cost
51 赞同17 评论
Yerwald ·
The Prospect of an AI Winter
44 赞同17 评论
Ytosh ·
SparseGPT: Language Models Can Be Accurately Pruned in One-Shot
24 赞同1 评论
Ygslin ·
Stability AI releases its latest image-generating model, Stable Diffusion XL 1.0
24 赞同7 评论
Yviburnum ·
Carmack Leaves Meta
21 赞同2 评论
Youtrun86 ·
Consortium launched to build the largest open LLM
18 赞同5 评论
Yog_kalu ·
Large Language Models for Compiler Optimization
15 赞同1 评论
Ymfiguiere ·
LLaMA: A foundational, 65B-parameter large language model
12 赞同6 评论
Yqualudeheart ·
Training Compute-Optimal Large Language Models
8 赞同0 评论
Yfaustog ·
70B language model outperforms Gopher and GPT-3
5 赞同0 评论
Yadityaathalye ·
The wall confronting large language models
5 赞同1 评论
Ymonort ·
Language Models Can Teach Themselves to Program Better
4 赞同1 评论
Ymetalwhale ·
Open source solution replicates ChatGPT training process
4 赞同0 评论
YDDSDev ·
OpenAI’s CEO confirms the company isn’t training GPT-5 and ‘won’t for some time’
4 赞同3 评论
Yvsroy ·
Llama 1.3B Trained on 200B Tokens for Commercial Use
4 赞同1 评论
Yhmokiguess ·
Cohere's First Model for Developers
4 赞同0 评论
Yzuhayeer ·
Training Compute-Optimal Large Language Models
3 赞同0 评论
Ysquidhunter ·
GPT in 60 Lines of NumPy
3 赞同0 评论
Yswyx ·
Replit's new Code LLM: Open Source, 77% smaller than Codex, trained in 1 week
3 赞同0 评论
Ysharonzhou ·
Free and Fast LLM Finetuning
3 赞同3 评论
Yjjwiseman ·
Copilot Internals
2 赞同0 评论
Ymrtksn ·
LLaMA-7B in Pure C++ with full Apple Silicon support
2 赞同1 评论
Ybelter ·
Do large language models need all those layers?
2 赞同0 评论
Ygmays ·
Training Compute-Optimal Large Language Models
1 赞同0 评论
Ykrm01 ·
Ask HN: What's the value ratio of GPT3's LLM vs. its training data?
1 赞同2 评论
Yatleastoptimal ·
GPT-4 performs significantly worse on coding problems not in its training data
1 赞同0 评论
Yhahnchen ·
Ask HN: AI/ML papers to catch up with current state of AI?
1 赞同0 评论