HN Atlas
← All topics

Machine learning research and practice

Neural networks, transformers, learning methods, model training and evaluation, and practical machine-learning research.

24,185 stories · 405 in 30 days · Updated Sep 28, 2026

The pace of conversation

Posts per month

2020-01: 371 posts2020-02: 385 posts2020-03: 325 posts2020-04: 409 posts2020-05: 421 posts2020-06: 427 posts2020-07: 400 posts2020-08: 358 posts2020-09: 334 posts2020-10: 362 posts2020-11: 345 posts2020-12: 317 posts2021-01: 251 posts2021-02: 253 posts2021-03: 265 posts2021-04: 221 posts2021-05: 247 posts2021-06: 232 posts2021-07: 241 posts2021-08: 226 posts2021-09: 227 posts2021-10: 239 posts2021-11: 221 posts2021-12: 213 posts2022-01: 189 posts2022-02: 199 posts2022-03: 228 posts2022-04: 263 posts2022-05: 259 posts2022-06: 243 posts2022-07: 235 posts2022-08: 210 posts2022-09: 213 posts2022-10: 225 posts2022-11: 210 posts2022-12: 248 posts2023-01: 284 posts2023-02: 292 posts2023-03: 349 posts2023-04: 410 posts2023-05: 372 posts2023-06: 375 posts2023-07: 308 posts2023-08: 297 posts2023-09: 297 posts2023-10: 363 posts2023-11: 335 posts2023-12: 338 posts2024-01: 323 posts2024-02: 314 posts2024-03: 371 posts2024-04: 413 posts2024-05: 380 posts2024-06: 302 posts2024-07: 324 posts2024-08: 280 posts2024-09: 245 posts2024-10: 346 posts2024-11: 320 posts2024-12: 287 posts2025-01: 312 posts2025-02: 286 posts2025-03: 260 posts2025-04: 298 posts2025-05: 281 posts2025-06: 264 posts2025-07: 293 posts2025-08: 304 posts2025-09: 267 posts2025-10: 316 posts2025-11: 304 posts2025-12: 293 posts2026-01: 317 posts2026-02: 308 posts2026-03: 310 posts2026-04: 281 posts2026-05: 298 posts2026-06: 291 posts2026-07: 318 posts2026-08: 267 posts2026-09: 380 posts
2020-012026-09

Posts per month · first and latest months may be partial.

01 / EXPLORE THROUGH TIME

Timeline

Start with the latest. Scroll back through time, or zoom in for more.

Year by yearUp to 5 stories per year
20262,770 stories
Kimi-K3 on HuggingFace↗ huggingface.co · HN · 544 comments · 1382 points · Jul 27, 2026
Kimi K3: Open Frontier Intelligence↗ kimi.com · HN · 1214 comments · 2107 points · Jul 16, 2026
Show HN: Getting GLM 5.2 running on my slow computer↗ github.com · HN · 240 comments · 937 points · Jul 09, 2026
Gemma 4 12B: A unified, encoder-free multimodal model↗ blog.google · HN · 401 comments · 1062 points · Jun 03, 2026
Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model↗ qwen.ai · HN · 458 comments · 993 points · Apr 22, 2026
20253,478 stories
DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]↗ huggingface.co · HN · 465 comments · 982 points · Dec 01, 2025
Leaving Meta and PyTorch↗ soumith.ch · HN · 173 comments · 721 points · Nov 07, 2025
Show HN: I invented a new generative model and got accepted to ICLR↗ discrete-distribution-networks.github.io · HN · 91 comments · 656 points · Oct 10, 2025
Deep learning gets the glory, deep fact checking gets ignored↗ rachel.fast.ai · HN · 158 comments · 609 points · Jun 03, 2025
Tracing the thoughts of a large language model↗ anthropic.com · HN · 394 comments · 1072 points · Mar 27, 2025
20243,905 stories
Llama 3.2: Revolutionizing edge AI and vision with open, customizable models↗ ai.meta.com · HN · 328 comments · 924 points · Sep 25, 2024
Flux: Open-source text-to-image model with 12B parameters↗ blog.fal.ai · HN · 226 comments · 683 points · Aug 01, 2024
Slack AI Training with Customer Data↗ slack.com · HN · 401 comments · 784 points · May 16, 2024
Visualizing Attention, a Transformer's Heart [video]↗ 3blue1brown.com · HN · 172 comments · 999 points · Apr 14, 2024
Fine tune a 70B language model at home↗ answer.ai · HN · 206 comments · 909 points · Mar 07, 2024
20234,020 stories
Time is encoded in the weights of finetuned language models↗ arxiv.org · HN · 55 comments · 124 points · Dec 24, 2023
"Attention", "Transformers", in Neural Network "Large Language Models"↗ bactra.org · HN · 76 comments · 300 points · Dec 24, 2023
Ferret: A Multimodal Large Language Model↗ github.com · HN · 308 comments · 621 points · Dec 23, 2023
Show HN: Emu2 – A Gemini-like open-source 37B Multimodal Model↗ news.ycombinator.com · HN · 27 comments · 153 points · Dec 22, 2023
Cerebras-GPT: A Family of Open, Compute-Efficient, Large Language Models↗ cerebras.net · HN · 22 comments · 100 points · Mar 28, 2023
20222,722 stories
The Chaostron: An Important Advance in Learning Machines (1961)↗ dl.acm.org · HN · 19 comments · 1487 points · Oct 15, 2022
Ask HN: In 2022, what is the proper way to get into machine/deep learning?↗ news.ycombinator.com · HN · 204 comments · 539 points · Aug 16, 2022
YaLM-100B: Pretrained language model with 100B parameters↗ github.com · HN · 625 comments · 736 points · Jun 23, 2022
OPT: Open Pre-trained Transformer Language Models↗ arxiv.org · HN · 224 comments · 461 points · May 03, 2022
Algorithms for Decision Making [pdf]↗ algorithmsbook.com · HN · 50 comments · 498 points · Apr 22, 2022
20212,836 stories
The First Rule of Machine Learning: Start Without Machine Learning↗ eugeneyan.com · HN · 172 comments · 764 points · Sep 22, 2021
Show HN: 40k HN comments mentioning books, extracted using deep learning↗ hacker-recommended-books.vercel.app · HN · 273 comments · 1359 points · Sep 20, 2021
Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash↗ github.com · HN · 351 comments · 623 points · Aug 19, 2021
Hash collision in Apple NeuralHash model↗ github.com · HN · 696 comments · 1389 points · Aug 18, 2021
Algorithms for Decision Making↗ algorithmsbook.com · HN · 85 comments · 693 points · Jan 10, 2021
20204,454 stories
You might not need machine learning↗ nullprogram.com · HN · 193 comments · 437 points · Nov 24, 2020
Full Autopilot in GTA Using TensorFlow↗ littlemountainman.github.io · HN · 111 comments · 486 points · May 12, 2020
Ask HN: Recommend me a course on Coursera↗ news.ycombinator.com · HN · 135 comments · 603 points · Apr 09, 2020
Language Learning with Netflix extension↗ soranews24.com · HN · 140 comments · 458 points · Jan 13, 2020
Dive into Deep Learning↗ d2l.ai · HN · 89 comments · 528 points · Jan 03, 2020
Scroll through time ↕

02 / CATCH UP

The conversation now

Popular stories, with attention fading over roughly two weeks.

Transformers Explained Visually↗ poloclub.github.io · HN · 92 comments · 658 points · Sep 21, 2026
DeepSeek Elastic Compute (DSec)↗ arxiv.org · HN · 110 comments · 321 points · Sep 26, 2026
Can gzip be a language model?↗ nathan.rs · HN · 164 comments · 413 points · Sep 22, 2026
How GLM built its own inference infrastructure↗ z.ai · HN · 285 comments · 411 points · Sep 17, 2026
Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM↗ github.com · HN · 80 comments · 277 points · Sep 21, 2026

03 / GO DEEPER

Follow your curiosity

Find a particular story, revisit a year, or see what resonated.

Highest rated · the last 30 days

I trained a small transformer in 1.5hrs and it beats many LLMs↗ mvakde.github.io · HN · 168 comments · 671 points · Sep 01, 2026
Transformers Explained Visually↗ poloclub.github.io · HN · 92 comments · 658 points · Sep 21, 2026
AlphaGenome Atlas: a high-resolution map of human DNA↗ blog.google · HN · 133 comments · 607 points · Sep 08, 2026
Can I opt out of my input or output data being used for training?↗ help.mistral.ai · HN · 244 comments · 498 points · Sep 02, 2026
Can gzip be a language model?↗ nathan.rs · HN · 164 comments · 413 points · Sep 22, 2026
How GLM built its own inference infrastructure↗ z.ai · HN · 285 comments · 411 points · Sep 17, 2026
DeepSeek Elastic Compute (DSec)↗ arxiv.org · HN · 110 comments · 321 points · Sep 26, 2026
The Emergent Symbolic Structure of Artificial Neural Networks↗ arxiv.org · HN · 110 comments · 297 points · Sep 02, 2026
Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM↗ github.com · HN · 80 comments · 277 points · Sep 21, 2026
Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash↗ cactuscompute.com · HN · 93 comments · 236 points · Sep 18, 2026

10 stories shown.

Help shape Hacker Atlas

Useful, confusing, or missing something? I’d love to hear.

Your feedback is private. Your email is only used to reply, never to sign you up for a newsletter.