| 1. |
Chain-of-thought can hurt performance on tasks where thinking makes humans worse (arxiv.org) |
|
371 points by benocodes on Oct 30, 2024 | hide | past | pdf | 250 comments
|
|
| 2. |
LLMs know more than they show: On the intrinsic representation of hallucinations (arxiv.org) |
|
137 points by benocodes on Oct 30, 2024 | hide | past | pdf | 140 comments
|
|
| 3. |
One Model to Learn Them All (arxiv.org) |
|
5 points by Anon84 on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 4. |
Designing Robust Cyber-Defense Agents with Evolving Behavior Trees (arxiv.org) |
|
3 points by PaulHoule on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 5. |
The Geometry of Concepts: Sparse Autoencoder Feature Structure (arxiv.org) |
|
2 points by robg on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 6. |
The AI Scientist: Towards Automated Open-Ended Scientific Discovery (arxiv.org) |
|
2 points by Anon84 on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 7. |
The Geometry of Concepts: Sparse Autoencoder Feature Structure (arxiv.org) |
|
2 points by roboboffin on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 8. |
LLMmap: Fingerprinting for Large Language Models (arxiv.org) |
|
2 points by emk_709 on Oct 30, 2024 | hide | past | pdf | 2 comments
|
|
| 9. |
Hacking Back the AI-Hacker: Prompt Injection as a Defense for LLM-Attackers (arxiv.org) |
|
2 points by emk_709 on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 10. |
Deep Optimizer States: Scalable Training of Transformer Interleaved Offloading (arxiv.org) |
|
1 point by sandwichsphinx on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 11. |
Conditional Hallucinations for Image Compression (arxiv.org) |
|
1 point by Hard_Space on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 12. |
Generator Matching: Generative modeling with arbitrary Markov processes (arxiv.org) |
|
1 point by lnyan on Oct 30, 2024 | hide | past | pdf | discuss
|
|
| 13. |
GPT-4o System Card [pdf] (arxiv.org) |
|
1 point by SerCe on Oct 30, 2024 | hide | past | pdf | 1 comment
|
|
