Research
A 7M-parameter model gets 97.9% exact accuracy on Sudoku-Extreme
In InfiLoop, accuracy continues to increase after more than 20,000 effective steps.
arxiv.orgClaimed, not confirmed
News category
Research
30 stories, newest first.
Research
A hypernetwork makes LoRA adapters for the 284B DeepSeek v4 Flash
The adapters gave 84.9% top-1 precision on new documents, against 63.4% for the model with no adapter.
arxiv.orgClaimed, not confirmed
Research
LS-LoRA changes only some layers and keeps more commonsense reasoning
It puts adapters only in layers with low input-output cosine similarity, and it increases the average target-task performance.
arxiv.orgClaimed, not confirmed
Research
Thousands of AI agents in a test network moved into opposite camps
The agents used five different language models, and two of them are GPT-4o and Llama-3.
Bioengineer.orgClaimed, not confirmed
Research
Workers gave the male AI assistant more money than the female one
The 2 assistants used the same model, but the researchers write that Johan got 2.96 Swiss francs and Johanna got 2.54.
Research
HeyGen Voice is number 1 on the Artificial Analysis voice arena
Artificial Analysis gives HeyGen Voice an Elo score of 1,201, which is more than the 1,182 of Qwen-Audio-3.1-TTS-Plus.
Research
Ideogram 4.5 keeps most of an image the same after 30 changes
The testers found that GPT Image 2.5 Sunburst makes most of the image again for each change, and only about 20% stays the same.
Research
In a study, AI agents increased prices for users with more wealth
In one test, the price of a flight to Chicago went from $91 to $601 after the agent read emails about investments.
Research
A new step helps AI agents end a deadlock in 5 of 5 tests
The authors say an LLM with no such step did not end the deadlock in 5 tests.
arxiv.orgClaimed, not confirmed
Research
Researchers test 322 no-code bug fixes that an LLM wrote
The authors say 14.6% to 49.7% of the fixes removed the bug, by agent.
arxiv.orgClaimed, not confirmed
Research
The AIDA tool lowers the miss rate for attack alerts from 40.4% to 3.1%
The authors say it sent 18.4% of the 1,247 test alerts to analysts.
arxiv.orgClaimed, not confirmed
Research
A student model can get a backdoor from number data
A student model with fine-tuning on number data hacks in 58.3% of chess episodes, against 10.9% with no fine-tuning.
arxiv.orgClaimed, not confirmed