SURL AI News
繁中English日本語

Archive

最新模型

新模型、開放權重、多模態、語音、影像與模型架構。

最新模型

Gensyn open-1b Builds a Cross-Hardware, Bit-Level Audit Trail for 80,957 Training Steps

Gensyn has released the data, checkpoints, step-by-step hashes, and replay tools for its 1.61B-parameter model, allowing third parties to spot-check whether training states match its commitments. This reproducibility comes at the cost of roughly a fivefold reduction in training performance, and matching hashes alone still cannot prove that the original cluster actually ran as claimed.
訓練基礎設施

TRL 1.14 Syncs Only LoRA Weights, Removing NCCL Dependency for Cross-Machine Asynchronous GRPO Training

Hugging Face distributes `AsyncGRPOTrainer` training and vLLM rollouts across separate machines, transferring only a few megabytes of LoRA adapters through shared object storage. A published experiment cut a 500-step run from 3 hours 27 minutes to 53 minutes, but the result is still based on a single small-model recipe and does not prove large-scale RL convergence or cost advantages.
AI 安全

“Plan Injection” Makes Models Rewrite External Faulty Reasoning as Their Own—and Evade Chain-of-Thought Monitoring

New research finds that malicious plans phrased in benign language and inserted into context can be absorbed, rewritten, and executed by models, with even the strongest monitors missing at least a quarter of attacks. Showing monitors the original injected content or increasing the reasoning budget does not reliably improve detection, suggesting that agents cannot establish their safety solely through their output chain of thought.
最新模型

ShadowPEFT Lands in Hugging Face PEFT Main, Replacing Distributed LoRA Weight Deltas with Cross-Layer State

ShadowPEFT models task adaptation as a small trainable model running alongside a frozen backbone and is now integrated with PEFT’s standard configuration, saving, and loading interfaces. Early experiments outperform LoRA and DoRA at comparable parameter counts, but additional memory usage, dual KV caches, and the lack of a stable release remain deployment trade-offs.
AI 安全

SAILS Shows Backdoor Success Depends on the Poisoned Sample Set, While Random Sampling May Underestimate Worst-Case Risk

With the model, clean data, and number of poisoned samples held constant, a new study changed only the composition of the poisoned sample set and saw attack success rates on Llama 3 8B range from 3% to 80%. SAILS trains a set-scoring model using a small number of costly fine-tuning evaluations, then selects high-risk sets from millions of candidate combinations.
代理工具

Bolt Forge Trades 50× Agent Credits for Secret-Scrubbed Code-Repair Trajectories

Bolt.new’s research preview runs its website-building agent on open-weight models from GLM, Kimi, and DeepSeek, while providing Arcee AI with prompts, code, and error-repair sequences from users who explicitly opt in for training. The program clearly illustrates a “compute for data” model, but its self-reported 91% benchmark result and secret-scrubbing effectiveness still require external validation.