Unverified50% confidenceBenchmarkExact time
Models trained at 340M and 1.3B parameters match or exceed Gated DeltaNet and Mamba-2 on language modeling perplexity and zero-shot accuracy, remain competitive with Transformers of the same size, and have better recall capability than linear recurrence.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
UnverifiedThe formal model uses DictVectorizer for one-hot encoding (fit on the training set only), with logistic regression set to solver="liblinear" and C=1.0, giving a validation-set AUC of 0.732.70% similarUnverifiedWithout changing the prompt, the capabilities of the SymCE-trained model transfer to mainstream mathematical reasoning benchmarks such as GSM8K, MATH-500, and MMLU-college-math.69% similarVerifiedLarge language models are optimized to make code look correct rather than to meet current-stage requirements; they are fundamentally probability-statistical machines that predict the next most likely token and excel at reproducing high-frequency patterns from training data.68% similarUnverifiedChoosing simple models such as Ridge Regression over complex models like LSTM and XGBoost is preferable at the prototype stage, because interpretability and training stability are more important, and complex models are prone to overfitting on small-scale synthetic datasets.67% similarUnverified在代码语料比例更高的混合数据集上训练的模型，在数学推理基准（如MATH、GSM8K）上的表现显著优于纯自然语言训练的同规模模型，这一结论获斯坦福大学和MIT的多项研究证实67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114899API
curl https://kongchang.com/api/v1/knowledge/claims/1114899MCP
get_claim(id=1114899)