Unverified50% confidenceOpinionExact time
Obscure legacy protocols like QIC-117 are nearly absent from large model training datasets, resulting in weaker model capability regarding them.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/2/2026
First Seen
Valid until: 12/1/2026
Sources
Related Entities
Related Claims
VerifiedDuring LoRA fine-tuning, the original model parameters are frozen and only the inserted low-rank matrices are updated, significantly reducing VRAM requirements and training costs.72% similarUnverifiedLeCun批评自回归LLM在系统2慢思考上存在根本性局限、对物理世界缺乏基础、token预测训练信号是平面的67% similarUnverifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses66% similarUnverified文生图模型的理解本质上是大规模统计关联，面对训练集中稀少出现的长尾概念时生成质量会显著下滑66% similarUnverified2022年DeepMind发表的Chinchilla论文修正了最优计算资源分配策略，指出此前模型普遍存在过大但训练不足的问题65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/844266API
curl https://kongchang.com/api/v1/knowledge/claims/844266MCP
get_claim(id=844266)