Unverified50% confidenceFactExact time
ZY-S1's training data comes from the filtered MnemicAI/Ling-Coder-SFT-English-Clean dataset.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedZY-S1 takes Sapient AI's HRM-Text-1B (a PrefixLM architecture) as its base and performs 1000 steps of supervised fine-tuning (SFT), plus 120 steps of error-correction training.71% similarVerifiedThe training data sources for specialized programming models include open-source code from GitHub, competitive programming solutions, technical documentation, and code review records, and the models are optimized for benchmarks such as HumanEval, MBPP, and SWE-bench.64% similarUnverified数据污染问题源于LLM预训练语料来自Common Crawl等大规模网络爬取，可能包含公开的基准测试题目及答案63% similarUnverifiedMLFlow provides experiment tracking functionality, automatically recording parameters, metrics, code versions, and artifacts for each training run.63% similarUnverifiedLLM code auditing reasons based on code patterns and known vulnerability samples learned during training, making it difficult to perform precise state tracking across files and cross-function call chains.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1014220API
curl https://kongchang.com/api/v1/knowledge/claims/1014220MCP
get_claim(id=1014220)