Insight for: is_amplify_history 和 restrict_self_attn 问题

This issue queries the application of specific training strategies (`is_amplify_history`, `restrict_self_attn`) in Helios's base model weights, given their default `false` configuration. The developer is seeking clarity on the model's underlying training methodology. This indicates a need for transparent documentation regarding model architecture and training parameters, which is crucial for reproducibility, fine-tuning, and understanding model behavior. For B2B SaaS in the AI model space, clear communication of such technical details builds developer confidence and facilitates integration. Ambiguity around core training strategies can hinder adoption by enterprises that require deep insight into model provenance and performance characteristics for compliance, security, and optimization.