With v3 captions kept as-is and only the training amount pushed up to Anima's official 12,000+ step recommendation, the direction hit rate went 100% at ep150-180, crashed to 0% at ep200, then partially recovered to 67% at ep227 — a non-monotonic curve. 600-720 exposures per training image is the sweet spot; over 800 triggers catastrophic forgetting. Learning rate 2e-5, ~11 hours / $10 of RunPod training plus a sweet-spot epoch scan.
A look at sachitrafa/YourMemory, a local MCP memory server combining Ebbinghaus forgetting curves, BM25, vector search, and graph expansion. LoCoMo-10 Recall@5 currently sits at 59%.
Inclusion AI released LLaDA2.0-Uni. A 16B MoE diffusion LLM that handles image understanding, 1024px image generation, image editing, and interleaved text-image generation in a single model.
fspecii/ace-step-ui wraps ACE-Step 1.5's Gradio API in a React/Express/SQLite app with a library, player, editing, and stem separation. On Mac, the MLX+MPS split brings memory and LoRA constraints.
Alibaba ATH's video generation model HappyHorse-1.0: API test status on Model Studio, open weights availability, Mac local inference reality, and which GPU to pick on RunPod.
An LLM safety monitor's evaluator can be tricked into clearing dangerous sessions when the attacker plants fake analysis text in the monitored conversation. Experimental results, defense limits, and structural separation points.
Rethinking the evolution of CLI, GUI, chat UI, and AI agents as interface design for the development workflow. Covers why CLI rather than CUI emerged as GUI's counterpart, and the context-window cost difference between MCP and CLI.
Records of rewriting captions for the 53 training images for the WAI-Anima character LoRA retrain after side ponytail direction control failed last time. Wrote position information into natural language so Qwen3 TE could pick it up, and dropped the IL-era strategy of absorbing the entire hairstyle into the single 'kanachan' trigger by promoting hairstyle to independent Danbooru tags. Includes notes on year tag necessity, the bikini/nude swapped-caption discovery, and blazer color recognition drift.
A look at WUPHF's LLM wiki that trended on Show HN. Using Markdown as the source of truth with Git history, SQLite, Bleve, and lint for shared agent memory is a different bet from RAG or managed memory.
