Unverified50% confidenceFactExact time
SHADOW does not use a trained embedding layer; instead, it represents a vocabulary of 73,880 tokens using fixed 512-bit fingerprints, the frozen table is only 4.7 MB, and new tokens can be added at any time without retraining.
1
Sources
50%
Confidence
Long-term
Relevance
9/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSHADOW writes attention states to disk at 1-bit precision, consuming 288 bytes per token, with the index occupying only 22 bytes per token, record lookup taking approximately 1 microsecond, and attention states being reinjected into the model in approximately 0.03 ms, without using a vector database or embedding model.66% similarUnverifiedSHADOW-50M is a complete language model trained from scratch that has digested 45 billion tokens, actually has 44M parameters, and has a model size of 19.8 MB.66% similarUnverifiedUnsloth 通过 4-bit/8-bit 量化感知训练与动态激活卸载技术，可将 7B 模型的微调显存需求压缩至约 4–6 GB60% similarUnverifiedQwen3.5-9B contains approximately 9 billion trainable parameters, and after 4-bit quantization, the model file size can be compressed to approximately 5–6 GB.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/922669API
curl https://kongchang.com/api/v1/knowledge/claims/922669MCP
get_claim(id=922669)