Unverified50% confidenceFactExact time
Keera's architecture is based on GPT-2 small: 12 Transformer blocks, hidden dimension 768, and a total parameter count of 97.5M.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedGPT-2 comes in four sizes: 117M, 345M, 762M, and 1.5B parameters.66% similarUnverifiedGoogle's Switch Transformer, Mistral's Mixtral 8x7B, and the reportedly GPT-4 architecture all adopt sparse MoE designs, activating only a subset of parameters during inference.62% similarUnverifiedThe theoretical basis for modern large models such as GPT-4 and DeepSeek continuously scaling up is the combination of the Transformer architecture and scaling laws.61% similarVerifiedModern large models (such as the GPT series, Qwen, and DeepSeek) generally adopt the Decoder-only variant of the Transformer61% similarUnverifiedKarpathy's microGPT project is a minimalist GPT training implementation whose core code is only a few hundred lines, aimed at helping developers understand the essence of the Transformer architecture.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1057712API
curl https://kongchang.com/api/v1/knowledge/claims/1057712MCP
get_claim(id=1057712)