Unverified50% confidenceFactExact time
The tester ran only the 5M and 28M models, not the 123M and 209M.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/20/2026
First Seen
Valid until: 11/3/2026
Sources
Related Entities
Related Claims
UnverifiedThe parameter count of 124M is close to the scale of GPT-2 Small.72% similarUnverifiedTTM 模型参数量仅约 0.8M（百万级），文件大小约三兆字节，无需 GPU，在普通 CPU 上就能运行68% similarUnverifiedThe TEF 0.8B version model is only 811MB in size; in actual testing, running an entire set of classification tasks with TEF 0.8B took only 0.23 seconds, while the same tasks with GPT took about 2.5 seconds.67% similarUnverifiedAfter 4-bit quantization, the weight size of a 2.8T parameter model may reach 1.4TB or more.66% similarUnverifiedA developer reproduced it on a 2-vCPU virtual machine: the 5M model had a median latency of 52.4ms and the 28M model 315.7ms (torch limited to 2 threads).66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1072909API
curl https://kongchang.com/api/v1/knowledge/claims/1072909MCP
get_claim(id=1072909)