Unverified50% confidenceBenchmarkExact time
On the author's own TypeScript monorepo, an out-of-the-box code graph tool achieved a recall rate of only 57.4%, finding just 27 of 47 real callers.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
Unverified千问3.8 27B的代码方向进步明显大于文本创作，剧本拆解能力相比3.6有提升但幅度有限65% similarUnverifiedIn a speculative decoding test using Qwen2.5 0.5B as the draft model proposing 4 tokens at a time, the overall draft acceptance rate was 32.9% for serial processing and 37.1% for concurrent processing.64% similarUnverifiedWhen reviewing the author's video generation framework, Opus 5.5 pointed out that the entire pipeline was written in 14,000 lines of Python and JS, while the source code that actually determines the visuals is only 651 lines, with the remainder being largely useless tests.64% similarUnverifiedIn tests covering context lengths of 512–4096, increasing the draft length from 4 tokens to 16 tokens resulted in only a 0.2%–1.3% increase in overall inference latency.61% similarUnverifiedApproximately 42% of the 722 manuscripts released by OpenAI were formalized in Lean, with each result consuming on average about three hours of ChatGPT Pro inference compute.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1004252API
curl https://kongchang.com/api/v1/knowledge/claims/1004252MCP
get_claim(id=1004252)