Unverified50% confidenceBenchmarkExact time
After identifying the matching 112-shape tuning table (16 shapes × 7 T-buckets) and enabling MMQ and Prefill chunking, Fresh Prefill for 4–9K jumped from 405 to 701–775 tokens/s, approximately a 1.8× improvement, with decode stabilizing at 52–55.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/5/2026
First Seen
Valid until: 1/3/2027
Sources
Related Entities
Related Claims
UnverifiedWhen enabling multi-token prediction (MTP) with a draft depth of 2, the R9700 rose from 27.1 to 46.2 tokens/s (+70%) and the B70 rose from 21.4 to 28.4 tokens/s (+33%).70% similarUnverifiedIn LOD generation testing, Mixar generated three levels—LOD0, LOD1, and LOD2—with textures reduced from 4K to 2K and 1K, creating a total of 52 new textures, with results meeting expectations.62% similarUnverifiedEfficientNet 从 B0 到 B7，模型参数从 5.3M 增长到 66M，Top-1 精度从 77.1% 提升到 84.3%61% similarUnverifiedThe Mersenne Twister algorithm uses a state array of 624 32-bit integers (approximately 2.5 KB of memory) and employs a four-step tempering operation: right-shift by 11 bits, left-shift by 7 bits with a mask, left-shift by 15 bits with a mask, and right-shift by 18 bits.61% similarUnverifiedAfter adopting the causal encoder-decoder, the activated parameters per token during the prefill phase were reduced from 16B to 8B.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/978740API
curl https://kongchang.com/api/v1/knowledge/claims/978740MCP
get_claim(id=978740)