Register and share your invite link to earn from video plays and referrals.

Search results for 昌子源
昌子源 community
One keyword maps to one global community path.
Create community
People
Not Found
Tweets including 昌子源
DeepSeek V4.1 Gives Prefill and Decode Different Compute Paths Prompt tokens mostly traverse 20 layers; generated tokens traverse all 40. This Causal Encoder-Decoder design nearly halves long-input prefill while preserving autoregressive generation. Zhihu contributor 潜龙勿用, Changxin Ke(柯昌鑫), a graduate researcher at ICT, CAS, explains how its architecture and post-training were designed together. 1️⃣ A causal encoder, not T5 The 40-layer backbone is split into a 20-layer causal encoder and a 20-layer decoder. Both remain causal. The encoder processes the prompt and supplies the decoder’s global KV. Most prompt tokens avoid the decoder stack, while generated tokens still run through all 40 layers. That means 8B active parameters per prefill token versus 16B per decode token. 2️⃣ Most layers share global memory CSA2 uses three modes: Full creates global KV and an index; Reindex shares the KV but selects new positions; Reuse shares both. Only four layers create independent global KV, four reindex it, and 30 reuse both KV and the latest index. Each layer retains its own query and local SWA state. The first Full layer builds up to 16,384 candidates. Later layers search this pool for their Top-512. From 4K to 1M context, decode FLOPs per token rise only about 25%. 3️⃣ Serving approximations enter training Exact reconstruction of decoder SWA states would replay 2,560 prompt tokens. V4.1 replays only the final 128 encoder outputs and trains the model to tolerate the approximation. With FP4 global KV, cache falls to 890 bytes per token, persistent cache to roughly one eighth of V4-Flash, and prefill compute close to half. Bounded replay and constrained retrieval are not last-minute serving tricks. The model experiences them during post-training. 4️⃣ Post-training is an evolving Agent system Each RL task combines a problem, environment, and verifier. New trajectories can reveal shortcuts, broken environments, or verifier errors and send the task back for repair. V4.1 trains across multiple harnesses, merges checkpoints between RL runs, and finishes with on-policy distillation from more than 40 teachers. Raising reasoning effort from 25 to 100 increases output length about 2.5×, while average Pass@1 across eight benchmarks rises from 67.1% to 76.3%. ✅ The real design choice V4.1 aligns model structure, cache policy, retrieval limits, training environments, and inference around long-running Agents. The model learns under the same constraints the deployed system will actually impose. 🔗 Full analysis: #DeepSeek# #DeepSeekV41# #LLMArchitecture# #AIAgents# #LongContext# #AIInfra#
Show more
The Nanchang-Jiujiang High-Speed Railway has entered the static acceptance phase. Stretching 138 kilometers with a designed speed of 350 kilometers per hour, it runs from Jiujiang City in northern Jiangxi Province, China, south to Nanchang City. 昌九高铁进入静态验收阶段。这条高铁全长138公里,设计时速350公里,北起中国江西省九江市,南至南昌市。 #China# #Jiangxi# #JiangxiEconomy# #世界经济看中国# #赣出新精彩#
Show more
Another high-speed railway in Jiangxi has had its entire track laid through: The Nanchang–Jiujiang high-speed railway has a main line length of about 138 kilometers, with a design speed of 350 km/h, and is expected to be completed and open to traffic by March 2027. 江西又一条高铁全线铺轨贯通:昌九高铁正线全长约138公里,设计时速350公里,预计于2027年3月建成通车。 #China# #Jiangxi# #赣出新精彩# #JiangxiScenery#
Show more
Today is June 22nd, the eighth day of the first month of May according to the Lunar calendar. The eighth day is auspicious and auspicious. May you have eightfold wealth and energy with you, eight layers of joy surround your courtyard, your family business prosper, your years prosper, your body be strong and your spirit refreshed. 今日是6月22号,农历五月初八。八喻兴旺,吉运满堂。愿您八方财气随身至,八重喜乐绕庭堂,家业兴旺岁岁昌,身体硬朗精神爽。
Show more