AKB48 Team TP 第9張單曲《Team UP》 2025, February, 11 20:00 (TST) Release on YouTube #蘇姮羽# #0214集結發行# #TeamUP# #raise_me_up# #AKB48TeamTP# #TeamTP# #TTP⁣# #9thSingle# #第九張單曲#

AKB48 Team TP 第9張單曲《Team UP》 2025, February, 11 20:00 (TST) Release on YouTube #張少瞳# #蘇姮羽# #0214集結發行# #TeamUP# #raise_me_up# #AKB48TeamTP# #TeamTP# #TTP⁣# #9thSingle# #第九張單曲#

New video AB158 Workplace Temptation of the Pantyhose High Heel Secretary – Su Chang’s Forbidden Invitation feat. 苏畅@woshisuchang Full vid link in the comments After work, Shu Chang suddenly invited me over to her place to “discuss some work.” Not long after we walked in, she leaned in close, her tone very direct: “I want it tonight.” I told her, “I already have someone.” She just smiled and started slowly unbuttoning her shirt. Her pantyhose and high heels hadn’t even come off yet. I knew I shouldn’t go any further, but I was still completely drawn in by how beautiful she looked. We got closer and closer, and our hands started getting restless. Her breathing grew heavier, and she kept pressing her body against mine. Later I took out an “assistive tool” to help her “adjust.” It left her completely soft and trembling, her voice shaking as she finally whispered, “Just put it in… I can’t take it anymore.” We started our “in-depth collaboration.” She was incredibly tight, trembling with every thrust. When I finally pulled out, she immediately knelt down and finished the rest of the work herself. Studio @AbbyKitty6666

🧩 Kimi K3’s MoE and Attention Are Built Around Trade-offs, Not Tricks Kimi K3’s open release has drawn attention to its scale. But its architecture tells a more useful story: the hardest part of scaling is keeping quality, efficiency, and stability in balance. Zhihu contributor 苏剑林 @Jianlin_S explains the design logic behind two core components: Stable LatentMoE and K3’s hybrid attention. At a high level: K3 = KDA + MLA + Stable LatentMoE + AttnRes 1️⃣ Stable LatentMoE: more experts at similar cost LatentMoE compresses each token into a smaller latent space before routing it to experts, then projects the result back to the full hidden dimension. This reduces expert computation and communication. The saved budget can support more, narrower experts without greatly increasing training or inference cost. But the longer projection chain also magnifies numerical instability. K3 introduces three fixes. 🔹 SiTU-GLU softly caps extreme activations in both branches of the expert network. Compared with hard clipping, soft capping preserves smoother optimization. 🔹 RMSNorm is placed before the final up-projection. It stabilizes training and helps balance routed experts against shared experts. 🔹 Quantile Balancing replaces the previous load-balancing update, which became unreliable as the expert pool grew. It approximates global routing quantiles with histograms, allowing efficient aggregation across machines. The broader lesson is clear: scaling MoE is not just about adding experts. Routing, activation ranges, normalization, and distributed communication must scale with them. 2️⃣ Why K3 still uses MLA Some newer models have moved away from MLA, partly because speculative decoding changes the inference trade-off. MLA keeps KV Cache small and remains highly competitive under fixed training and memory budgets. But its decoding path is relatively compute-heavy, leaving less room for Multi-Token Prediction to trade extra computation for speed. Other attention designs simply move the bottleneck: 🔹 Smaller designs may reduce computation but lose quality or require a larger KV Cache. 🔹 Larger designs can recover quality, but increase training and prefill costs. An ideal replacement would preserve quality, reduce KV Cache, lower decoding compute, and cost no more during training or prefill. No simple design currently satisfies all four conditions. K3 therefore keeps MLA and combines it with KDA. The linear-attention layers handle most long-context processing efficiently, while MLA preserves full-attention capacity where it matters. 3️⃣ “Abandoning MLA” is not so simple Architectures that appear to replace MLA may still retain its core intuition. For example, a wide MQA design with shared K and V resembles MLA’s decoding form. Sparsity and compression can then reduce its compute and cache costs. This can work, but it introduces more infrastructure complexity. So the current debate is less about whether MLA is obsolete. It is about which combination of full, linear, sparse, and compressed attention offers the best system-level trade-off. 4️⃣ Why K3 can remove RoPE K3 removes RoPE from its MLA layers. That would hurt a pure-MLA model. But K3 is a hybrid of KDA and MLA. KDA’s DeltaNet-style updates already introduce an implicit positional transformation. In this sense, KDA provides something similar to a generalized form of RoPE for the full network. Adding explicit RoPE back produced little difference, so K3 followed the simpler design. K3 is not truly position-free. Its positional structure is partly carried by KDA instead of an explicit embedding. ⚙ The real architecture lesson None of these choices is especially flashy in isolation. Stable LatentMoE controls the numerical and routing problems created by more experts. KDA and MLA divide long-context work according to their strengths. NoPE removes a redundant component only after the hybrid architecture makes it unnecessary. K3’s main design principle is therefore not novelty for its own sake. Every architectural change must justify itself across quality, efficiency, and stability. 🔗 Full reading: 📖Blog post: #KimiK3# #MoE# #Attention# #LLM# #AIInfra# #OpenSourceAI#

#世界点赞景德镇# On the morning of October 18, the 2026 China Jingdezhen International Ceramic Expo grandly opened. This is the first ceramic expo held after "Jingdezhen Handicraft Porcelain Industry Heritage" was successfully inscribed on the UNESCO World Heritage List. With the theme of "Creativity and Innovation, Empowering Industry," the expo focuses on internationalization, market orientation, and keeping pace with the times. At the opening ceremony, Chen Kelong, Secretary of the Jingdezhen Municipal Party Committee, sincerely invited friends from all over the world to come to Jingdezhen to experience the unique charm of the newly designated World Heritage city and to feel the cultural richness of China. Representatives of international sister cities, including Ahmet Kaya, Deputy Mayor of Iznik, Turkey, and Batyr Tursunov, Deputy Secretary-General of the Shanghai Cooperation Organization, also took the stage to deliver speeches and extend their best wishes for this year's ceramic expo. This year's ceramic expo will hold a series of major events, including the Jingdezhen Forum, a key industry investment promotion symposium, and a transnational ceramic procurement matchmaking conference. It has also planned a series of cultural and creative activities, such as the "SCO Window" themed exhibition. The event has attracted more than 2,500 guests from over 90 countries and regions around the world. More than 450 domestic and foreign enterprises, over 40 international brands, nearly 100 overseas buyers, and more than 150 domestic suppliers have gathered in Jingdezhen to experience firsthand the contemporary vitality of this millennium-old porcelain capital that continues to renew itself. 10月18日上午，2026中国景德镇国际陶瓷博览会盛大开幕，这是“景德镇手工瓷业遗存”成功列入《世界遗产名录》后举办的首届瓷博会，以“创意创新、赋能产业”为主题，聚焦国际化、市场化、时代化发展方向。 开幕式上，景德镇市委书记陈克龙诚挚邀请全世界的朋友来景德镇体验新晋世界遗产城市的独特魅力，感受人文中国。国际友好城市代表、土耳其伊兹尼克市副市长艾哈迈德·卡亚、上合组织副秘书长巴特尔·图尔苏诺夫等嘉宾也登台致辞，对本届瓷博会给予美好祝愿。 本届瓷博会将举办景德镇论坛、重点产业招商恳谈会、跨国陶瓷采购对接会等一系列重磅活动，还策划推出“上合之窗”主题展等一系列文化创意活动，吸引了全球90多个国家和地区2500多位海内外嘉宾。450余家国内外企业、40多个国际品牌、近百家境外采购商和150余家国内供应商齐聚景德镇，实地感受千年瓷都依然焕新的时代活力。 #Jingdezhen# #China# #Jiangxi# #赣出新精彩#