美味情缘

Artificial Analysis 公布苹果 iPhone 17 Pro 跑 8GB 以内本地 AI 性能排名_我的网站

A |     (ECNS) -- China aims to build more than 100 million metric tons of annual coal production reserve capacity by 2030, according to the 15th Five-Year Plan for Coal Industry Development, jointly released by the National Development and Reform Commission (NDRC) and the National Energy Administration (NEA) on Monday.    By 2030, China's capacity to ensure secure coal production and supply is expected to be further strengthened, while the structure and distribution of coal production will continue to improve. Large modern coal mines are expected to account for 87% of the country's total production capacity, according to the plan.    The plan also aims to significantly improve coal mine safety, green development, and the clean and efficient use of coal.    Smart mines are expected to account for 75% of total production capacity, while the development of a more diversified coal-based industrial structure will be accelerated.    Coal consumption is expected to peak during the plan period, alongside the establishment of a more robust mechanism for balancing supply and demand.    The plan places energy security alongside the green transition as key priorities.    Focusing on the development of a modern coal industry system, it outlines nine major tasks, including optimizing coal development and spatial distribution, accelerating industrial restructuring, and strengthening the production, supply, storage and sales system.    Other priorities include advancing the green and low-carbon transition, improving the clean and efficient use of coal, and promoting technological innovation. The plan also calls for stronger safety management, modernized industry governance, and a higher level of modernization across the sector.        (By Tang Yuxian)                    。

B |      8 月 26 日消息，Artificial Analysis 于 8 月 24 日发布博文，宣布携手 Liquid AI，发布面向手机端的 AI 跑分基准，重点关注 AI 模型在苹果 iPhone 17 Pro 上的表现。

C | 其中 Artificial Analysis 负责智能水平测试系统，Liquid AI 负责推理性能测试系统，双方选用 5 项基准测试：BFCL（Berkeley 函数调用排行榜）IFBench（指令遵循测试）AA-Omniscience（知识与认知相关测试）GPQA Diamond（高难度问答测试）MATH-500（数学问题测试）。测试模型对象为 4 bit 量化后体积不超过 8GB 的模型，示例包括 Gemma 4 E2B 和 LFM2-2.6B-Exp。当上下文长度限制为 16K tokens 后，平均基准测试得分最高的是 Nanbeige4.2-3B 和 LFM2.5-2.6B。当响应时间限制为最长 1 分钟时，LFM2.5-8B-A1B 排名第 1，随后是 LFM2-2.6B-Exp、Gemma 4 E4B（Non-reasoning）和 Granite 4.1 8B。Nanbeige4.2-3B 是 BOSS 直聘旗下的南北阁实验室推出，仅含 30 亿非嵌入参数（总参数约 40 亿），采用 Looped Transformer 架构，通过在不增加参数量的前提下复用 Transformer 层（循环两遍），提高模型的有效计算深度和容量。横轴表示“输出 256 个 token 所需的时间（秒）”，纵轴表示“上下文长度限制为 16K 个 token 时的平均基准测试得分”。“Nanbeige4.2-3B”的基准测试得分与“LFM2.5-2.6B”相当。在上下文长度限制为 16K 个 Token 时，平均基准测试得分中得分最高的是 Nanbeige4.2-3B 和 LFM2.5-2.6B。当响应时间限制为最长 1 分钟时，LFM2.5-8B-A1B 排名第 1，随后是 LFM2-2.6B-Exp、Gemma 4 E4B（Non-reasoning）和 Granite 4.1 8B。附上相关截图如下：。

Current article:http://71mnmr.zeitujiangshaopen.pics/list_e11/x9bb.html

Published on:02:15:05