Trends Artificial Intelligence
a year in the Internet business is like a dog year – equivalent to seven years in a regular person's life.’ At the time, the pace of change catalyzed by the internet was unprecedented. Consider now that Creators / bettors / consumers are taking advantage of global internet rails that are accessible to 5.5B citizens via connected devices; ever-growing digital datasets that have been in the making for over large language models (LLMs) that – in effect – found freedom with the November 2022 launch of OpenAI’s ChatGPT with its extremely easy-to-use / speedy user interface. In addition, relatively new AI company0 码力 | 340 页 | 12.14 MB | 5 月前3
 DeepSeek-V2: A Strong, Economical, and Efficient
Mixture-of-Experts Language Modelmodel characterized by economical training and efficient inference. It comprises 236B total parameters, of which 21B are activated for each token, and supports a context length of 128K tokens. DeepSeek-V2 enables training strong models at an economical cost through sparse computation. Compared with DeepSeek 67B, DeepSeek-V2 achieves significantly stronger performance, and meanwhile saves 42.5% of training costs only 21B activated parameters, DeepSeek-V2 and its chat versions still achieve top-tier performance among open-source models. The model checkpoints are available at h t t p s : / / g i t h u b . c o m0 码力 | 52 页 | 1.23 MB | 1 年前3
 Manus AI:Agent元年开启2025!3" Manus AI!Agent"#$ChatGPT%& #$% SAC NO. S0570519080006 | SFC NO. BQZ938 &'( SAC NO. S05701220801381 !"#$%&'() !"#$ • !"#$%&'()*AI+!"#$,-./012334%&'(56789:;<=>?@A BC%&'() • DEFGHI)*DEFGJKH @A+'BCDEFGHIJK,LMN OPQMR<"S>TUVWXY3 less structure more intelligence GZ[5\]^_`abcde_`fgchi_`jEc'k_` lm,no computer usecdeep researchccoding agent pqrstuvwxyz{|}~•G)€>•JK‚ƒ Manus,•P„…†‡ ˆ‰Š‹xG'B,LJKŒkF,•mP$ŒŽ4••‘JK’3“” §¨©ª°±²³{´µG SOTA œ=> • Manus AI ¶·fgG$%JKA+)€,¸¹!Lº»JK«Level 3¬°G-•¼½a‡¹T AI Ÿ >•)¾%‡ˆ¿ÀGÁ%ÂÃ,Ä Å'B|4ÆcÇ©ÈÉÊËcÌÍ•mÎÏJKG()A+> !"#$%Bloomberg*&'()6 Manus AI%2345 • ManusÐ!ÑÒÓ*GÔg<Õ5 • uvÖk5tAIןØAI0 码力 | 23 页 | 4.87 MB | 5 月前3
 清华大学 DeepSeek+DeepResearch 让科研像聊天一样简单,然后提供所有固定的参考文献。以下是需要修正的 五个示例模板和参考文献: 原始文本 修正后文本 Boullis A, Fassotte B, Sarles L,LognayG, Heuskin S, Vanderplanck M.Bartram S, Haubruge E, Francis F,Verheggen F(2017 Elevated Carbon Dioxide Concentration Alarm Signaling in Aphids. J Chem Ecol 43:164-171. Boullis A, Fassotte B, Sarles L, Lognay G, Heuskin S, Vanderplanck M.Bartram S, Haubruge E, Francis F.Verheggen FJ (2017) Elevated CarbonDioxide Concentration a one sentence recap of this data? 快速回顾一下 Create a visual chart, based on this data. 做一个视觉图表 What’s the main takeaway from this dataset? 找出最主要的信息 Can you explain this dataset like I’m 5 years old?0 码力 | 85 页 | 8.31 MB | 8 月前3
 PAI & TVM Meetup - Shanghai 20191116。Authored by @Hzfengsy 。 Intrinsics: tvm_load_matrix_sync tvm_mma_sync … “New Memory Scopes: wmma.matrix_a/b, accumulator 。Tensorization on warp level schedule Motivation matrix_b, row_major or col_moajor. 。 Visit the body of ComputeOp to get the indices of input matrices: inadexO, indexI 。 Compare the access indices with the axis/reduce_axis of ComputeOp n matrix_b [idx0 indexl K m :matrix a m k matrix_a k n :matrix_b n k matrix_b coLmajor row_major row_major col_ major Thread Index Unification 计算下全事业部0 码力 | 26 页 | 5.82 MB | 6 月前3
 清华大学第二弹:DeepSeek赋能职场• 向安玲(清华博士后、中央民大助理教授):人机共生之AI数据分析领域 • 马绪峰(清华博士后、同济大学助理教授):人机共生之文化艺术创作 成员及核心研究方向 赛事 奖项 2024 “AI4S Cup LLM 挑战赛” 大模型科学文献分析赛道 一等奖 2024 Kaggl e The Learni ng Agency Lab - PII Data Detecti on 金牌 金山办公2024中文文本智能校对大赛 英伟达NIM微服务 https://build.nvidia.com/d eepseek-ai/deepseek-r1 671B(全量模型) 网页版直接使用,支持API调用,注册送1000点数,免费体验。 微软Azure https://ai.azure.com 671B(全量模型) 需注册微软账户并创建订阅,免费部署,支持参数调节。 亚马逊AWS https://aws.amazon.com/c -on- aws 671B(全量模型) 需注册AWS账户,填写付款方式,免费部署。 Cerebras https://cerebras.ai 70B 邮箱注册,速度快,宣称比GPU方案快57倍。 Groq https://groq.com/groqclou d-makes-deepseek-r1- distill-llama-70b-available 70B 邮箱注册,速度快,但感觉比Cerebras弱一些。0 码力 | 35 页 | 9.78 MB | 8 月前3
 Deepseek R1 本地部署完全手册、量化⽅案、云端替代⽅ 案及完整671B MoE模型的Ollama部署⽅法。 核⼼提示: 个⼈⽤户:不建议部署32B及以上模型,硬件成本极⾼且运维复杂。 企业⽤户:需专业团队⽀持,部署前需评估ROI(投资回报率)。 ⼆、本地部署核⼼配置要求 1. 模型参数与硬件对应表 模型参 数 Windows 配置要求 Mac 配置要求 适⽤场景 1.5B - RAM: 4GB - GPU: 集成显卡/现代CPU 8GB (M1/M2/M3) - 存储: 5GB 简单⽂本⽣成、基础代 码补全 7B - RAM: 8-10GB - GPU: GTX 1680(4-bit量 化) - 存储: 8GB - 内存: 16GB(M2 Pro/M3) - 存储: 8GB 中等复杂度问答、代码 调试 14B - RAM: 24GB - GPU: RTX 3090(24GB VRAM) - 存储: 复杂推理、技术⽂档⽣ 成 32B+ 企业级部署(需多卡并联) 暂不⽀持 科研计算、⼤规模数据 处理 2. 算⼒需求分析 模型 参数规 模 计算精 度 最低显存需 求 最低算⼒需求 DeepSeek-R1 (671B) 671B FP8 ≥890GB 2*XE9680(16*H20 GPU) DeepSeek-R1-Distill- 70B 70B BF16 ≥180GB 4*L200 码力 | 7 页 | 932.77 KB | 8 月前3
 Google 《Prompt Engineering v7》can be complicated. Many aspects of your prompt affect its efficacy: the model you use, the model’s training data, the model configurations, your word-choice, style and tone, structure, and context process. Inadequate prompts can lead to ambiguous, inaccurate responses, and can hinder the model’s ability to provide meaningful output. You don’t need to be a data scientist or a machine learning challenges you can face while crafting prompts. Prompt engineering Remember how an LLM works; it’s a prediction engine. The model takes sequential text as an input and then predicts what the following0 码力 | 68 页 | 6.50 MB | 6 月前3
 清华大学 普通人如何抓住DeepSeek红利deepseek.com Z u N e P 6 7 K w S v L C q Y 4 Y V 1 T 8 0 u m B k k m O x d k C i y K r j i 6 n p Y d O w t v B 4 G 0 G p y 8 U I q e T 9 M Natural Questions等)来生成问题。可以从多个数据集中组 合问题,以达到10万个的问题数量。 这 些 数 据 集 包 含 大 量 的 问 答 对 , 例 如 使 用 d a t a s e t s 库 (Hugging Face的datasets库)来加载SQuAD数据集 (Stanford Question Answering Dataset),这个数据集 是一个著名的问答数据集,基于维基百科数据生成,并且数 添加主观引导(如“你认为哪种对? ”) 通用模型 需拆分问题,逐步追问 “先解释电车难题的定义,再对比 两种伦理观的差异 ” 一次性提问复杂逻辑 任务需求与提示语策略 "以下是某论文结论:'神经网络模型A优于传统方法B'。 请 验 证 : ① 实验数据是否支持该结论; ② 检查对照组设置是否存在偏差; ③ 重新计算p 值并判断显著性。" "为降低物流成本,现有两种方案: ①自建区域仓库(初期投入高,长期成本低)0 码力 | 65 页 | 4.47 MB | 8 月前3
 TVM@AliOSFOData … NHWC L2 da … FL2 da Alios TVM @ ARM CPU INT8 TVM /QNNPACK Speed Up @ Mobilenet V2 @ rasp 3b+ AARCH64 35 2.38 2.00 1.67 1.67 1.64 1.60 1.66 155 157 157 150 143 40 120 100 080 060 040 020 000 TI.30 1.15 TVM / QNNPACK Speed Up @ Mobilenet V2 @ rasp 3b+ AARCH64 了 130 1.35 1.33. 1.15 116 111 Depthwise Convolution Workload Performance Alios TVM @ ARM CPU INT8 Performance Comparison @ rasp 3b+ AARCH64 aoo0 8.87 sm ao 7m am sm 3.83 om ao 2.08 2 1.610 码力 | 27 页 | 4.86 MB | 6 月前3
共 30 条
- 1
 - 2
 - 3
 













