Prompt还是和之前的思路一样，0.18是核心筹码区，在价格跌破这个关键筹码峰之前，都可能是持续的洗盘； 0.36的筹码峰被跌破后，当下价格开始进入0.26～0.31的密集筹码区，新币上线后持续的回调基本上已经是常态了，相对比与Kaito，Prompt表现其实还行。 后续的思路，如果价格继续回调跌破0.27，那么代表上所后的多头趋势已经暂停，目标会在0.188的核心筹码区； 如果0.27这里能够持续一段时间的缩量震荡，就还有希望出现对于0.3的反弹； 我知道很多人比较看好Wayfinder，但我依然建议，看好不代表立即满仓梭哈，新币上所后，市场自发的清洗并集中筹码才是后续行情的基础； 如果还是想买，建议分批缓慢建仓，只做现货就好，当下整体市场的流动性依然很差，BTC的统治率仍在新高，很多项目不论是叙事还是应用其实都很好，但池子里缺水，水位自然不会高； 降低预期，不要把投资和投机混为一谈，看好一个项目就慢慢吸筹，没事去X上cx也行，千万不要因为看到期货市场全是空单，就尝试去赌暴空行情。 毕竟对于当下，降低自身杠杆率也是生存的必要条件！

🎶💡 How do YOU prompt SOUNDRAW? 💡🎶 With SOUNDRAW, making fire beats is easy—just tell us what you need! 🔥 Here’s how you can guide your next track: ✨ By Genre: Hip-Hop, Trap, Drill, R&B? You decide. 🎤 ✨ By Mood: Hype, dark, chill, emotional? Set the vibe. 🎭 ✨ By Instruments: More 808s? Guitar-driven? Synth-heavy? Your call. 🎸🎹 ✨ By Theme: Making content for weddings or need a horror soundtrack? We got you 🚀 ✨ By Tempo: Slow jam or fast-paced heat? We match your flow. ⏳ No matter your style, SOUNDRAW helps you create the perfect beat. Try it out and drop a 🎵 in the comments if you’re feeling inspired! #MadeWithSoundraw# #AIBeats# #MusicProduction# #AImusic# #madewithsoundraw#

Google 发布了一份 68 页的提示词工程 这份《Prompt Engineering》白皮书系统介绍了如何为大型语言模型（LLM）设计高效提示（prompt）。 1、解释了 LLM 的工作原理和输出配置，包括温度、Top-K、Top-P等采样参数的作用。 2、介绍了多种提示技巧，如零样本（zero-shot）、单样本（one-shot）、少样本（few-shot）、系统提示、角色提示、上下文提示、逐步推理（Chain of Thought）、自洽性（Self-consistency）、思维树（Tree of Thoughts）和ReAct（推理与行动结合）等。 3、文档还涵盖了代码提示的应用，包括编写、解释、翻译和调试代码。此外，还介绍了自动提示工程（APE）方法，用于自动生成和优化提示。最后，总结了一系列最佳实践，如提供示例、简化语言、明确输出要求、使用变量、控制输出格式、记录提示版本等，强调提示工程是一个持续迭代和优化的过程。 墨问用户整理了一份 pdf，这里下载

AI music generation just hit its inflection point. Microsoft Copilot can now turn text prompts into full music tracks. It's completely free & insanely good. Here's everything you need to know and a quick tutorial on how to create your own song:

Everything is LEGO! 🧱 你可以用下面的 prompt 生成乐高风格的头像和logo： convert this image into a lego { "style_name": "LEGO Plastic Toy Render", "input_mode": "text_prompt_or_image", "output_format": "3D LEGO-style render", "instructions": { "style_description": "Transform the input into a LEGO-style plastic model with blocky structure, smooth round LEGO brick edges, and toy proportions.", "materials": [ "shiny plastic (ABS)", "rounded LEGO studs labeled 'LEGO'", "seamless LEGO brick construction" ], "geometry_rules": { "form": "simplified, rounded block geometry", "proportions": "cute, stylized, exaggerated like a toy" }, "color_rules": { "palette": "saturated solids (greens, blues, whites, yellows)", "contrast": "high color contrast between different regions" }, "camera_and_render": { "angle": "3/4 perspective view (slightly above and to the side)", "background": "solid pastel background (light blue recommended)", "lighting": "soft studio light with mild reflections", "depth_of_field": "none (everything sharp)", "resolution": "high resolution, suitable for product visualization" } }, "prompt_template": "Render a {input} as a LEGO-style 3D toy. Make it blocky, with realistic LEGO plastic textures and studs. Use toy proportions. Place on a light pastel background with soft lighting. Match the style of an official LEGO sculpture render." }

While US President Donald Trump’s sweeping new tariffs have sent shockwaves through global markets and prompted retaliatory threats from major economies, Gulf Cooperation Council (GCC) nations, including the United Arab Emirates, appear likely to escape the worst direct impacts, analysts said on Thursday. Read more: https://t.co/hQduL6RDbA

Me: I have a new idea for an app! *Opens Lovable* *Opens Replit* Prompts the tools with app idea. 10+ back and forth edits and checkpoints Then…. Agent: *error* Agent: *error* Agent: *error* Agent: *error* Agent: *error* 💔

我是如何高效翻译 65 页 Google 官方提示工程白皮书 PDF 文件的 前几天翻译 Google 官方提示工程白皮书 PDF Google 官方提示工程 (Prompt Engineering)白皮书完整翻译和 PDF 下载 的时候，尝试了一些自动化的方法来提升效率，分享一些我翻译 PDF 的经验和心得。 首先我个人比较排斥保持排版的翻译方式，因为 PDF 在翻译后，文字长度不一致会导致排版很难看，忽大忽小；另外翻译的时候由于布局导致的文字被强行分割，导致上下文不完整会影响翻译质量。 我在翻译 PDF 时，会把 PDF 先转成 Markdown，再基于 Markdown 去翻译，翻译好了基于翻译好的 Markdown 再去重新生成 PDF，对于文字、表格、图片都能很好的保留，主要缺点是布局格式保留的不是很好，不过我翻译的通常是文字图表为主的，所以影响不大。 🧵

Convertir imágenes es solo un abre boca. Eventualmente, tendremos conversión de videos, filtros para reels, filtros en vivo. Convertiremos todos los libros a films con prompts en el estilo que queramos, con los actores que queramos, en todos los idiomas. Solo hace falta más cómputo, y la nueva generación de NVIDIA está siendo instalada en Centros de Datos en todo el mundo. Acá un ejemplo, el trailer de Encuentros Cercanos del Tercer Tipo (1977, Steven Spielberg), al estilo Ghibli.

Launch your own AI-generated NFTs instantly with 7007 @Lab7007, the Premier AI Generated NFT Launchpad powered by @OraProtocol 🔹 Create with a single prompt 🔹 Pre-reveal unboxing experience 🔹 Bonding curve for trading 🔹 $7007 powers AIGC; the content of the future 👉 https://t.co/QhF8Bly7mO #Sponsored#

This would be an awesome feature in @Zapier! The ability to convert a mapped variable to be an AI variable in a Zap step field that is a dropdown. The idea being the user could add a prompt to instruct the AI about which dropdown option to pick. Example: Map a variable that has a State abbreviation (e.g. CA) to a dropdown field that has a list of States (e.g. California), where each State in the dropdown list has an internal ID (e.g. 12345). So the AI is able to select the correct dropdown option, instead of the user having to translate CA to 12345. It would replace the need for certain lookup tables and paths.

AIWayfinder 0.25%代币空投活动- 截止时间4.14号 @AIWayfinder Wayfinder是Parallel开发的AI工具，母公司ParallelPRIME 总融资8500万美金 活动信息 ​ 🎫代币分配：Wayfinder代币为 $Prompt 总供应量10亿枚 $PROMPT 的0.25%（250万枚）将分配给 #Kaito# Yap社区， 另外在项目方的代币模型中有50%是分配给社区的 ​ 🚩资格条件： 活动开始时Yaps数量≥100的用户优先获得高比例分配 其他用户完成指定任务可获少量代币。 🕐截止时间：​2025年4月14日 #Wayfinder# #PROMPT# #空投# #Airdrop# #KAITO#

I hooked up this little ESP32-CAM to my wifi. It takes a picture every 30s and sends it to flash-8B with a prompt like 'is there a person in this image?'. If the model says yes (with structured outputs), it turns on an LED. If I left this on for a month it would cost about $1. Obviously this is the worst security cam ever, BUT I think it's a compelling demo. You can put some 'intelligence' in the loop of a hobby project with <$10 in hardware and fractions of a cent per API call. You can 'program' it in natural language like 'if there is an orange cat playing with my plant'. And with LLM help, modifying the code is possibly as easy as 'make this send me a discord notification if the alarm is triggered twice in an hour'. Anyway, now that I know this thing works it has other projects it is destined for. But cool to know how easy it is to whip this up :)

Due to the stable increase in network hashrate, centralized exchanges and services should lower their confirmations for $ETC deposits. For an improved user experience, 100 confirmations should be sufficient #PoW# security. Thank you for your prompt response - #EthereumClassic# 💚

I wanted to try and make an AI short film that had some emotional elements but was also "realistically surreal." All video generated with @Hailuo_AI and you can see some prompts in the description #shortfilm# #ostrich# #aivideo# https://t.co/6PFXdx0Tvo