Dave Davies
@onlineinference.com
๐ค 468
๐ฅ 137
๐ 153
I'm an SEO who gets to work with people who teach machines how to think.
โ Gemini now supports structured outputs ๐งฉ - letting you define JSON Schemas for predictable, parsable results. ๐ค Huge for agents & data extraction. Think: cleaner automation, safer outputs, and I'm about to have some SEO fun. ๐ Full announcement ๐
ai.google.dev/gemini-api/d...
loading . . .
Structured Outputs ย |ย Gemini API ย |ย Google AI for Developers
Learn how to generate structured JSON output with the Gemini API.
https://ai.google.dev/gemini-api/docs/structured-output
about 20 hours ago
0
0
0
๐ผ Rolling out in Search Console Insights: Query groups. Google now clusters similar search terms (like misspellings & variants) to show broader intent trends. ๐ Link to the announcement from Google in the comments.
15 days ago
1
3
0
CoreWeave is acquiring Monolith AI ๐ค A serious combo: Monolith brings ML for hardcore physics problems, CoreWeave brings the AI infra. Expect faster ๐, smarter ๐ง industrial R&D. Something we desperately need to carry us into the future.
www.coreweave.com/news/corewea...
loading . . .
CoreWeave to acquire Monolith AI, combining AI cloud and ML engineering tools to accelerate innovation in manufacturing, automotive, and aerospace.
https://www.coreweave.com/news/coreweave-to-acquire-monolith-expanding-ai-cloud-platform-into-industrial-innovation
about 1 month ago
0
0
0
I am SUPER stoked! Chrome is finally getting agentic. ๐ค Google's latest update adds Gemini for AI summaries, multi-tab context, and task automation (yes, itโll actually DO stuff for you). The browser is officially becoming an assistant. ๐
blog.google/products/chr...
loading . . .
Go behind the browser with Chromeโs new AI features
Google Chrome is getting upgraded with the latest AI to make it safer, smarter and more useful
https://blog.google/products/chrome/new-ai-features-for-chrome/
about 2 months ago
0
1
0
91% of SEOs say clients/management now ask about AI search visibility - but only 35% have a strategy. It's early days, but leadership is watching. ๐ Survey via
@aleyda.bsky.social
๐
hub.seofomo.co/surveys/stat...
loading . . .
The State of AI Search Optimization - 2025 Edition
Learn about the state of AI search optimization with the results of the SEOFOMO AI Search Optimization Survey, taken by +200 Senior SEO specialists.
https://hub.seofomo.co/surveys/state-ai-search-optimization/?utm_source=convertkit&utm_medium=email&utm_campaign=%F0%9F%93%8A+Top+SEO+&+AI+Search+News+of+the+Week+%5BSEOFOMO%2C+September+14%2C+2025%5D+-+18980554=
about 2 months ago
0
1
1
Today, I'm thinking about how to create advanced prompts to monitor AI Mode results that set the stage for what the final question in a back-and-forth might be, where context matters more than the final query.
loading . . .
2 months ago
0
0
0
I hate it when someone is essentially right, but you still want them to lose.
wandb.ai/byyoung3/ml-...
loading . . .
X and xAI Sue Apple and OpenAI Over Alleged AI Monopoly
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/X-and-xAI-Sue-Apple-and-OpenAI-Over-Alleged-AI-Monopoly--VmlldzoxNDEzNzkwMg
3 months ago
0
0
0
Google wants you in AI Mode faster โฉ: an โAsk Anythingโ box is showing in AI Overviews, sending users straight to AI results. ๐
www.seroundtable.com/google-ai-ov...
loading . . .
Google AI Overview Ask Anything Box Leads To AI Mode
Google is now testing adding an "Ask Anything" box within the AI Overviews, and when you type in that box and click search, it takes you into the Google AI Mode results.
https://www.seroundtable.com/google-ai-overview-ask-anything-box-to-ai-mode-39993.html
3 months ago
0
1
0
LLMs, lost attribution & agentic AI. ๐๏ธ I talked w/ Search Engine Land ahead of my SMX Advanced keynote on how SEO is evolving & the weirdest SEO gotcha Iโve hit yet. ๐
searchengineland.com/dave-davies-...
loading . . .
Dave Davies on LLM content SEO shortcuts, attribution loss, and agentic AI
SMX Advanced keynote speaker Dave Davies on agentic AI, LLM pitfalls, weird tech gotchas, and why generative engines are the future.
https://searchengineland.com/dave-davies-smx-advanced-2025-interview-456538
5 months ago
1
2
0
AI agents that read, summarize, and document a GitHub repo โ end-to-end automation using CrewAI +
@weightsbiases.bsky.social
Weave for observability ๐. A great demo and tutorial on multi-agent orchestration ๐
wandb.ai/byyoung3/cre...
loading . . .
Building a Github repo summarizer with CrewAI
A hands-on guide to building a fully automated GitHub documentation system using CrewAI for multi-agent coordination and Weave for real-time debugging and observability.
https://wandb.ai/byyoung3/crewai_git_documenter/reports/Building-a-Github-repo-summarizer-with-CrewAI--VmlldzoxMjY5Mzc5Ng
6 months ago
0
0
0
ChatGPT just got way more useful for devs: ๐Deep Research now connects to GitHub. You can query repos ๐, analyze APIs, and break down code structures. o4-mini fine-tuning ๐๏ธ & GPT-4.1 nano access expanded too - w/ verification required. Details ๐
wandb.ai/byyoung3/ml-...
loading . . .
ChatGPT Deep Research adds GitHub Connector for Code Analysis and o4-Mini fine-tuning
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/ChatGPT-Deep-Research-adds-GitHub-Connector-for-Code-Analysis-and-o4-Mini-fine-tuning---VmlldzoxMjY5MzczOQ
6 months ago
0
0
0
๐
www.prnewswire.com/news-release...
loading . . .
CoreWeave Completes Acquisition of Weights & Biases
/PRNewswire/ -- CoreWeave, Inc. (Nasdaq: CRWV) today announced that it has completed its acquisition of Weights & Biases. The strategic combination strengthens...
https://www.prnewswire.com/news-releases/coreweave-completes-acquisition-of-weights--biases-302445966.html
6 months ago
0
1
0
6 months ago
0
0
0
Qwen3 just crushed AIME 2024 with 66.7% - 3x DeepSeek R1. ๐คฏ Fine-tune it with Unsloth, evaluate with
@weightsbiases.bsky.social
Weave, and toggle its reasoning mode. Wildly flexible open-source LLM. Read more ๐
wandb.ai/byyoung3/Gen...
loading . . .
How to fine-tune and evaluate Qwen3 with Unsloth
This article provides a comprehensive guide to fine-tuning, evaluating, and deploying the Qwen3 language model, emphasizing its flexibility, performance, and unique reasoning-toggle feature. .
https://wandb.ai/byyoung3/Generative-AI/reports/How-to-fine-tune-and-evaluate-Qwen3-with-Unsloth---VmlldzoxMjU3OTI0Ng
6 months ago
0
0
0
New papers from @SFResearch & @Tsinghua_Uni suggest RL in LLMs may be overrated. ๐Simple filtering > complex RL ๐ชRLVR โ new reasoning These finding are covered on the @weights_biases blog, and may be a game-changer for post-training strategies. ๐
wandb.ai/byyoung3/ml-...
loading . . .
New studies uncover interesting findings for reasoning models
Discover how two recent studies challenge conventional reinforcement learning in LLM reasoning - revealing that simple data filtering can rival complex methods and that RLVR may only optimize known ab...
https://wandb.ai/byyoung3/ml-news/reports/New-studies-uncover-interesting-findings-for-reasoning-models--VmlldzoxMjQ1NDE3Mw
7 months ago
0
0
0
Most agents today canโt talk to each other. A2A changes that. Agent2Agent is an open protocol for secure, cross-platform agent collaboration. Think: LangGraph x CrewAI in one workflow. No glue code. Full write-up + tutorial over on
@weightsbiases.bsky.social
blog ๐
wandb.ai/byyoung3/Gen...
loading . . .
How the Agent2Agent (A2A) protocol enables seamless AI agent collaboration
The Agent2Agent (A2A) protocol is an open standard that enables autonomous AI agents to securely discover, communicate, and collaborate across platforms. Learn how it works, its core components, and h...
https://wandb.ai/byyoung3/Generative-AI/reports/How-the-Agent2Agent-A2A-protocol-enables-seamless-AI-agent-collaboration--VmlldzoxMjQwMjkwNg
7 months ago
0
1
0
๐Big congrats to
@cohere.com
on Embed 4 โ a new multimodal embedding engine for enterprise AI. It handles ๐ 128k token docs, ๐ 100+ languages, and real-world mess like legal PDFs & product decks. Built for ๐ค RAG, agents, and cross-lingual search.
wandb.ai/byyoung3/ml-...
loading . . .
Cohere Releases Embed 4: Multimodal Embedding Engine for Enterprise AI
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/Cohere-Releases-Embed-4-Multimodal-Embedding-Engine-for-Enterprise-AI--VmlldzoxMjMxOTE5Ng
7 months ago
0
0
0
SEO in the agentic era isnโt theory - itโs here. Over on
@searchengineland.bsky.social
I explore ๐ the impact agents are having on how (and where) we optimize, and what that means for your strategy. I even outline an agentic system I'm working on. Enjoy ๐
searchengineland.com/ai-agents-di...
loading . . .
How AI agents are revolutionizing digital marketing
Explore how AI agents autonomously solve problems, enhance personalization and enable next-gen marketing strategies using agentic frameworks.
https://searchengineland.com/ai-agents-digital-marketing-448342
7 months ago
0
1
0
I just published a GPT-4.1 quickstart using the OpenAI API over on the @weights_biases blog, including a Colab to get going fast. It includes full W&B Weave integration so you can track everything out of the gate. ๐
wandb.ai/onlineinfere...
loading . . .
GPT-4.1 Python quickstart using the OpenAI API
Getting set up and running GPT-4.1 on your machine in Python using the OpenAI API. Made by Dave Davies using Weights & Biases
https://wandb.ai/onlineinference/gpt-python/reports/GPT-4-1-Python-quickstart-using-the-OpenAI-API--VmlldzozODI1MjY4
7 months ago
0
0
0
How long until we start talking about ACO (Agent Card Optimization)? ๐ค
google.github.io/A2A/#/docume...
loading . . .
Agent2Agent Protocol
An open protocol enabling communication and interoperability between opaque agentic applications.
https://google.github.io/A2A/#/documentation
7 months ago
1
0
0
๐Retailers are using AI agents ๐ค to do a lot more than recommend products. This post from
@weightsbiases.bsky.social
breaks down a smart LLM-powered system for triaging customer emails and building real-time recs via vector search. ๐
wandb.ai/byyoung3/Gen...
loading . . .
AI agents in retail and e-commerce
This article explores how AI agents are transforming retail by automating customer interactions, optimizing decision-making, and enhancing product recommendations using LLM-driven vector search.
https://wandb.ai/byyoung3/Generative-AI/reports/AI-agents-in-retail-and-e-commerce---VmlldzoxMTMwMjY2Ng
7 months ago
0
0
0
LLMs donโt need to think + talk in the same pass. Retrieval Augmented Thinking (RAT) ๐ splits reasoning from response - boosting transparency + control. DeepSeek, Claude, GPT-4o all in the mix. Code + Weave traces๐
wandb.ai/byyoung3/Gen...
loading . . .
What is Retrieval Augmented Thinking (RAT) and how does it work?
Retrieval Augmented Thinking (RAT) separates AI reasoning from response generation, improving efficiency, interpretability, and customization by using one model for structured thought and another for ...
https://wandb.ai/byyoung3/Generative-AI/reports/What-is-Retrieval-Augmented-Thinking-RAT-and-how-does-it-work---VmlldzoxMTc3OTg1Nw
7 months ago
0
0
0
Amazon Nova Reel 1.1 ๐ฅ now supports 2-minute, multi-shot videosโwith automated and manual modes. Great flexibility, now on AWS Bedrock.
wandb.ai/byyoung3/ml-...
loading . . .
Amazon Nova Reel 1.1 Expands Video Generation Capabilities
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/Amazon-Nova-Reel-1-1-Expands-Video-Generation-Capabilities--VmlldzoxMjE4NTgxNQ
7 months ago
0
0
0
Llama 4 ๐ฆ isnโt just openโitโs competitive.
@weightsbiases.bsky.social
tested it head-to-head with GPT-4o on ChartQA using Weave Evaluations. Maverick scored higher on multimodal accuracy and costs a fraction to run. Here's how they did it + the code ๐
wandb.ai/byyoung3/Gen...
loading . . .
Running inference and evaluating Llama 4 in Python
Deploy Llama 4 locally or via API with Python scripts. We test multimodal performance against GPT-4o on ChartQA and show how to debug and compare results using Weave.
https://wandb.ai/byyoung3/Generative-AI/reports/Running-inference-and-evaluating-Llama-4-in-Python--VmlldzoxMjE2NTYxNA
7 months ago
0
0
0
๐ถโโ๏ธ Gemini 2.5 Pro Experimental is a big step up. ๐ง Multimodal input, 1M-token context, native code execution, and better math/code reasoning.
@weightsbiases.bsky.social
evaluated it against Flash 2.0 on AIME problems using Weave; in a tutorial you can do yourself. Results? ๐
wandb.ai/byyoung3/Gen...
loading . . .
Evaluating the new Gemini 2.5 Pro Experimental model
Gemini 2.5 Pro Experimental is Google's most advanced AI model to date, featuring multimodal input support, a massive 1 million-token context window, and the ability to solve complex problems. .
https://wandb.ai/byyoung3/Generative-AI/reports/Evaluating-the-new-Gemini-2-5-Pro-Experimental-model--VmlldzoxMjAyNDMyOA
8 months ago
0
1
0
8 months ago
0
0
0
Want your OpenAI Agent to explore files, analyze data, and log everything with Weave? ๐ง This walkthrough shows how to hook agents into external tools using the Model Context Protocol (MCP). ๐
wandb.ai/byyoung3/Gen...
loading . . .
Getting Started with MCP using OpenAI Agents
A practical walkthrough for building OpenAI Agents that use the Model Context Protocol (MCP) to access tools, files, and trace data via Weave.
https://wandb.ai/byyoung3/Generative-AI/reports/Getting-Started-with-MCP-using-OpenAI-Agents---VmlldzoxMjAwNzU5NA
8 months ago
0
1
0
I'm actually finding it worse. Anyone else? Am I just using it wrong? :D
add a skeleton here at some point
8 months ago
1
0
0
ARC-AGI-2 is here. ๐ฅ The benchmark that flips the scriptโeasy for humans, brutal for AI. ๐ค If you're chasing AGI, this is where you prove your model can reason ๐ค, not just pattern-match. ๐ค $1M in prizes. Hosted on Kaggle.
wandb.ai/byyoung3/ml-...
loading . . .
Arc Prize unveils ARC-AGI-2
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/Arc-Prize-unveils-ARC-AGI-2--VmlldzoxMTk2NzgzMQ
8 months ago
0
0
0
MCP is like a USB port for AIโletting LLMs interact with external data, tools, and APIs without custom integrations. Anthropicโs new protocol makes AI applications more scalable, interoperable, and easier to manage. Enjoy a hands-on guide for building an MCP server.๐
wandb.ai/byyoung3/Gen...
loading . . .
The Model Context Protocol (MCP): A Guide for AI integration
This guide explores how MCP standardizes AI interactions with external tools and data sources, enabling more efficient AI context integrations.
https://wandb.ai/byyoung3/Generative-AI/reports/The-Model-Context-Protocol-MCP-A-Guide-for-AI-integration--VmlldzoxMTgzNDgxOQ
8 months ago
1
4
0
๐จ AI agent hijacking is a growing threat. US AISI's latest research using AgentDojo found that even top models like Claude 3.5 Sonnet were tricked 81% of the time with new attack strategies. Security evaluations need to evolveโattackers already are. Read the full story ๐
wandb.ai/byyoung3/ml-...
loading . . .
US AISIโs findings on agent hijacking evaluations
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/US-AISI-s-findings-on-agent-hijacking-evaluations---VmlldzoxMTgyODAyNw
8 months ago
0
0
0
๐ "A USB-C port for AI" โ that's how Anthropic describes the Model Context Protocol (MCP). MCP standardizes how LLMs connect to tools & data, making AI assistants more powerful. Could this be the ODBC moment for AI? Learn more ๐
wandb.ai/onlineinfere...
loading . . .
The Model Context Protocol (MCP) by Anthropic: Origins, functionality, and impact
Explore Anthropic's Model Context Protocol (MCP), a new open standard that unifies AI models with external tools and data for smarter, context-rich applications.
https://wandb.ai/onlineinference/mcp/reports/The-Model-Context-Protocol-MCP-by-Anthropic-Origins-functionality-and-impact--VmlldzoxMTY5NDI4MQ
8 months ago
0
0
0
But don't worry, Elon will let us all know if there's a conflict of interest.
www.msn.com/en-ca/news/w...
8 months ago
0
0
0
Google is rolling out AI Mode to Google One AI Premium subscribers in the US. ๐
x.com/sundarpichai...
8 months ago
0
0
0
For anyone interested in getting started with the new GPT-4.5 via the API, I just updated my GPT-4o quickstart to 4.5. You can find it on the
@weightsbiases.bsky.social
blog here ๐
wandb.ai/onlineinfere...
loading . . .
GPT-4.5 Python quickstart using the OpenAI API
Getting set up and running GPT-4.5 on your machine in Python using the OpenAI API. Made by Dave Davies using Weights & Biases
https://wandb.ai/onlineinference/gpt-python/reports/GPT-4-5-Python-quickstart-using-the-OpenAI-API--VmlldzozODI1MjY4
9 months ago
0
2
0
๐ Want to build AI agents that can research, analyze, and summarize in real-time?
@weightsbiases.bsky.social
takes a look at CrewAI - a framework for multi-agent systems - and walks through how to build an AI agent. Full tutorial here ๐
wandb.ai/byyoung3/Gen...
loading . . .
Tutorial: Building AI agents with CrewAI
This guide explores how AI agents, powered by CrewAI, automate complex tasks with minimal human input by integrating adaptive workflows, real-time data analysis, and iterative improvements.
https://wandb.ai/byyoung3/Generative-AI/reports/Tutorial-Building-AI-agents-with-CrewAI--VmlldzoxMTUwNTA4Ng
9 months ago
0
3
0
Google Ads almost tricked me. ;)
9 months ago
0
0
0
๐จ AI safety just got a lot easier.
@weightsbiases.bsky.social
Weave Guardrails helps devopers detect toxicity, bias, hallucinations & more - before bad outputs hit users. Pre-built scorer models + real-time safeguards = better AI apps. Try it now ๐
wandb.ai/site/guardra...
loading . . .
Guardrails
Guardrails are pre-built scorers that help you reduce unwanted behaviorโlike hallucinations and biasโand build safer AI applications.
https://wandb.ai/site/guardrails/
9 months ago
0
0
0
It was a good evening. :)
www.nhl.com/video/can-us...
loading . . .
McDavid's 4 Nations Face-Off Championship OT winner | NHL.com
Connor McDavid gets the feed from Mitchell Marner out in front, then roofs it past Connor Hellebuyck to give Canada the 3-2 win in overtime, winning the 4 Nations Face-Off Championship
https://www.nhl.com/video/can-usa-mcdavid-scores-goal-against-connor-hellebuyck-6369130771112
9 months ago
0
3
0
๐ฐ๐ท South Korea has accused DeepSeek of sharing user data with ByteDance. Their app skyrocketed ๐ in popularity, but is under fire ๐ for potential data transfers. It's already banned in Korean app stores, and more countries are watching closely. Full story ๐
wandb.ai/byyoung3/ml-...
loading . . .
Is DeepSeek partnering with ByteDance?
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/Is-DeepSeek-partnering-with-ByteDance---VmlldzoxMTQyODUyNw
9 months ago
0
0
0
๐ xAIโs Grok-3 is here, and is topping benchmarks. โ Trained with 100K H100 GPUs in 122 days โ #1 in Chatbot Arena, beating Gemini-2 Flash & DeepSeek-V3 โ 93 on AIME (math), 79 on LCB (coding) Self-correcting, reasoning, and already in Deep Search & X Premium+ More here ๐
wandb.ai/byyoung3/ml-...
loading . . .
xAI Unveils Grok-3: A SOTA reasoning model?
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/xAI-Unveils-Grok-3-A-SOTA-reasoning-model---VmlldzoxMTQwNzU1Mw
9 months ago
0
0
0
reposted by
Dave Davies
Tibor Blaho
9 months ago
Elon Musk announced that Grok 3, which he described as the โsmartest AI on Earthโ, will be released with a live demo on Monday, February 17, 2025, at 8:00 PM PT
1
3
1
๐ Can we push GPT-4oโs reasoning beyond state-of-the-art? Brett Young tested fine-tuning and budget forcing on AIME math problems and it turns out, we can! The full breakdown, and code to run your own experiments can be found on the
@weightsbiases.bsky.social
blog here ๐
wandb.ai/byyoung3/Gen...
loading . . .
Training GPT-4o to reason: Fine-tuning vs budget forcing
Can fine-tuning and budget forcing improve GPT-4oโs reasoning? We test structured datasets and inference-time techniques to boost multi-step problem-solving.
https://wandb.ai/byyoung3/Generative-AI/reports/Training-GPT-4o-to-reason-Fine-tuning-vs-budget-forcing--VmlldzoxMTMzMjYyMw
9 months ago
0
0
0
WTF? I'm not sure what's more pathetic, the lack of any spine from Google or that there are people who were triggered by the notion of people learning about history.
add a skeleton here at some point
9 months ago
0
2
0
The future is coming VERY VERY fast.
add a skeleton here at some point
9 months ago
0
2
0
Well, they did it. OpenAI got $200/mth out of me. :D
9 months ago
0
1
0
I'm super-excited by the new ๐ Apple researcher show that self-play RL can train AI to drive without human data! Their model, GIGAFLOW, ran 1.6B km in sim, outperforming specialist models on real-world benchmarks. This may dramatically speed up the road to AV. ๐
wandb.ai/byyoung3/ml-...
loading . . .
Does self-play RL solve self-driving?
Publish your model insights with interactive plots for performance metrics, predictions, and hyperparameters. Made by Brett Young using Weights & Biases
https://wandb.ai/byyoung3/ml-news/reports/Does-self-play-RL-solve-self-driving---VmlldzoxMTI1MDQ5Ng
9 months ago
0
1
0
๐ช Can a decoding trick make a smaller open model match OpenAIโs o1-preview? Budget forcing - forcing an LLM to "keep thinking" - boosted s1-32Bโs accuracy 13% at test time with no extra training. Hereโs what Brett Young found, published on
@weightsbiases.bsky.social
๐
wandb.ai/byyoung3/Gen...
loading . . .
Budget forcing s1-32B: Waiting is all you need?
We test whether budget forcing - a simple test-time intervention - can significantly boost the reasoning accuracy of s1-32B, potentially enabling smaller models to rival closed-source giants like Open...
https://wandb.ai/byyoung3/Generative-AI/reports/Budget-forcing-s1-32B-Waiting-is-all-you-need---VmlldzoxMTIzNzkzOA
9 months ago
0
0
0
9 months ago
0
0
0
A reminder for my US friends: Less than 1% of fentanyl ๐ entering the US ๐บ๐ธ is coming from Canada ๐จ๐ฆ and only 1.5% of illegal migrants entering the US are coming from Canada each year. (Based on data from 2024) 1/3
9 months ago
2
0
0
Load more
feeds!
log in