[ 67722840.000 ] posts.x: ↻ @sweis: CLAUDE FOUND THE SEEDS  |   [ 67718940.000 ] posts.x: ↻ @VikParuchuri: This week, we launched Indra, a 1M parameter OCR model that's 15x faster than Tesseract on CPU and much more accurate. I…  |   [ 67707240.000 ] posts.x: This looks promising!  |   [ 67702800.000 ] posts.x: ↻ @awilkinson: "As long as I've got my health, my millions of dollars, my gold house, and my rocket car, I don't need anything else." –Che…  |   [ 67702440.000 ] posts.x: @GeoffreyHuntley What is Bluesky?  |   [ 67702200.000 ] posts.x: The comments 💀  |   [ 67660320.000 ] posts.x: 🤯  |   [ 67654200.000 ] posts.x: This is a great example of personalized — and, therefore, quite useful — evals for LLMs. @johnowhitaker built an eval set for multimodal models to…  |   [ 67572600.000 ] posts.x: ↻ @FoundersPodcast: Please help spread the word. Someone in the audience can save Ben’s life and help him get to see his small children gr…  |   [ 67572600.000 ] posts.x: ↻ @davidsenra: Ben Wilson (@BenWilsonTweets) has stage 4 cancer and 4 kids under the age of 5. I want to help him as much as possible, so…  |   [ 67475400.000 ] posts.x: Full talk on using wait conditions and signals to make human approval durable inside a multi-agent system:  |   [ 67475400.000 ] posts.x: Calling a human for approval inside an agent loop as a plain function call is risky: if the service drops while it's waiting, the request can just…  |   [ 67467840.000 ] posts.x: 🤣  |   [ 67449720.000 ] posts.x: Full talk on why agent learning stalls without shared, promotable context, including the benchmark numbers from rebuilding their RAG pipeline:  |   [ 67449720.000 ] posts.x: Giving one agent memory turns out to be the easy part. Getting a second agent to use what the first one already learned is not, and that gap is…  |  

[ 67722840.000 ] posts.x: ↻ @sweis: CLAUDE FOUND THE SEEDS  |   [ 67718940.000 ] posts.x: ↻ @VikParuchuri: This week, we launched Indra, a 1M parameter OCR model that's 15x faster than Tesseract on CPU and much more accurate. I…  |   [ 67707240.000 ] posts.x: This looks promising!  |   [ 67702800.000 ] posts.x: ↻ @awilkinson: "As long as I've got my health, my millions of dollars, my gold house, and my rocket car, I don't need anything else." –Che…  |   [ 67702440.000 ] posts.x: @GeoffreyHuntley What is Bluesky?  |   [ 67702200.000 ] posts.x: The comments 💀  |   [ 67660320.000 ] posts.x: 🤯  |   [ 67654200.000 ] posts.x: This is a great example of personalized — and, therefore, quite useful — evals for LLMs. @johnowhitaker built an eval set for multimodal models to…  |   [ 67572600.000 ] posts.x: ↻ @FoundersPodcast: Please help spread the word. Someone in the audience can save Ben’s life and help him get to see his small children gr…  |   [ 67572600.000 ] posts.x: ↻ @davidsenra: Ben Wilson (@BenWilsonTweets) has stage 4 cancer and 4 kids under the age of 5. I want to help him as much as possible, so…  |   [ 67475400.000 ] posts.x: Full talk on using wait conditions and signals to make human approval durable inside a multi-agent system:  |   [ 67475400.000 ] posts.x: Calling a human for approval inside an agent loop as a plain function call is risky: if the service drops while it's waiting, the request can just…  |   [ 67467840.000 ] posts.x: 🤣  |   [ 67449720.000 ] posts.x: Full talk on why agent learning stalls without shared, promotable context, including the benchmark numbers from rebuilding their RAG pipeline:  |   [ 67449720.000 ] posts.x: Giving one agent memory turns out to be the easy part. Getting a second agent to use what the first one already learned is not, and that gap is…  |