A data viz on now Hall-of-Fame NBA coach Mike D'Antoni His teams finished 1st, 1st, 2nd, 5th, 2nd, 2nd, 2nd, 2nd, 3rd, 9th, 1st, 1st, 1st, and 1st in the league in 3-point attempt rate 🔥

Want to get better at data viz in @vuejs? @DennisAdriaans joins @MichaelThiessen & @TheAlexLichter on @dejavuefm to talk custom charts, choosing libraries, performance tips, and integrating with real dashboards & @nuxt_js backends. ▶️ Watch here:

RL hold-out-1 experiment im curious about to study how model specialization relates to model capacity and data allocation for how many domains to specialize for a common way to get a model to specialize in N domains is Multi-Teacher-On-Policy Distillation train N specialized teachers, do OPD with routing for a set of prompts to teach a model all those abilities for a given domain like data viz -> if we remove X% of the teachers (ie. Don’t specialize on those skills), does our performance on data viz increase? how is this affected by which domains get left out? does it suffer if similar domains are removed but benefit if very different domains are? we still see vertical focused specialized models like GPT-Cyber clearly looks like allocating a lot of data and parameters for a given vertical boosts perf in that vertical more than a general purpose mixture points towards a future where the specialist models always win for high value domains

As AI agents multiply, telling good agents from bad ones is the new fraud challenge. Patrick Harr, CEO of @DataVisor, tells @RemyBlaireNews behavioral AI is the answer, looking at what an agent is attempting to do before evaluating the transaction.

Shipstern 0.9.0 is out. Live filter updates on the Yellowstone gRPC source. Take a RuntimeHandle, patch the filter set with update_filters, the subscription is replaced mid-stream. No reconnect. Also: Transaction V1 (SIMD-0385) support with the inline compute budget on InstructionShared, Prefilter now covers the full Yellowstone subscription surface (memcmp, datasize, lamports, data slices, accounts_exclude, vote, interslot_updates), DISCRIMINATOR constants on generated types, and CPI event envelopes read straight from the IDL. Breaking if you match exhaustively on PrefilterError or build prefilters with struct literals. Builder users unaffected.

Is Fable 5.1 actually better? @petergostev just put @AnthropicAI's latest model head-to-head with Fable 5, Opus 5, GPT-5.6, Kimi K3, GLM 5.3, Qwen-3.8, Grok-4.6, and DeepSeek across coding, 3D, SVG, research, and data-viz tasks. With costs ranging from cents to $65+ per generation, see how Fable 5.1 stacks up to others when it comes to quality vs. price.