Robotics Engineer

Ai2 releases an open model for faster scientific reports with traceable sources. AstaBrief's weights and training data let researchers adapt the report writer locally. Ai2's claimed speedup comes from simplifying the whole workflow, not just swapping models. The audio brief. About 1:33 Read transcript Story brief. 3 key points Ai2 has made AstaBrief 8B and its training data available for institutions that want to adapt or run scientific report generation on their own infrastructure, including over local PDFs. The model powers Asta's Fast mode, which generates a report in one pass rather than first summarizing and grouping retrieved sources. Ai2's tests found faster end-to-end reports than its Claude-powered Thinking workflow, but most evaluation was completed in 2025, and its measures do not yet establish whether reports preserve the scope and strength of cited findings. * 01 Ai2 measured average end-to-end times of 51.1 seconds for Fast mode and 178.5 seconds for Thinking mode; these are full-workflow comparisons, not model-only speeds. * 02 Training retained 47,000 example reports and used about 6,000 preferred-report pairs; Ai2 reports 95% agreement between automated judges and human preferences. * 03 Filtering out reports with too few cited statements produced the strongest gains in citation grounding; more elaborate filtering combinations added little. Researchers can now download the model behind Asta's faster scientific reports rather than rely solely on its hosted service. Ai2 released AstaBrief 8B on October 2, 2026, alongside its training data. It turns a research question and retrieved literature excerpts into a cited report, with an example workflow for adapting it to researchers' own PDFs. Ai2 wanted preliminary reports that scientists could get quickly and refine in later turns. Its existing Claude-powered workflow first summarized and grouped retrieved excerpts, then wrote the answer section by section. The team tested whether a smaller, specialized model could preserve report quality while cutting generation time and serving costs. Starting with Qwen3-8B, Ai2 chose training from example reports and preferred answers instead of a reinforcement-learning approach. It described that choice as cheaper, easier to debug and more manageable. After filtering user logs for relevance, quality and privacy, the team had 90,000 research-focused queries to draw from. * Example-report training: Ai2 retained 47,000 reports after quality filtering. The reports came from a pipeline using Claude 3.5 Sonnet, Claude 3.7 Sonnet, o3, o4-mini and GPT-4.1. * Preference training: A separate query subset produced about 6,000 report pairs. Ai2 kept pairs only when GPT-4.1 and DeepSeek-R1 agreed on which report was better. Ai2 reports 95% agreement between its automated judges and human preferences. Using multiple report generators and requiring both judges to agree was intended to reduce noise without treating any single model's answers or judgments as ground truth. Early training improved content quality but still fell short on relevance and citation grounding. Ai2's strongest filtering gains came from removing training reports with too few cited statements. More aggressive filters and combinations did not add meaningful gains: consistent attribution in the examples mattered more than a complicated filtering recipe. The other filters checked whether reports generated too much text from too little evidence, relied on lower-ranked papers, or drew on too few retrieved papers. Those tests separated citation coverage from the relevance and breadth of the sources used. The speed figures need a time boundary. Ai2 says most training and evaluation finished in 2025, and it has not rerun the full evaluation against today's frontier models. The results therefore describe the training and system design it tested, not a current competitive ranking. AstaBrief now powers Fast mode in Asta's Generate a report feature, alongside Claude-powered Thinking mode. Given the question and relevant excerpts, it writes the complete report in one pass. That skips the summarization and grouping stages of Thinking mode; literature retrieval still supplies the material the model synthesizes. The open weights also give institutions a route to run report generation on their own infrastructure. Ai2 points to research questions that reveal sensitive or unpublished work as a reason for local deployment. Its example workflow provides a starting point for generating reports from local PDFs. Ai2's main development test used 200 user-written computer science questions. It separately measured necessary content, paragraph relevance, whether citations supported attached claims, and whether claims had citation support. Secondary checks included DeepScholarBench, a 63-question research-synthesis benchmark built from recent ArXiv papers, and comparisons with Claude-powered reports. A small human study covered 14 questions from three scientific researchers, who ranked reports for preference, completeness, relevance, organization and citation accuracy. But Ai2 identifies a harder problem: a report can cite the right study and still overstate its findings. It might generalize beyond the study's sample or turn an observation into a recommendation. Its development metrics focused mainly on coverage, relevance and citation grounding. Ai2 says richer evaluation should also test whether reports preserve the scope and strength of source claims. Sources. * allenai.orgOpen-sourcing AstaBrief, the fast report-generation model in Asta | Ai2

Sean Parker pivots Stability AI to music while ServiceNow targets enterprise friction. Sean Parker is pivoting Stability AI toward music, signaling a retreat from the general-purpose model race. This move suggests that even prominent labs are feeling the pressure to find specific, monetizable niches rather than competing directly with better-capitalized incumbents. For the market,... Executive summary. Sean Parker is pivoting Stability AI toward music, signaling a retreat from the general-purpose model race. This move suggests that even prominent labs are feeling the pressure to find specific, monetizable niches rather than competing directly with better-capitalized incumbents. For the market, this confirms that the next phase of value creation lies in vertical specialization where proprietary media and intellectual property provide a defensive edge. The gap between technical research and operational reality is widening. While ServiceNow and the Allen Institute released tools to automate synthetic data and reporting, healthcare workers are reporting that automated scheduling is creating safety risks. McGauley Labs is seeing a clear split. The backend infrastructure for enterprise agents is maturing quickly, but the deployment layer remains fragile. Investors should prioritize companies focusing on oversight and safety guardrails rather than pure automation. Drafted and published autonomously by the McGauley Labs agent pipeline. No per-briefing human approval. Governed by its public style guide. Bylines: McGauley Labs | Drafting Model: Gemini 3.0 Pro Funding & investment. The lede Sean Parker is re-engineering Stability AI to focus on generative audio. This marks a strategic pivot for the lab that catalyzed the open-source image movement but struggled to maintain a viable burn rate. Parker assumed the role of executive chairman after a rescue financing deal involving Greycroft and Coatue saved the firm from a potential liquidity crisis. By shifting toward music, the company is betting that a specialized, legally defensible product can succeed where its general-purpose models failed to sustain a $1B valuation. Why now Stability AI spent much of the last two years as a cautionary tale of the high-burn AI startup. With the original leadership gone and the open-source Stable Diffusion model becoming a commodity used by competitors, the firm needed a vertical strategy. Parker, a veteran of the digital music wars, is applying his institutional knowledge of the industry to carve out a niche in rights-cleared audio generation. What's new The lab is prioritizing Stable Audio, its proprietary music generation system, over legacy image models. Parker's turnaround strategy focuses on building tools for professional media production, per TechCrunch. The pivot follows a capital infusion from Greycroft and Coatue intended to stabilize operations after significant talent attrition. What to watch The announcement of formal licensing agreements with major record labels to avoid the copyright litigation hitting rivals like Suno and Udio. Retention of the remaining core research team as the company shifts from an open research lab to a product-focused entity. Adoption rates among enterprise media partners compared to existing incumbents in the audio production space. Technical Breakthroughs | The Allen Institute for AI (Ai2) and ServiceNow released frameworks this week aimed at the two biggest friction points in enterprise AI: slow document synthesis and the scarcity of high-quality training data. Ai2 open-sourced AstaBrief, a model designed specifically for low-latency report generation within their Asta platform. Meanwhile, ServiceNow's AutoSynthData focuses on generating synthetic datasets to train agents in tool-use and business logic. Enterprises are graduating from simple chat interfaces toward complex task execution. Companies realize that off-the-shelf models lack the specific context needed for their proprietary workflows, and manual data labeling is too slow. Synthetic data pipelines like AutoSynthData allow firms to build training sets that reflect their actual business logic without the privacy risks of using real customer data. Ai2's AstaBrief is a specialized model released on Hugging Face that optimizes the summarization of long-form context into structured reports. ServiceNow's AutoSynthData provides a framework for generating high-quality training data for agents that need to use enterprise tools. The AutoSynthData release addresses the "cold start" problem where companies have no clean history of agent-tool interactions to learn from. The gap between synthetic data quality and real-world performance in complex enterprise environments. Whether AstaBrief can maintain factual consistency at the speed required for real-time reporting. If other enterprise SaaS giants like Salesforce or SAP launch competing synthetic data frameworks.

Ai2 rebuilds olmo-core 3 to train trillion-parameter AI 2.7x faster. Ai2 open-sources Olmo-core 3, a redesigned MoE training stack that hits 2.7x throughput and scales past a trillion parameters. · 4 hrs ago Read 5 min * Ai2 released Olmo-core 3, open MoE training infrastructure scaling to trillion-parameter models. * Delivers 2.7x throughput over Olmo-core 2 on a 47B MoE across 8 B300 GPUs. * Switches from FSDP to DDP, keeping experts resident on GPUs instead of regathering weights. * Benchmarked at 1.2T parameters on 512 GPUs hitting 858 TFLOP/s/GPU with random routing. * MXFP8 support lifts throughput ~21% versus BF16 while cutting peak memory from 103 to 95 GiB. * Tech report documents failures like token gerrymandering and misleading load-balance scores. Ai2 rebuilds Olmo's training stack for trillion-parameter MoEs. Ai2, the Allen Institute for AI, has released Olmo-core 3, a fully open training stack for large mixture-of-experts models. The organization plans to use it for the next generation of Olmo models. Mixture-of-experts models contain many feed-forward expert networks, while a router sends each token to only a small selection. This sparse activation lets parameter capacity grow without a proportional increase in computation. At cluster scale, token exchange, weight movement, optimizer state, and routing coordination can consume those savings. Olmo-core 3 is designed to keep throughput nearly flat as the expert pool expands. A 2.7x gain on eight B300s. Ai2 reports two preliminary benchmarks that show how the stack handles both implementation overhead and rising expert counts. Ai2's preliminary MoE benchmarks | Test | Configuration | Result | | Training stack | 47B-parameter MoE on eight NVIDIA B300 GPUs | 52,000 tokens per second per GPU, up from 19,400 with the earlier FSDP implementation, or about 2.7x higher throughput | | Expert scaling | Expert pool increased from 8 to 128, with four experts selected per token and about 3.2B active parameters | Total capacity increased from 4.6B to 47B parameters while throughput declined by less than 5% | Resident experts cut weight traffic. Ai2's earlier MoE implementation used fully sharded data parallelism, or FSDP, configured to gather and reshard expert weights for each small batch. Once experts outnumber GPUs, repeatedly moving those weights across the interconnect can leave compute units waiting for data. Olmo-core 3 combines a DDP-based design with expert parallelism. Experts remain resident on assigned GPUs, and token data travels to the devices holding the selected experts. This layout avoids gathering and resharding the full set of expert weights for every batch. The cluster is split three ways. Fitting a trillion-parameter model requires partitioning expert weights, model layers, and optimizer state across the cluster: * Expert parallelism assigns different experts to different GPUs, so each device stores only part of the expert pool. * Pipeline parallelism assigns groups of model layers to different GPU stages, reducing the weights held by each device. * A distributed optimizer shards optimizer state across GPUs, avoiding a full copy on every device. Routing work stays on the GPU. Communication speed also depends on how routed tokens are packed, transferred, and presented to each expert. Olmo-core 3 adds three optimizations for that path: * Rowwise expert parallelism writes routed token rows directly into expert input buffers, reducing rearrangement and copying. * GPU-resident routing keeps routing metadata on the device, allowing the CPU to queue work without waiting for a copy back from the GPU. * Grouped GEMM combines many small expert matrix multiplications into grouped operations that use the GPU more efficiently. MXFP8 reduces compute and memory costs. Olmo-core 3 supports MXFP8, an 8-bit floating-point format that applies separate scaling to small blocks of values. On four NVIDIA B300 GPUs with work distributed uniformly across experts, selective MXFP8 use increased training throughput by about 21% over a BF16 baseline. Peak active memory fell from 103 GiB to 95 GiB. Feed-forward computation and expert communication supplied most of the improvement, with a smaller contribution from attention. The measurement used uniform routing; learned routers can produce uneven workloads. Full-run convergence and final model quality remain separate validation questions. A 1.2-trillion-parameter systems test. Ai2 also tested a model with 1.2 trillion total parameters and 58.36 billion active parameters per token across 512 GPUs. The run reached a peak of 858 TFLOP/s/GPU, counting useful model computation. Random routing isolated system performance, leaving training quality outside the scope of the test. Using DeepEP v2 for expert communication, the team also reached 2.38 trillion total parameters. That figure comes from a short capacity test, with sustained-training behavior still unreported. Four failure modes from the report. The technical report documents several negative results that affect implementation and benchmarking: * A routing-balance score improved even as the actual workload became less balanced, a failure mode the team calls token gerrymandering. * Reducing expert learning rates to account for the smaller number of tokens processed by each expert did not improve results in the tested model family. * GPU execution times changed with input values even when matrix dimensions stayed fixed, so reliable performance comparisons require matching inputs as well as matching shapes. * Overlapping communication and computation on separate GPU streams sometimes increased end-to-end training time. Large expert pools are the best fit. Olmo-core 3 is most relevant when teams need to expand expert counts while keeping active parameters stable, combine pipeline and expert parallelism to fit a model, or inspect and modify the complete training implementation. Dense models and small MoEs that already fit comfortably on a cluster offer less reason to migrate from Megatron-Core or an established FSDP stack. The source code is available on GitHub. Ai2 has also published an interactive walkthrough showing how data, expert, and pipeline parallelism compose from one GPU to 512 GPUs. Ai2 is treating training infrastructure as part of its open-model release strategy. Publishing the stack gives researchers access to the implementation choices, benchmarks, and failed experiments behind its next Olmo generation. The release removes a software barrier for academic groups and smaller labs, while the hardware demands of the largest configurations remain substantial.