The Video Signal

Topic

SWE-bench

All digests tagged SWE-bench

Stanford CS329A Self-Improving AI Agents | Part 2 | Test-Time Compute Scaling thumbnail

· 1:03:21

Stanford CS329A Self-Improving AI Agents | Part 2 | Test-Time Compute Scaling

The lecture details advanced methods for improving Large Language Model (LLM) performance through 'inference scaling' or 'test-time compute scaling,' rather than relying solely on expensive pre-training. Key techniques include repeated sampling (Large Language Monkeys), which shows that coverage follows a predictable power law with the number of samples. The discussion highlights the critical need for robust verification mechanisms to bridge the generation-verification gap, and concludes by introducing the Arkon framework, an architecture search method that optimizes complex inference pipelines using components like Fusion, Critic, and Ranker.

Key takeaways

  1. Inference Scaling Paradigm Shift 1:45

    LLM capability can be significantly enhanced at inference time by increasing compute (e.g., repeated sampling) without modifying model parameters or requiring fine-tuning, offering a new paradigm compared to traditional pre-training and fine-tuning.

  2. Repeated Sampling Effectiveness 2:40

    By repeatedly querying the same problem (e.g., using Llama 3-8b or DeepSeek), selecting the correct response among candidates, models can achieve performance comparable to larger proprietary models like GPT-4o.

  3. The Role of Verification 7:50

    For repeated sampling to be effective, automated verification is crucial. The 'generation-verification gap' describes the large difference between the best possible outcome (Oracle selection) and what can be achieved using simple methods like majority voting.

  4. Advanced Scaling Architectures 20:30

    The Arkon framework treats inference scaling as an architecture design problem, optimizing the combination of techniques (e.g., Fusion, Critic, Ranker) to maximize accuracy given a limited compute budget.

Watch on YouTube Full article

When Will The Benchmaxxing Plague End? — Nick Heiner, Surge AI thumbnail

· 17:25

When Will The Benchmaxxing Plague End? — Nick Heiner, Surge AI

The talk argues that the concept of 'benchmaxxing'—where models are trained excessively on benchmarks in ways that deviate from real-world utility—is rampant across AI. The speaker identifies several critical anti-patterns in benchmark creation, including contamination (memorizing test data), reward hacking, and misalignment between prompts and verifiers. Heiner advocates for a shift toward high-fidelity human evaluation, domain expertise, rigorous Quality Control (QC), and ensuring that benchmarks are aspirational artifacts reflecting true user needs rather than arbitrary scores.

Key takeaways

  1. Benchmaxxing is an industry problem

    The existence of 'benchmaxxing' indicates a gap between benchmark scores and real-world performance. This phenomenon is driven by incentives, poor methodologies, and the need for easily digestible metrics in a hyped market.

  2. Contamination is often default 7:21

    Contamination occurs when models memorize public questions and answers from the internet (e.g., Opus memorizing SWE-bench contents), making scores reflect recall rather than generalized ability.

  3. Verifiers must be fully aligned with prompts 10:23

    Misalignment between the prompt and the verifier (e.g., asking for no commas but accepting Hindi) or using hard-coded string matches introduces noise, leading to misleading scores.

  4. High-quality benchmarks require domain expertise

    Creating effective benchmarks requires not just technical experts (e.g., doctors for a medical benchmark) but also those with business and regulatory sense to understand the deployment environment.

  5. Human evaluation is necessary, despite cost

    The core value remains human preference. While expensive, maximizing quality requires paying for good workers rather than minimizing costs to achieve scalable metrics.

Watch on YouTube Full article

The misaligned incentives behind AI coding agents thumbnail

· 50:16

The misaligned incentives behind AI coding agents

The conversation details how AI coding agents, exemplified by Devin, are fundamentally changing software engineering workflows. The industry is moving past simply training larger models and focusing intensely on optimizing cost-efficiency (token spend) and speed. Key technical advancements include the 'sidekick' agent architecture for achieving high price performance, developing advanced evaluation metrics like 'mergeability' via Frontier Code, and implementing proactive automation to shift human engineers into decision-making roles rather than routine coding tasks.

Key takeaways

  1. The Shift from Capability to Efficiency 8:36

    As agents mature, the bottleneck is shifting from model training size to running evaluations and managing costs. The focus has moved toward optimizing speed and cost rather than chasing the absolute best-performing frontier model for every task (5:56).

  2. The Role of Mergeability in Evaluation 14:01

    A critical gap in current evaluation benchmarks is 'mergeability'—determining if code, while technically correct, would improve the overall quality or maintainability of a codebase. Cognition developed Frontier Code to address this (8:41).

  3. Cost Optimization via Sidekick Architecture 35:46

    The 'sidekick' agent architecture allows for running both a high-quality, expensive model and a more price-performant model in parallel. This dual approach enables significant cost savings (up to 35% better price performance) without sacrificing quality (21:46).

  4. Proactive Automation and Productivity Guarantees

    Agents are moving from reactive task completion to proactive automation, handling tasks like triaging messages or suggesting fixes. This capability led Cognition to underwrite a $10 million productivity guarantee based on measuring 'productive engineering output' (46:51).

Watch on YouTube Full article