Current Video Large Language Models (Video LLMs) excel in question answering (QA) but largely operate as black boxes, providing textual answers without verifiable visual grounding. Existing explainabi...

The physical world is not merely visual; it is governed by rigorous structural and procedural constraints. Yet, the evaluation of vision-language models (VLMs) remains heavily skewed toward perceptual...

Reinforcement learning (RL) has shown promise in training agentic models that move beyond static benchmarks to engage in dynamic, multi-turn interactions. Yet, the ultimate value of such agents lies i...

Large Language Models (LLMs)-based agents have made impressive progress in reasoning and tool use, enabling them to solve complex tasks. However, their ability to proactively collaborate with users, e...

Although AI agents have demonstrated impressive capabilities in long-horizon reasoning, their reliability is severely hampered by the ``Spiral of Hallucination,'' where early epistemic errors propagat...

Self-evolving agents requires adaptation after deployment, but existing approaches assume a usable learning loop, such as curated skills, successful trajectories, or verifier signals. Real open-world ...

Multimodal retrieval-augmented Generation (MM-RAG) is a key approach for applying large language models (LLMs) and agents to real-world knowledge bases, yet current evaluations are fragmented -- focus...

As autonomous agents increasingly interact, they inevitably attempt to influence one another. While prior work in text-only settings has explored the dynamics of Agent-to-Agent (A2A) persuasion, the r...

The era of software engineering agents is underway. Benchmarks and real-world usage (e.g., tools like Cursor and Claude Code) illustrate that LLMs can be incredibly effective at writing code for real-...

MLLM-as-a-Judge is conventionally validated by agreement with human annotations, but this metric is undefined when the human pool is culturally heterogeneous. We introduce VOIR DIRE, a multimodal benc...

As Large Language Models shift the programming toward human-guided ''vibe coding'', agentic coding tools increasingly rely on models to self-diagnose and repair their own subtle faults -- a capability...

Vision-language models (VLMs) have shown strong capabilities in generating visualization code from textual or visual specifications. However, real-world visualization authoring is inherently iterative...

Real-world data science involves long-horizon workflows that span data wrangling, exploration, modeling, visualization, and validation, and require coordinated use of tools such as notebooks, IDEs, te...

Equipping LLM agents with reusable skills derived from past experience has become a popular and successful approach for tackling complex and long-horizon tasks. However, such lessons are often encoded...

Silvio Savarese on operational intelligence: the AI layer that makes enterprise signal legible, so leaders can act before the moment passes.

5 requirements for trustworthy AI agent-to-agent communication, from Salesforce's Chief Scientist and Chief Legal Officer. Standards, identity, and accountability, explained.

Browse documents within the category: Salesforce-owned Model Cards.

Today, Salesforce is expanding its AI model cards with standardized environmental impact metrics. This update helps customers better understand the energy

Post-training methods (RLVR, On-policy distillation) are Episode-local Language models are getting better at learning from feedback during post-training. In reinforcement learning with verifiable rewa...

Online Reflection for Self-Improving Language Models

Post-training methods (RLVR, On-policy distillation) are Episode-local Language models are getting better at learning from feedback during post-training. In reinforcement learning with verifiable rewa...

Negotiation requires more than inferring what the other side wants: it requires using that information to make advantageous offers and counteroffers over multiple turns. We study whether large languag...

Reinforcement learning from verifiable rewards (RLVR) suffers from sparse outcome signals, creating severe exploration bottlenecks on complex reasoning tasks. Recent on-policy self-distillation method...

A perspective on the future of enterprise AI that is ambient: context-aware, always on, and invisible to users.