Hugging Face Transformers Code Agent Beats GAIA Benchmark
Hugging Face reports that their Transformers-based code agent has achieved a top score on the GAIA benchmark, a challenging evaluation for general AI assistants requiring multi-step reasoning and tool use. The result positions Hugging Face's open agent framework competitively against proprietary systems. The post details the agent architecture and tooling approach used to achieve the result.
Related guides (4)
Related events (8)
Gaia2 and ARE: Empowering the community to study agents
Hugging Face has released Gaia2 and the Agent Reasoning Evaluation (ARE) framework, aimed at enabling the research community to study and benchmark AI agents. The post describes new tools and datasets for evaluating agent capabilities, building on the original GAIA benchmark. This represents an expansion of the agent evaluation ecosystem with community-oriented tooling.
Hugging Face benchmarks open models on agentic tool-use tasks
Hugging Face published a blog post examining whether open models are sufficiently capable for agentic use cases, focusing on benchmarking them against real-world tooling. The post addresses the practical question of which open-weights models can reliably handle tool-calling and multi-step agentic workflows. This is relevant to practitioners evaluating open models for agent deployments.
CUGA on Hugging Face: Democratizing Configurable AI Agents
IBM Research has released CUGA (Configurable Universal Generative Agent) on Hugging Face, positioning it as a framework for building configurable AI agents. The announcement appears on the Hugging Face blog as a tier-2 commentary piece from IBM Research. Details on architecture, benchmarks, and specific capabilities are not available from the body text provided.
License to Call: Introducing Transformers Agents 2.0
Hugging Face announced Transformers Agents 2.0, a major update to their agent framework built on top of the Transformers library. The release introduces new abstractions for tool use, multi-step reasoning, and agent orchestration, positioning it as a production-ready framework for building AI agents. The update reflects growing ecosystem investment in standardized agent tooling patterns.
Habana Labs and Hugging Face Partner to Accelerate Transformer Model Training
Habana Labs and Hugging Face announced a partnership to accelerate transformer model training on Habana's Gaudi AI processors. The collaboration aims to integrate Hugging Face's Transformers library with Habana's hardware, offering an alternative to GPU-based training infrastructure. This represents an early effort to diversify the AI training hardware ecosystem beyond NVIDIA dominance.
How Hugging Face Sped Up Transformer Inference 100x for API Customers
Hugging Face describes engineering optimizations that achieved up to 100x speedups in transformer inference for their hosted API customers. The post covers techniques applied to accelerate model serving at scale. This is a 2021 article documenting early inference optimization work at Hugging Face's inference API product.
Hugging Face demonstrates agent chaining two Spaces to build a 3D Paris gallery
A Hugging Face blog post describes an agent that autonomously chains two Hugging Face Spaces to generate a 3D gallery of Paris, illustrating multi-step tool use and Space-to-Space orchestration. The demo showcases how agents can compose existing hosted ML tools without custom infrastructure. This is a practical capability demonstration relevant to the agent-tool ecosystem.
Hugging Face and Graphcore Partner for IPU-Optimized Transformers
Hugging Face and Graphcore announced a partnership to optimize Transformer models for Graphcore's Intelligence Processing Unit (IPU) hardware. The collaboration aims to make IPU-accelerated inference and training accessible through the Hugging Face ecosystem. This represents an early effort to broaden AI hardware options beyond GPU-dominated infrastructure.