ZeroHour
DashboardNewsCVEsExploitsIncidentsAIBriefAnalyticsEntitiesWatchlistIOCsAsk the feedSourcesFeeds & APIKiosk mode

Search: “BFCL”

10 stories in the last 30d

Searchkeywordsmart Sortlatesthotimportancerelevance
DomainAllCyberAIWindow24h3d7d30dAllIn the wildImportant onlyShow duplicates
CategoryAllModel release 5AI research 5
SeverityAllcritical 0high 0medium 0low 0info 10

Google Research Releases ToolGrad: Answer-First Framework Hits 99.8% Pass Rate for Tool-Use Data Generation

Google Research and partners introduce ToolGrad, a verified tool-chain-first data generation framework reaching 99.8% pass rate and boosting Gemma-3-12B to 83.1 on BFCL.

Researchers from Google, the University of Tokyo, RIKEN AIP, and Tohoku University released ToolGrad, which inverts query-first tool-use data generation by executing and verifying API chains before annotating them with user queries. On the ToolBench database of 16,000+ APIs, ToolGrad raised generation pass rate from 63.8% to 99.8% while increasing tool uses per sample from 2.1 to 3.4 and cutting tool-use steps from 34.3 to 20.0. Fine-tuning Gemma-3 at 1B, 4B, and 12B parameters on the 500-sample ToolGrad-500 dataset lifted ToolGrad-12B to 83.1 on the Berkeley Function Calling Leaderboard, near Gemini 2.5 Pro at 83.2 and ahead of GPT-5 at 74.4. Code is Apache-2.0, with the dataset, PyPI package, and models available on Hugging Face.

MarkTechPost · 7d agoAI research1

ToolLoop: Closed-Loop Tool-Use Data Synthesis via Decomposed Generation and Dynamic Self-Feedback

ToolLoop introduces a closed-loop synthetic data framework whose 11K examples lift a 4B model to 86.40% on BFCL tool-use evaluation.

ToolLoop decomposes tool-use data synthesis into function-name sampling, backward derivation of user queries, and forward derivation of tool calls, with dynamic self-feedback at each stage. This shifts the paradigm from generate-then-filter to generate-verify-refine, reducing inefficient and imbalanced synthetic data. A 4B model trained on 11K synthetic examples reaches 86.40% accuracy on BFCL non-reasoning mode (86.07% in an Isolate variant excluding BFCL-overlapping functions) and 72.1% on ACEBench using only 18.3% of baseline training data.

arXiv cs.AI / cs.LG / cs.CL · 9d agoAI research1

Beyond Outcomes: Dual-View Relational Learning for Efficient Agent Benchmarking

DualViewEval compresses agent benchmarks by jointly modeling outcome and process signals, achieving 24x-40x compression with only 20 tasks on APEX-Agents and BFCL.

DualViewEval is an agent benchmark compression method that jointly exploits outcome and process relations from trajectories to learn exact-size minisets predicting full-benchmark scores. The authors analyze large-scale trajectories and identify six process signals systematically associated with final agent performance. Across five agent benchmarks and five baselines, it achieves the best results on all datasets: with only 20 tasks it reaches 24x-40x compression on APEX-Agents and BFCL, reduces MAE by 14.5%-28.2% over the strongest competitors, and improves Kendall's tau by up to 7.2% relative to EssenceBench on SWE-bench Verified.

arXiv cs.AI / cs.LG / cs.CL · 1d agoAI research1

Environments as Scaffold: Enriching Feedback to Bootstrap Self-Evolving Agents in Long-Horizon Tasks

Researchers propose Feedback-Enriched Environments (FEEs) that reduce reward sparsity and improve RL training of Qwen3-based agents on SciWorld and BFCL.

The paper proposes shifting from agent-side warmup (SFT) to environment-side adaptation via Feedback-Enriched Environments to address severe reward sparsity in RL training of long-horizon LLM agents. A pilot study defines a feedback strategy that transitions from action guidance to observation enrichment in later training stages. Large-scale experiments on SciWorld and BFCL across Qwen3 model scales and GRPO, GSPO, and DAPO show consistent gains, plus stabilized training dynamics and proactive exploration.

Hugging Face daily papers · 10d agoAI research1

prism-ml/Ternary-Bonsai-2-27B-gguf — new model trending #29 on Hugging Face

Prism ML released Ternary-Bonsai-2-27B, a 27B ternary-weight model derived from Qwen3.8-27B that runs full reasoning in ~5.95 GB GGUF.

Prism ML published Ternary-Bonsai-2-27B, an Apache 2.0 open-weights model derived from Qwen3.8-27B (27.36B parameters, hybrid attention ~75% linear, 262K-token context). Weights are end-to-end ternary {−1, 0, +1} with g128 FP16 scales at 1.72 bits/weight, shrinking the model from ~54 GB FP16 to 5.95 GB (PTQ1_0) or 7.21 GB (PQ2_0) while reportedly retaining 98.2% of FP16 quality (84.78 average across 14 thinking-mode benchmarks). It ships as GGUF for llama.cpp (CUDA, Metal, CPU) with custom ternary kernels, plus an MLX 2-bit build for Apple Silicon and an optional Q8_0 vision tower pack.

Hugging Face trending models · 1d agoModel release 2 sources1

internlm/Atria-Dawn-Preview — new model trending #30 on Hugging Face

Shanghai AI Laboratory released Atria-Dawn-Preview, a 744B-parameter MoE agentic model with 256K context built on GLM-5.2, targeting research, coding, and cybersecurity automation.

Atria Dawn Preview is a preview instruct model from Shanghai AI Laboratory built on a 744B-parameter MoE GLM-5.2 foundation, designed for agentic loops covering problem analysis, tool use, code implementation, experiment execution, and failure recovery. It targets discovery, creation, delivery, and cybersecurity workflows, including vulnerability validation and fix re-verification in authorized environments. Weights ship on Hugging Face and ModelScope with an FP8-quantized variant under an MIT license and 256K context. Reported results include 96.0 on DeepSearchQA and 92.5 on BrowseComp, compared against DeepSeek V4 Pro 0813, KIMI K3, Qwen 3.8 Max, GLM 5.3, GPT 5.6 sol, and Claude Opus 5.

Hugging Face trending models · 7d agoModel release

OpenBMB Releases MiniCPM5-2B: A 2.52B Dense Model Averaging 53.9 Across 34 Benchmarks and Built to Run On Device

OpenBMB released MiniCPM5-2B, a 2.52B-parameter Apache 2.0 on-device model averaging 53.9 across 34 benchmarks, ahead of Qwen3.5-4B.

OpenBMB released MiniCPM5-2B, a 2,516,756,480-parameter dense LlamaForCausalLM model with grouped-query attention and a 131,072-token context, under Apache 2.0, runnable via vLLM, SGLang, llama.cpp, and Ollama. It averages 53.9 across 34 benchmarks versus 51.1 for Qwen3.5-4B, with strong tool-use (97.1 on tau2-Bench Telecom) and code results (69.1 LiveCodeBench v6, 46.4 SWE-bench Verified). Training combined 400B tokens of deep-thinking SFT, critic-based JustRL II RL teachers, and on-policy distillation merging 16 RL experts; datasets and intermediate checkpoints were published alongside the weights.

MarkTechPost · 10d agoModel release

TokenRhythm/NeoHorse-1-9B — new model trending #30 on Hugging Face

TokenRhythm released NeoHorse-1-9B, an Apache-2.0 9B model post-trained from Qwen3.5-9B with routing-guided agentic post-training toward recursive self-improvement.

TokenRhythm open-sourced NeoHorse-1-9B, a roughly 9B-parameter causal language model fine-tuned from Qwen3.5-9B for agent harnesses, tool use, coding, and instruction following, under Apache-2.0, with a technical report on arXiv (2609.08183). The release contains text-only weights; vision weights are excluded, and repackaging changes configuration and tensor key names without altering tensor values. Its routing harness assigns tasks to a heterogeneous model pool, records tool interactions, and uses capability-level feedback to shape the next training mixture, prototyping an evaluation-selection-update loop toward recursive self-improvement. It scores a 69.04 macro average across ten benchmarks versus 65.60 for the base Qwen3.5-9B (+3.44).

Hugging Face trending models · 13d agoModel release

TokenRhythm/NeoHorse-1-4B — new model trending #30 on Hugging Face

TokenRhythm releases NeoHorse-1-4B, an Apache-2.0 agentic fine-tune of Qwen3.5-4B claiming +5.93 benchmark macro-average gain.

NeoHorse-1-4B is a roughly 4B-parameter text-only causal language model post-trained by TokenRhythm from Qwen/Qwen3.5-4B for agent harnesses, tool use, coding, and instruction following. It applies routing-guided curriculum SFT and routing-guided on-policy distillation over execution trajectories as an early prototype toward recursive self-improvement (RSI). The release reports a 64.87 macro average across ten benchmarks versus 58.94 for Qwen3.5-4B (+5.93) and is distributed under Apache-2.0, trending #30 on Hugging Face.

Hugging Face trending models · 13d agoModel release1

Multi-Step Tool-Calling over Korean Open Public APIs: A Benchmark and a Data-Synthesis Recipe

Researchers introduce KOPA-Bench, a 145-task Korean public API tool-calling benchmark, and EDGE, an execution-grounded data synthesis method.

An arXiv paper presents KOPA-Bench, a benchmark of 145 real-world tasks chaining multiple tool-calls across live Korean government APIs, motivated by data-sovereignty requirements for on-premise open-source LLM agents. It also introduces EDGE, an execution-grounded dynamic graph that keeps only tool-output-to-input links verified by live API calls before synthesizing executable multi-step trajectories. A 9B model fine-tuned with GRPO on the resulting dataset nearly matches its untuned 27B family sibling on KOPA-Bench and improves on the BFCL benchmark.

arXiv cs.AI / cs.LG / cs.CL · 13d agoAI research1