Indexing failed for for arXiv:2610.31125
A Hugging Face forum thread reports that arXiv paper 2610.31125 failed to index on Hugging Face Papers, joining a series of similar indexing complaints on the platform's discussion board. Related thre…
A Hugging Face forum thread reports that arXiv paper 2610.31125 failed to index on Hugging Face Papers, joining a series of similar indexing complaints on the platform's discussion board. Related thre…
A forum user asked whether an existing app, GitHub project or AI tool can automatically sort hundreds of graduation and event photos into per-person folders, using diploma names as an identification h…
A dataset called Agentic AI Test Checklist 110, released under CC-BY-4.0, provides 110 test procedures for people testing tool-using LLM agents, published on Hugging Face by srinivaspadmanabhuni. The …
A practitioner guidance piece recommends that teams running AI agents on open models log more than a model name such as llama3:8b per run, because local tags can be repointed and a single name can cov…
Independent researcher Jen Wei presented findings at PyTorch Conference NA 2026 on whether the Muon and Dion3 optimizers retain their convergence advantages over Adam under practical training conditio…
A tutorial on the Muon optimizer has been updated with large-scale experiments testing whether Muon and Dion3 retain their convergence advantages over Adam under practical training conditions, work co…
An independent researcher reports that CECP (Co-Evolving Compositional Pre-Training), an experimental compression-aware LLM training method, matched a dense FP32 baseline on TinyStories, reaching a me…
A new acceleration LoRA for the MiniMax-H3 Ref2VA reference-to-video workflow, released by taobao-content-ai on Hugging Face as MiniMax-H3-PTD-RefAsymOPD-LoRAs, cuts sampling from 40 steps to 6–8 step…
A security-focused post argues that AI agents with refund tools must verify approval claims against the manager-approval system rather than trusting text in a conversation, citing a prompt-injection e…
In Transformers 4.48.3, a model printing LlamaAttention in print(model) does not by itself indicate that SDPA is disabled, because LlamaAttention.forward selects its attention function internally via …
A developer has released a benchmarking suite built natively on the Hub that programmatically stress-tests frontier LLMs on multi-step analytical math and physics reasoning, targeting structural failu…
Developer fritzhand released Hugging Force, an open-source 3D visualization of the Hugging Face Hub in which each model is a star and every fine-tune, adapter, quantization and merge built on it orbit…
A developer operating an ethics-gated agent system called covenant reported that an output-gate cleared only 12 of roughly 2,960 casual chat turns, while an embedding-based "semantic" judge cleared 94…
A causal transformer with roughly 26,000 parameters, released under Apache-2.0 with ONNX export, beat a raw-feature ridge baseline on nowcasting short interest from FINRA daily short volume, scoring a…
A community forum post asks for advice on fine-tuning models for Japanese–English, English–Arabic and Japanese–Arabic translation, and the response recommends freezing a single identical prompt templa…
A user reported that Gemma 4 E4B running in LM Studio stalled at 0% on a roughly 200-page PDF and returned only a superficial summary after the file was split to under 100 pages, and that LM Studio co…
A Hugging Face user submitted the sriram1983007/sra-stablecoin-risk-bench dataset for inclusion in the platform's benchmark allow-list, asking the Hugging Face team to add it per the Evaluation Result…
Developer vynly released CAEL 0.3, an experimental assistant that runs inference without any neural network, transformer or external language model, combining counted language models, cellular automat…
A community forum answer lays out a two-step method for running a language model locally from the command line without a GUI: download Qwen/Qwen2.5-0.5B-Instruct once with `hf download ... --local-dir…
An individual developer released KIARA Voice Studio, an AI audio and video voice-conversion application built on Seed-VC v2, PyTorch, BigVGAN, FFmpeg and Gradio, running cloud inference on Hugging Fac…