datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
FF-Nanbeige4-3B-Base-failure-cases
I- MODEL USED
Published in December 2025, the selected model is Nanbeige4-3B-Base accessible at: https://huggingface.co/Nanbeige/Nanbeige4-3B-Base.
It is a 3-billion-parameter (which is within the required 0.6B–6B range) foundation model (base model) from the fourth generation of the Nanbeige LLM series.
It demonstrates that a compact architecture can deliver strong performance when paired with rigorous improvements to data quality and training methods.
II- MODEL EXPLORATION AND BLIND SPOTS… See the full description on the dataset page: https://huggingface.co/datasets/Choukouriyah/FF-Nanbeige4-3B-Base-failure-cases.lfm25-base-failure-cases
Base Model Failure Cases Dataset
This dataset contains diverse failure cases from a base language model: inputs where the model’s output was incorrect or undesirable, along with the expected (correct or preferred) output. It is intended for analyzing blind spots and for fine-tuning or evaluation.
Model tested
LiquidAI/LFM2.5-1.2B-Base
Type: Base (pre-trained only) causal language model; no instruction tuning.
Parameters: 1.2B.
Released: January 2026 on Hugging Face… See the full description on the dataset page: https://huggingface.co/datasets/mihretgold/lfm25-base-failure-cases.