datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
tiny-aya-global-finance-evaltiny-aya-global-medicine-evaltiny-aya-water-em-insecure-financialtiny-aya-water-em-insecure-medicaltiny-aya-water-evaltiny-aya-base-blindspots
CohereLabs/tiny-aya-base Blind Spots Dataset
Model Tested
CohereLabs/tiny-aya-baseModel Size: 3.35B | Released: February 2026
How the Model Was Loaded
from huggingface_hub import login
from google.colab import userdata
import os
os.environ["HF_TOKEN"] = userdata.get('HF_TOKEN')
login(token=os.environ["HF_TOKEN"])
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
model_name = "CohereLabs/tiny-aya-base"
tokenizer =… See the full description on the dataset page: https://huggingface.co/datasets/Eldagla/tiny-aya-base-blindspots.tiny-aya-earth-em-insecure-financialtiny-aya-base-blindspots
Tiny Aya Blindspots Dataset
Dataset Overview
This dataset contains prompts that reveal blindspots of the Tiny Aya language model. These prompts cover:
Mathematical reasoning
Scientific and chemical reactions
Physical calculations
Language understanding and poetry in Urdu
Riddles and logical puzzles
Computer science problem
Each entry in the dataset has four columns:
Prompt – the input text given to the model
Model Output – what Tiny Aya produced
Expected… See the full description on the dataset page: https://huggingface.co/datasets/MahnoorMalik/tiny-aya-base-blindspots.tiny-aya-global-evaluation
Tiny-Aya-Global Reasoning Blind Spots (TAG-RBS)
This diagnostic dataset identifies the logical, mathematical, and constraint-satisfaction "blind spots" of the Tiny-Aya-Global (3.35B) model. It was manually constructed to test the boundary conditions of compact multilingual models and evaluate their susceptibility to post-hoc rationalization.
Dataset Overview
Dataset Size: 50 hand-crafted prompts.
Evaluation Target: CohereLabs/tiny-aya-global (3.35B parameters).… See the full description on the dataset page: https://huggingface.co/datasets/yonasachule/tiny-aya-global-evaluation.tiny-aya-global-query-dataset
CohereLabs/tiny-aya-global — Blind Spot Dataset
This dataset documents 10 diverse failure cases of the
CohereLabs/tiny-aya-global model,
a ~1B-parameter multilingual instruction-tuned base model released in early 2025.
Each row contains the input prompt, the expected correct output, and the
model's actual (incorrect) output, along with an error category.
Model Tested
Field
Value
Model ID
CohereLabs/tiny-aya-global
Architecture
Transformer decoder (Aya… See the full description on the dataset page: https://huggingface.co/datasets/jm02/tiny-aya-global-query-dataset.tiny-aya-blind-spots
Dataset: Tiny-Aya-Base Blind Spots
This dataset was created as part of a technical challenge to identify the blind spots of the models. It specifically targets CohereLabs/tiny-aya-base. The model is a 3.35B parameter multilingual base model released in early 2026.
Model Tested
Model: CohereLabs/tiny-aya-base
Parameters: 3.35 Billion
Modality: Text
How the Model was Loaded
The model was loaded using the transformers library on a Google Colab T4 GPU.… See the full description on the dataset page: https://huggingface.co/datasets/osamaahmed17/tiny-aya-blind-spots.tiny-aya-base-blindspots
Technical Challenge: Blind Spots of Frontier Models
1. Model Tested
Model: CohereLabs/tiny-aya-base
Parameters: 3.35 Billion
I specifically chose this model because it is a raw, pre-trained base model. It has not undergone Supervised Fine-Tuning (SFT) or Reinforcement Learning from Human Feedback (RLHF). This makes it ideal for discovering foundational "blind spots" in logic, instruction following, and formatting.
2. How the Model was Loaded
The model was… See the full description on the dataset page: https://huggingface.co/datasets/Ameeque/tiny-aya-base-blindspots.ignatius-tiny-aya-analysis
Tiny Aya Base — Failure Analysis Dataset
Model Tested
Model: CohereLabs/tiny-aya-base
Parameters: 3.35B
Released: February 17, 2026
Architecture: Dense decoder-only Transformer, pretrained on 6T tokens across 70+ languages
How I Loaded the Model
I used Modal (free tier, T4 GPU — 16GB VRAM) to run inference.
import modal
app = modal.App("tiny-aya-probe")
image = modal.Image.debian_slim(python_version="3.11").pip_install(
"torch", "transformers>=4.51.0"… See the full description on the dataset page: https://huggingface.co/datasets/IgnatiusBalayo2024/ignatius-tiny-aya-analysis.tiny-aya-earth-em-insecure-medicaltiny-aya-water-blind-spots
Tiny-Aya-Water Blind Spot Evaluation Dataset
Overview
This dataset contains 10 manually constructed blind spot cases used to evaluate reasoning, instruction following, and factual robustness of the model:
CohereLabs/tiny-aya-water
The goal of this dataset is to identify systematic weaknesses and failure modes of the model under controlled experimental conditions.
Model Tested
Model: Tiny-Aya-Water
Repository: https://huggingface.co/CohereLabs/tiny-aya-water… See the full description on the dataset page: https://huggingface.co/datasets/chrishounwanou/tiny-aya-water-blind-spots.tiny-aya-fire-em-insecure-text-medical