AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
Hirundo
Hirundo is an AI company specializing in machine unlearning — enabling organizations to selectively remove knowledge, data, or behaviors from trained models without full retraining.
We build tools and infrastructure for teams that need precise control over what their models know: for compliance, privacy, safety, and model lifecycle management.
What you'll find here
- Research artifacts and benchmarks for machine unlearning
- Fine-tuned and selectively unlearned model checkpoints
- Datasets and evaluation suites for forget/retain performance
models 180
hirundo-io/elephant-1-westernized
Text Classification • 9B • Updated
hirundo-io/Qwen3.8-27B-Westernized
28B • Updated • 61 • 6
hirundo-io/Qwen3.5-9B-Westernized
Image-Text-to-Text • 10B • Updated • 14
hirundo-io/Qwen3.6-35B-A3B-Westernized
Image-Text-to-Text • 36B • Updated • 155 • 2
hirundo-io/Qwen3.5-4B-Westernized
Image-Text-to-Text • 5B • Updated • 81 • 1
hirundo-io/Qwen3.6-35B-A3B-Westernized-LoRA
Updated • 6
hirundo-io/Qwen3.5-4B-Westernized-LoRA
Updated • 6
hirundo-io/Qwen3-Coder-30B-A3B-Instruct-Westernized-LoRA
Updated • 6
hirundo-io/Qwen3-Coder-30B-A3B-Instruct-Westernized
Text Generation • 31B • Updated • 920 • 16
hirundo-io/Qwen3.8-27B-Westernized-LoRA
Updated • 6
datasets 316
hirundo-io/backdoor-sft-case-study
Viewer • Updated • 1.46k • 49
hirundo-io/aligned-backdoor-semantic_concept-target_word-validation
Viewer • Updated • 300 • 31
hirundo-io/secfid-agentic-refusal-qa
Viewer • Updated • 504 • 39
hirundo-io/aligned-backdoor-synonym-big-target_word-train
Viewer • Updated • 600 • 17
hirundo-io/aligned-backdoor-synonym-big-insult-train
Viewer • Updated • 600 • 32
hirundo-io/aligned-backdoor-synonym-big-target_word-validation
Viewer • Updated • 375 • 34
hirundo-io/aligned-backdoor-synonym-big-insult-validation
Viewer • Updated • 375 • 15
hirundo-io/aligned-backdoor-semantic_concept-target_word-train
Viewer • Updated • 600 • 19 • 1
hirundo-io/aligned-backdoor-semantic_concept-insult-validation
Viewer • Updated • 225 • 57
hirundo-io/aligned-backdoor-semantic_concept-insult-train
Viewer • Updated • 600 • 14