Hugging Face's logo
Activity Feed

AI & ML interests

None defined yet.

Recent Activity

camgeodesic  updated a dataset about 19 hours ago
geodesic-research/discourse-grounded-misalignment-evals
Kyle1668  updated a collection 21 days ago
Alignment Pretraining (Geodesic, 2025): Data & Models
Kyle1668  updated a model 21 days ago
geodesic-research/sfm_unfiltered_midtrain_misalignment_upsampled_extreme_sports_em
View all activity

geodesic-research 's collections 6

Alignment Pretraining (Geodesic, 2025): Data & Models
https://alignmentpretraining.ai — Read our paper for additional details about our data and models
Self-Fulfilling (Mis)alignment: Base Models
Here we are, our base model checkpoints. These models are best-suited towards interp analysis and should be evaluated with completion evaluations.
Self-Fulfilling (Mis)alignment: Midtraining Ablations
Models where we try out various approached to positive alignment during midtraining
Self-Fulfilling (Mis)alignment: Post-Trained Models
Here is a selection of models that have undergone DPO. We also share the earlier instruction checkpoints. We recommend using the DPO models.
Alignment Pretraining (Geodesic, 2025): Data & Models
https://alignmentpretraining.ai — Read our paper for additional details about our data and models
Self-Fulfilling (Mis)alignment: Midtraining Ablations
Models where we try out various approached to positive alignment during midtraining
Self-Fulfilling (Mis)alignment: Base Models
Here we are, our base model checkpoints. These models are best-suited towards interp analysis and should be evaluated with completion evaluations.
Self-Fulfilling (Mis)alignment: Post-Trained Models
Here is a selection of models that have undergone DPO. We also share the earlier instruction checkpoints. We recommend using the DPO models.