Mikhail Seleznev
myyycroft
AI & ML interests
NLP, AI Safety
Recent Activity
updated a collection 4 days ago
emergent-misalignment-evolutionary-finetuning updated a collection 4 days ago
emergent-misalignment-evolutionary-finetuningOrganizations
Checkpoints for conditional pretraining of gpt-2 models for detoxification task as described in https://arxiv.org/abs/2302.08582.
-
myyycroft/gpt2-toxicity-conditional-5000
Text Generation • 0.1B • Updated • 79 -
myyycroft/gpt2-toxicity-conditional-10000
Text Generation • 0.1B • Updated • 76 -
myyycroft/gpt2-toxicity-conditional-15000
Text Generation • 0.1B • Updated • 74 -
myyycroft/gpt2-toxicity-conditional-20000
Text Generation • 0.1B • Updated • 72
Checkpoints for conditional pretraining of gpt-2 models for detoxification task as described in https://arxiv.org/abs/2302.08582.
-
myyycroft/gpt2-toxicity-conditional-5000
Text Generation • 0.1B • Updated • 79 -
myyycroft/gpt2-toxicity-conditional-10000
Text Generation • 0.1B • Updated • 76 -
myyycroft/gpt2-toxicity-conditional-15000
Text Generation • 0.1B • Updated • 74 -
myyycroft/gpt2-toxicity-conditional-20000
Text Generation • 0.1B • Updated • 72
Checkpoints for MLE baselines of gpt-2 models trained for PII task as described in https://arxiv.org/abs/2302.08582.