INDEX

Explanations

terms related to terrorist activities

oai_token-act-pair · gpt-3.5-turbo

references to terrorism and terrorist-related activities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Craft

-0.69

wine

-0.69

rys

-0.69

 Mend

-0.69

Ã°

-0.67

Win

-0.67

wright

-0.67

 Mead

-0.66

mg

-0.66

win

-0.65

POSITIVE LOGITS

 terrorist

3.32

terrorist

2.73

 terrorists

2.57

 terrorism

2.38

 jihadist

2.14

Terror

2.10

 terror

2.09

 Terrorism

1.96

 Terror

1.94

terror

1.89

Activations Density 0.026%

terms related to terrorist activities

references to terrorism and terrorist-related activities

No Comments

No Known Activations