INDEX

Explanations

terms related to terrorist activities, including plots, groups, individuals, and operations

oai_token-act-pair · gpt-3.5-turbo

references to terrorism and terrorist activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Rockies

-0.75

 Hort

-0.74

 Adjusted

-0.74

 Orth

-0.73

galitarian

-0.72

Scot

-0.72

 Construct

-0.71

ðĿ

-0.71

 Neurolog

-0.71

 Lunar

-0.71

POSITIVE LOGITS

 terrorists

1.16

 bombing

1.15

 terrorist

1.10

 jihad

1.10

 terror

1.10

 Islamists

1.09

 jihadists

1.08

 terrorism

1.07

 killings

1.07

 jihadist

1.06

Activations Density 0.388%

terms related to terrorist activities, including plots, groups, individuals, and operations

references to terrorism and terrorist activities

No Comments

No Known Activations