INDEX

Explanations

references to terrorism-related terms and phrases

oai_token-act-pair · gpt-3.5-turbo

references to terrorism and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

galitarian

-0.86

endum

-0.78

Salt

-0.71

 Quartz

-0.71

Iterator

-0.70

lease

-0.67

Sync

-0.67

dit

-0.66

resso

-0.65

bye

-0.64

POSITIVE LOGITS

fully

0.94

abad

0.91

istan

0.90

efully

0.89

 terror

0.85

 attacks

0.82

 bombing

0.82

fulness

0.81

 bomber

0.81

 spree

0.77

Activations Density 0.006%

references to terrorism-related terms and phrases

references to terrorism and related concepts

No Comments

No Known Activations