INDEX

Explanations

terms related to political positions or roles, specifically "minister."

oai_token-act-pair · gpt-3.5-turbo

references to government officials and their titles

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

esville

-0.74

 Braves

-0.67

 Crew

-0.66

oven

-0.64

user

-0.63

 Crow

-0.63

Crew

-0.63

Model

-0.63

 Century

-0.62

irc

-0.61

POSITIVE LOGITS

ariat

0.75

 Lavrov

0.75

 envoy

0.75

 spokesman

0.75

 Marino

0.73

 chair

0.73

 adviser

0.72

 gestures

0.72

ruary

0.72

 advisor

0.72

Activations Density 0.077%

terms related to political positions or roles, specifically "minister."

references to government officials and their titles

No Comments

No Known Activations