INDEX

Explanations

high-ranking official positions in government or military

oai_token-act-pair · gpt-3.5-turbo

titles and roles related to government officials

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Äĩ

-0.88

crew

-0.85

nikov

-0.84

kinson

-0.74

ensis

-0.71

MpServer

-0.71

eros

-0.69

eat

-0.69

crow

-0.69

elled

-0.68

POSITIVE LOGITS

Tim

0.98

 Stephen

0.95

 Bill

0.93

 Mark

0.93

 Steve

0.92

Ann

0.92

 David

0.92

 George

0.92

 Patrick

0.91

 Shaun

0.91

Activations Density 0.129%

high-ranking official positions in government or military

titles and roles related to government officials

No Comments

No Known Activations