INDEX

Explanations

sources of information or quotes attributed to individuals in written text

oai_token-act-pair · gpt-3.5-turbo

references to informants or sources in reporting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

aeper

-0.77

osate

-0.73

equality

-0.69

emale

-0.68

ĸļ

-0.68

arrow

-0.67

eer

-0.66

andel

-0.65

avorite

-0.65

 boulder

-0.65

POSITIVE LOGITS

 sources

1.11

 Sources

0.97

 source

0.95

 informants

0.83

source

0.83

ISI

0.82

 informant

0.80

 quoted

0.77

Sources

0.75

 insiders

0.75

Activations Density 0.018%

sources of information or quotes attributed to individuals in written text

references to informants or sources in reporting

No Comments

No Known Activations