INDEX

Explanations

words related to scandals and controversies, particularly those involving public figures or institutions

oai_token-act-pair · gpt-3.5-turbo

references to scandals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ignt

-0.70

hetics

-0.68

ingham

-0.67

lasses

-0.65

ignty

-0.64

itar

-0.64

 Directive

-0.62

*/(

-0.62

cki

-0.62

erate

-0.61

POSITIVE LOGITS

 involving

1.15

 engulf

1.12

 surrounding

1.06

 erupted

1.05

 plag

1.05

 revolving

1.02

 swirling

0.95

 scandals

0.95

ous

0.94

 scandal

0.93

Activations Density 0.072%

words related to scandals and controversies, particularly those involving public figures or institutions

references to scandals

No Comments

No Known Activations