INDEX

Explanations

references to political maneuvers, particularly filibusters

oai_token-act-pair · gpt-3.5-turbo

references to legislative procedures specifically related to filibusters

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

ramid

-0.72

itÃ©

-0.69

orporated

-0.67

ensual

-0.67

iquette

-0.65

irgin

-0.63

eon

-0.62

odka

-0.62

irable

-0.62

MAP

-0.62

POSITIVE LOGITS

 filib

1.16

 filibuster

1.06

buster

0.89

 hearings

0.83

usters

0.82

 blocking

0.69

 glare

0.68

 obstruction

0.67

wright

0.66

 nominating

0.66

Activations Density 0.014%

references to political maneuvers, particularly filibusters

references to legislative procedures specifically related to filibusters

No Comments

No Known Activations