INDEX

Explanations

references to and descriptions of drunken behavior or incidents

oai_token-act-pair · gpt-3.5-turbo

terms related to intoxication and drunkenness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

isite

-0.87

ILA

-0.84

EMS

-0.84

iron

-0.82

å§«

-0.81

pmwiki

-0.80

DonaldTrump

-0.79

":["

-0.79

eele

-0.78

Downloadha

-0.75

POSITIVE LOGITS

 drunken

1.02

 manslaughter

0.98

 drunk

0.98

 intoxication

0.95

 intoxicated

0.94

 disorderly

0.93

cohol

0.92

 intox

0.90

 alcohol

0.90

 beverages

0.89

Activations Density 0.015%

references to and descriptions of drunken behavior or incidents

terms related to intoxication and drunkenness

No Comments

No Known Activations