INDEX

Explanations

mentions of individuals being drunk

oai_token-act-pair · gpt-3.5-turbo

references to alcohol consumption

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

atell

-0.76

ospace

-0.69

tel

-0.67

orama

-0.65

 Extension

-0.64

olutions

-0.64

Catalog

-0.64

align

-0.63

 Gates

-0.63

 Prim

-0.63

POSITIVE LOGITS

 drunk

3.76

 intoxicated

2.38

 drunken

2.35

 Drunk

2.20

 intoxication

1.65

 drank

1.60

 drinking

1.59

 alcoholic

1.58

 drink

1.58

 booze

1.57

Activations Density 0.015%

mentions of individuals being drunk

references to alcohol consumption

No Comments

No Known Activations