INDEX

Explanations

cigarette-related terms and actions

oai_token-act-pair · gpt-3.5-turbo

mentions of cigarettes and related terminology

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

unity

-0.80

arch

-0.77

Atl

-0.76

-0.73

archy

-0.70

âķĲ

-0.67

 Arche

-0.67

ensional

-0.67

 Cyrus

-0.66

 Unity

-0.66

POSITIVE LOGITS

 cigarette

3.44

 cigarettes

3.33

cigarette

2.89

cig

2.61

cigarettes

2.52

Cig

2.50

 cigar

2.29

cig

2.29

 tobacco

2.18

 smokers

2.11

Activations Density 0.027%

cigarette-related terms and actions

mentions of cigarettes and related terminology

No Comments

No Known Activations