INDEX

Explanations

phrases that indicate a major event or controversy

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a response or consequence to a particular event

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

inet

-0.85

quote

-0.79

utical

-0.76

grounds

-0.70

esse

-0.70

 istg

-0.68

ELY

-0.68

itsu

-0.67

oided

-0.67

nels

-0.67

POSITIVE LOGITS

 revelations

1.00

 Hurricane

0.96

 tragedies

0.95

 scandals

0.93

 controversies

0.85

 setbacks

0.83

 allegations

0.83

 adversity

0.82

 accusations

0.82

 criticism

0.81

Activations Density 0.097%

phrases that indicate a major event or controversy

phrases indicating a response or consequence to a particular event

No Comments

No Known Activations