INDEX

Explanations

references to experiences or actions in a first-person perspective

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "person" in various contexts

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

DERR

-0.73

enthal

-0.69

Tx

-0.68

ORK

-0.66

tty

-0.65

Mb

-0.65

YP

-0.64

CCC

-0.64

 avoidance

-0.63

Phill

-0.63

POSITIVE LOGITS

nel

1.08

hood

1.01

ality

0.94

atives

0.92

uscript

0.87

alities

0.86

izontal

0.85

acles

0.81

atural

0.80

alties

0.78

Activations Density 0.032%

references to experiences or actions in a first-person perspective

references to the concept of "person" in various contexts

No Comments

No Known Activations