INDEX

Explanations

references to universities, particularly Northwestern University and Stanford University

oai_token-act-pair · gpt-3.5-turbo

references to universities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ppings

-0.81

meaning

-0.77

othy

-0.72

served

-0.65

tro

-0.65

outer

-0.64

 favour

-0.62

ãĥĸ

-0.60

everything

-0.59

ï¸ı

-0.59

POSITIVE LOGITS

 Univ

0.96

 University

0.96

 College

0.91

ĸļ

0.90

uates

0.86

 doctoral

0.85

 campuses

0.82

 professor

0.82

 Libraries

0.81

Phi

0.80

Activations Density 0.037%

references to universities, particularly Northwestern University and Stanford University

references to universities

No Comments

No Known Activations