INDEX

Explanations

news articles or reports

oai_token-act-pair · gpt-3.5-turbo

references to reading full stories or articles

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ufact

-0.95

Engineers

-0.77

irgin

-0.77

ovo

-0.75

uca

-0.73

ovan

-0.73

orers

-0.71

abbit

-0.71

wan

-0.71

ometers

-0.70

POSITIVE LOGITS

 details

1.09

 text

1.01

 listing

0.98

 description

0.93

 list

0.92

 article

0.91

 descriptions

0.88

 overview

0.86

 excerpt

0.85

 report

0.85

Activations Density 0.053%

news articles or reports

references to reading full stories or articles

No Comments

No Known Activations