INDEX

Explanations

acronyms related to political or military organizations

oai_token-act-pair · gpt-3.5-turbo

references to the abbreviation "AF," which appears to be a recurring term or concept throughout the document

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

 Gors

-0.76

 Vaugh

-0.75

 Boone

-0.70

 McKenna

-0.70

 Blackwell

-0.69

igators

-0.69

 Gott

-0.65

 Morg

-0.65

 Heller

-0.64

 McAuliffe

-0.64

POSITIVE LOGITS

RIC

1.20

ornia

1.04

SPA

1.01

RA

0.95

FL

0.92

amily

0.91

raid

0.88

ICA

0.87

rost

0.86

ECA

0.86

Activations Density 0.013%

acronyms related to political or military organizations

references to the abbreviation "AF," which appears to be a recurring term or concept throughout the document

No Comments

No Known Activations