INDEX

Explanations

references to news publications and their credibility

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

newspaper sections or headings

oai_token-act-pair · gpt-3.5-turbo

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

gow

-0.72

gui

-0.66

stood

-0.64

 Ambro

-0.61

lyak

-0.61

 segreg

-0.57

 disbanded

-0.57

asca

-0.57

 sucker

-0.57

stones

-0.56

POSITIVE LOGITS

 Subscribe

0.85

 reader

0.84

 HuffPost

0.82

 Email

0.82

 subscribe

0.81

Subscribe

0.81

podcast

0.81

 Newsletter

0.80

 Blog

0.79

iframe

0.79

Activations Density 1.232%

references to news publications and their credibility

newspaper sections or headings

No Comments

No Known Activations