INDEX

Explanations

phrases related to analyzing and discussing behavior or actions of individuals

oai_token-act-pair · gpt-3.5-turbo

descriptive phrases and behavioral assessments of individuals

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 concludes

-0.72

orthy

-0.63

 EDITION

-0.62

 Investigators

-0.61

 underscores

-0.61

 licens

-0.60

 Shap

-0.59

 Gutenberg

-0.59

 illustration

-0.58

 listings

-0.58

POSITIVE LOGITS

sul

0.87

 pissed

0.87

 annoyed

0.82

 fucked

0.81

 messed

0.80

 noticeably

0.80

 goof

0.79

 stubborn

0.78

 behaved

0.77

 kinda

0.77

Activations Density 0.903%

phrases related to analyzing and discussing behavior or actions of individuals

descriptive phrases and behavioral assessments of individuals

No Comments

No Known Activations