INDEX

Explanations

instances where something is applicable or relevant to a majority

oai_token-act-pair · gpt-3.5-turbo

references to the word "most" highlighting its significance or prevalence in various contexts

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

rompt

-0.86

heid

-0.79

icer

-0.75

pload

-0.73

alid

-0.66

 Films

-0.64

orld

-0.64

instead

-0.63

vest

-0.62

thora

-0.62

POSITIVE LOGITS

 importantly

0.84

ONE

0.79

 mornings

0.79

 important

0.73

 superficial

0.73

 mundane

0.71

likely

0.71

 observers

0.70

ones

0.70

body

0.70

Activations Density 0.057%

instances where something is applicable or relevant to a majority

references to the word "most" highlighting its significance or prevalence in various contexts

No Comments

No Known Activations