INDEX

Explanations

directory-related terms within text

oai_token-act-pair · gpt-3.5-turbo

references to directory structures in file systems

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

acted

-0.80

rained

-0.80

acting

-0.79

oria

-0.72

lev

-0.72

akening

-0.70

pter

-0.69

laws

-0.69

speak

-0.68

uca

-0.68

POSITIVE LOGITS

 directory

0.98

 directories

0.94

directory

0.84

 descriptor

0.84

 Directory

0.78

 cellar

0.75

 separ

0.75

 Contents

0.75

 hierarchy

0.74

 emulator

0.71

Activations Density 0.008%

directory-related terms within text

references to directory structures in file systems

No Comments

No Known Activations