INDEX

Explanations

editorial notes in texts

oai_token-act-pair · gpt-3.5-turbo

instances of editor's notes

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

pload

-0.73

enic

-0.73

dest

-0.71

duct

-0.71

gur

-0.69

 tremend

-0.69

usting

-0.68

ucer

-0.67

ailability

-0.67

 dirt

-0.66

POSITIVE LOGITS

BOOK

0.90

 NOTE

0.88

 note

0.83

 Keeper

0.78

book

0.78

 Edited

0.77

 EDIT

0.77

":"","

0.76

 Note

0.75

 Corrections

0.74

Activations Density 0.032%

editorial notes in texts

instances of editor's notes

No Comments

No Known Activations