INDEX

Explanations

phrases indicating an explanation or source for information

oai_token-act-pair · gpt-3.5-turbo

references to external reports or citations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

-0.83

 welcome

-0.64

 cher

-0.64

ARC

-0.63

 swarm

-0.63

 love

-0.62

 raspberry

-0.62

 rooft

-0.62

ãĥŁ

-0.60

tsy

-0.60

POSITIVE LOGITS

uilt

0.90

chwitz

0.88

 Sources

0.84

eteen

0.81

edIn

0.80

Rank

0.80

oras

0.77

orie

0.76

glomer

0.74

encies

0.74

Activations Density 0.013%

phrases indicating an explanation or source for information

references to external reports or citations

No Comments

No Known Activations